THE internet ENCYCLOPEDIA 1 volume 3 phần 7 pdf

HOW STREAMING WORKS Streaming involves taking video or audio ﬁles, breaking them down into packets of information, and sending them to their destination.. Each streaming technology menti

Trang 2

Tuttle WL040/Bidgolio-Vol I WL040-Sample.cls June 20, 2003 17:30 Char Count= 0

able to use the iconoscope to imitate the ways that

hu-man eyes view images for television broadcast (Inventors

Online Museum, 2002) This technology was a key

com-ponent in the advancement of electronic television and

serves as the foundation for the design of the modern

elec-tronic televisions in use today (Fortner, 2002)

Although television may have provided the foundationfor the technology of streaming video, the Internet has

provided the means that has made it available to

con-sumers in their homes and to businesses The Internet has

revolutionized the computer and communications world

as never before It has become a worldwide medium for

broadcasting, information dissemination, collaboration,

and interaction between individuals without regard to

location (Leiner et al., 2000)

NETWORKING CONCEPTS

Because streaming video is delivered to the user over a

network, it is important to understand the basics of how

the information is handled and transmitted through a

net-work In essence, networking involves one computer

ex-changing information with another computer Most

Inter-net address begins with http:// HTTP stands for hypertext

transfer protocol and is a standard or protocol

(RealNet-works, 2000) It tells a browser and computer that HTML

has been sent to it so it can read the incoming information

In the case of some streaming video locations on the ternet, the addresses start with PNM://, RTP://, or RTSP://

In-PNM stands for progressive networks media and it is an

older protocol However, there are still a number of video

clips in use that use this protocol (RealNetworks, 2000)

RTP stands for real-time protocol, and it is one of the most

commonly used protocols for streaming media on the

In-ternet (Compaq Computer Corporation [Compaq], 1998)

RTSP stands for real-time streaming protocol, which is the

newest protocol (RealNetworks, 2000) In all three cases,

these addresses tell a browser and computer that

stream-ing video has been sent to it It should be noted that any

computer receiving streaming video must have a special

application installed that can read and play the video This

topic will be discussed in more detail in the next section

HOW STREAMING WORKS

Streaming involves taking video or audio ﬁles, breaking

them down into packets of information, and sending them

to their destination At the receiving end, the viewer can

then play the video as it is being downloaded Because

of the way that information ﬂows on the network, it is

easy to see that there would be a number of interruptions

and delays in playing the video To address this issue, a

technology, called buffering, was developed to ensure that

the playing of the video on the receiving end is smooth

Buffering is the process where a large number of

infor-mation packets are collected before the playing of the

video begins Once enough packets have been collected,

the playing of the video will begin As the video plays, the

buffering will continue until all of the information has

been received It is important to note that the video is not

stored on the user’s computer; it is received, buffered, and

played

The process described above is referred to as truestreaming It should not be confused with a method calledpseudo-streaming or progressive download Pseudo-streaming users wait until a significant portion of videofile has been downloaded to their computer before view-ing the video This method allows users to save files tothe hard drives on their computer for later viewing Pro-gressive download works best with very short media clipsand a small number of simultaneous users ( DoIt & WISC,2002)

Streaming video may involve a video with or withoutsound In the case of a video with sound, the visual por-tion of the video is delivered on one stream while the au-dio is delivered on another stream Technology has beendeveloped to synchronize these streams at the destina-tion to ensure that the sound matches up with the ac-tion being viewed Streaming ﬁles that include more thanone medium are known as rich media It should be notedthat streaming can include slide presentations, text, video,audio, or any combination of these

A number of components are required in order to makestreaming video work on the Internet First, the user musthave a computer connection to the Internet via a localarea network or modem The user must also have a Webbrowser with the appropriate video player or plug-in in-stalled Many plug-ins can be downloaded from the Webfor free A plug-in works in conjunction with a browser toplay streaming video ﬁles A Web server stores Web pages

or HTML ﬁles Streaming video ﬁles are usually kept on

a separate dedicated streaming server When a streamingvideo link is clicked on a Web page, the browser reads theHTML code and the lets the player/plug-in take over (DoIt

& WISC, 2002) The player accesses the selected video onthe streaming server using the video protocols (RTP andRTSP) discussed in the previous section After a few sec-onds of buffering, the video will start and play

STREAMING TECHNOLOGIES AND SYSTEMS

A number of technologies are available for streamingvideo The three major technologies are RealOne, Quick-Time, and Windows Media (DoIt & WISC, 2002) Eachstreaming technology has three common hardware and/orsoftware components: (1) servers and video ﬁles; (2) videoplayers and plug-ins; and (3) compression, encoding, andcreation tools (DoIt & WISC, 2002) The speciﬁcs of eachtechnology will be discussed in more depth in a later sec-tion in this paper

Each streaming technology mentioned above may haveits own proprietary server and media file types that theyuse Also, RealOne, QuickTime, and Windows Media havetheir own servers that stream files in their own formats.Therefore, it is important to create video files in a for-mat that are compatible with the technology and serverthat will be used to stream the files However, and rela-tively newer product called Helix offers open, comprhen-sive didital media communication for all players

In order to play the video ﬁle, the user must have thesecond component, the player, installed on their com-puter Users can download the player from the Web forfree or, sometimes, it is included with the browser As

Trang 3

V IDEO S TREAMING

556

with the video ﬁles and servers, each technology may have

its own proprietary player In some cases, one

technol-ogy’s media ﬁles cannot be played by another technoltechnol-ogy’s

player

As indicated above, the third common streaming

tech-nology component is ﬁle creation, compression, and

en-coding It involves the process of creating video ﬁles

for streaming Again, each technology may have its own

proprietary way of creating, compressing, and encoding

streaming video ﬁles Therefore, special software may be

needed to create streaming video ﬁles that are compatible

with the video player on the receiving end

The above discussion has focused on the system

re-quirements for streaming video At this point, it is worth

noting that the typical streaming video system has ﬁve

basic functions First, the video must be captured,

digi-tized, and then stored on a disk Second, after the video

is stored on a disk, it can be edited to improve its quality

and content Third, the video ﬁle must be compressed and

encoded to the appropriate streaming format Fourth, the

video is delivered to the user via the video server And, ﬁfth,

the user receives, decodes, buffers, and plays the video on

the computer

CAPTURING AND DIGITIZING VIDEO

In working with streaming video, the ﬁrst step is to record

the video or obtain a recorded video There are two types

of video that can be recorded The ﬁrst is analog video,

which is produced with a vhs, hi-8, or beta cam format

The second is digital video, which is produced with a

dig-ital recorder or camera (DoIt & WISC, 2002)

Analog video contains video information in frames

consisting of varying analog voltage values It tends to

degrade over time and it can contain imperfections such

as snow in the picture Digital video contains video

infor-mation in a series of digital numbers that can be stored

and transmitted without imperfections Digital video does

not degrade over time The recent advances in the digital

technology make it easier to store, retrieve, and edit digital

video (Compaq, 1998)

If the video is from an analog source, it will have to be

converted and compressed into a digital format In order

to do this conversion, an analog video capture card and

the appropriate software will have to be installed on the

computer The video capture card is an expansion card

that works in conjunction with, or replaces, the

graph-ics adapter inside the computer If the video is digital,

a FireWire capture card can be used and the

analog-to-digital step is not needed (Videomaker Magazine, 2001)

A side note on the digital video format that is

worth-while to review is that digital video often uses a different

color format than the format used for computer monitors

Computer monitors display the color information for each

pixel on the screen using the RGB (red, green, blue)

for-mat (Pixels can be deﬁned as the small elements or points

that make up the frame.) Digital video frequently uses a

format known as YCrCb, where Y represents the

bright-ness (or luma) of a pixel, and Cr and Cb represent the pure

color In the different color schemes used in digital video,

each pixel will have a brightness component but groups

of pixels may share the CrCb color data Hence, the terms

24-bit, 16-bit, and 12-bit color schemes refer to the ber of color bits required per pixel (Compaq, 1998).With the capture and conversion of the video, the video

num-is transferred into a format that can be edited and thenencoded for streaming A number of formats are avail-able One of the most common of these is the AVI format.AVI stands for Audio Video Interlaced and was created

by Microsoft It is one of the oldest formats in use and isincluded with Microsoft’s Windows applications (Fischer

& Schroeder, 1996) This format was used in many of theearly video editing systems and software However, thereare restrictions in using this format; the most notable ofthese is compatibility issues with some of the more ad-vanced editing systems Even with these issues, many edit-ing systems and software can still use this format.Another format is the MOV format, which was origi-nally developed for the Macintosh computer by Apple Itthen became the proprietary standard of Apple’s Quick-Time streaming technology (Fischer & Schroeder, 1996).One of the most recent formats is the MPEG format.MPEG is a newer format and it is becoming very popularwith streaming video users MPEG stands for Motion Pic-tures Experts Group, which is an international organiza-tion that developed standards for the encoding of movingimages (Fischer & Schroeder, 1996) There are a number

of MPEG standards available, primarily for the encodingand compression of streaming video These will be dis-cussed in more detail later in this paper However, one ofthe initial standards that was developed, MPEG-1, is usedfor the storage of video

In capturing and converting video for streaming, it isrecommended to maintain the highest quality video pos-sible The result will be very large video ﬁles that will have

to be edited and streamed However, it is better to startwith the highest quality that can be maintained and thenscale down to the quality that can be streamed Startingwith a lower quality leaves fewer options for editing, com-pression, and encoding

EDITING THE VIDEO

Once the video has been captured and converted to adigital format, it can be edited with a variety of edit-ing tools As mentioned above, each of the three mainstreaming technologies—RealOne, QuickTime, and Win-dows Media—has editing tools Editing is critical as itimpacts how the video is ultimately received by the userand the end user’s needs are paramount (see ProducingStreaming Video for more.)

In editing a video, one of the ﬁrst things that mayhave to be done is cropping the video Cropping thevideo involves removing the edges, where electronic er-rors, glitches, and black bars may be seen These usuallyappear during the process of recording and converting thevideo In most cases, removing about 5% of the edges willeliminate the glitches In cropping a video, it is important

to remember that the ﬁnal dimensions of the video must

be compatible with the encoding technology (Kennedy,2001)

Television systems use a technique called interlacing

to display a picture on the screen This process involvesdisplaying the picture on every other line on the television

Trang 4

screen Then lines are inserted between the ﬁrst set This

process of alternating the picture lines eliminates any

ﬂicker on the screen Videos also have this feature

How-ever, for streaming video that will be displayed on a

computer screen, interlacing is not needed Some

cap-ture cards have a deinterlacing feacap-ture and some

cam-corders will record video without interlacing However, if

the video is interlaced at the editing step, and the ﬁle is

very large, it is advisable to deinterlace the video during

editing (Kennedy, 2001)

Also, when ﬁlm is converted to video, additional framesare added because ﬁlm is shot at 24 frames per second and

depending on the video standard, television may run from

25 to 30 frames per second (Kennedy, 2001) The process

of converting ﬁlm to video, where the additional frames

are put in, is called Telecine It is best to avoid adding

frames that are not needed Therefore, if it is available, an

Inverse Telecine conversion should be used to reduce the

video back to 24 frames per second (Kennedy, 2001)

If a video has been shot with a lot of motion, the videocould appear to be shaky or fuzzy, and not ideal for stream-

ing If this is the case, the best option may to use a still

frame or slow motion A still frame or slow motion may

not look very natural, but it is better than streamed video

that is not viewable

Although special effects are great when viewed in amovie, they do not work well in streaming video because

they utilize a lot of memory and impact the quality of the

video It is generally recommended that special effects be

removed from the video Streaming video is limited in its

ability to deliver smooth video for any motion such as

dance that relies on ﬂuid movements Also, if text is used

in the video, it should be concise, legible, and easy to read

Audio is a very important part of streaming video Ifthe video has an audio portion, the quality of the audio

needs to be reviewed For example, it is advisable to avoid

the use of background music or other noise in order to

ensure that speakers can be heard clearly It is also good

to prepare the audio to work on the worst speaker

sys-tem that any potential user may have If the audio is not

clear then the usefulness of the video is greatly

dimin-ished

BANDWIDTH

Before covering the topic of compressing and encoding, it

is essential to understand the concept of bandwidth The

reason is that bandwidth is a critical factor in the

trans-mission and reception of streaming video Bandwidth

is, simply put, the amount of information that can pass

through a particular point of the wire in a speciﬁc amount

of time (RealNetworks, 2000) Network bandwidth can be

compared to a water pipe and a ﬁle to a tank of water If

the pipe is very narrow, then it will take a long time for the

water from the tank to ﬂow through the pipe If the pipe

is larger, then it will take less time for the water to ﬂow

through (Microsoft.com, 2000) Therefore, the higher the

bandwidth, the greater the amount of information that

can ﬂow through the network to the destination At the

destination, the speed of the modem or other device used

to connect to the Internet determines the bandwidth of

the stream that is received

Table 1 Available Bandwidths

Single Channel ISDN 64 Kbps

Because video ﬁles are large and many networks havelimited bandwidths, there are many issues involved intransmitting these ﬁles over networks Although manycomputer networks have installed new devices and tech-nology to improve their bandwidths, this is one of thebiggest challenges to streaming a video over a net-work The Internet was not designed to handle streamingvideo

File sizes are measured in kilobytes (abbreviated as K

or KB) A kilobyte contains 1,024 bytes When this sion is applied to large video ﬁles and the math is done todetermine transmission rates, it is apparent that these ﬁleshave a huge amount of information that has to be trans-mitted For example, a full-screen, full-motion video canrequire a data transmission rate of up to 216 Megabits persecond (Mbps) (Compaq, 1998) This exceeds the highestavailable data rates in most networks Table 1 shows theavailable bandwidth for several methods of data delivery,according to Compaq (1998)

conver-In reviewing the above exhibit, it should be noted thatthe throughput listed for each technology represents anupper limit for that technology In most cases, the actualthroughput will be below this limit due to the amount oftrafﬁc on the network Depending on the conditions oftheir connections, many users will see their data ﬂuctu-ate up and down One minute, they may have a 10 Kbpsrate; the next minute, it may jump to 24 Kbps (Kennedy,2000) Therefore, it is important for the provider of thestreaming video to match the data rate to the conditionsand limitations of the potential users

Also, the Fast Ethernet and Ethernet technologieslisted in Table 1 are used primarily in businesses and orga-nizations Single channel ISDN (integrated services digi-tal network) is also used by businesses for video phonesand video conferencing Cable modems and ASDL (asym-metrical digital subscriber loops) are available to individ-ual Internet users, but they are newer, more expensivetechnologies and are not as widely available as modems.Thus, it is safe to say that most Internet users have ei-ther a 56-Kbps high-speed modem or a 28-Kbps standardmodem

Two options are available for successfully deliveringstreaming video over networks The ﬁrst option involvesscaling the video to smaller window sizes This is impor-tant for low-bandwidth networks where many clients havemodem access The second option involves compressingthe video using compression algorithms designed for thispurpose This is needed for most networks because of the

Trang 5

558

high bandwidth requirements of videos that have not been

compressed

Scaling and compressing video do affect the quality of

the video The quality of the video is impacted by frame

rate, color, and resolution Frame rate is the number of

still images that make up one second of a moving video

image Images move ﬂuidly and naturally at 30 frames

per second, which is the National Television Standards

Committee (NTSC) standard for full motion video

How-ever, ﬁlm is usually 24 frames per second (Compaq, 1998l)

Videos with a frame rate of less than 15 frames per second

become noticeably jumpy It should be noted that most

phone and modem technology limits the frame rate to 10

frames per second (Videomaker Magazine, 2001)

The second quality variable, color depth, is the

num-ber of bits of data the computer assigns to each pixel of

the frame The more bits of color data assigned to each

pixel, the more colors can be displayed on the screen Most

videos are either 8-bit 256-color, 16-bit 64,000-color, or

24 bit 16.8-million color The 8-bit color is very grainy

and not suitable for video The 24-bit color is the best, but

it greatly increases the size of the streaming ﬁle, so the

16-bit color is normally used (Videomaker Magazine, 2001)

The third quality variable, resolution, is measured by

the number of pixels contained in the frame Each pixel

displays the brightness and color information that it

re-ceives from the video signal The more pixels in the frame,

the higher the resolution For example, if the video is

640× 480, there are 640 pixels across each of the 480

ver-tical lines of pixels Streamed video ranges from postage

stamp size, which is 49× 49 pixels, to full PC monitor

screen, which is 640× 480 pixels, and beyond

(Video-maker Magazine, 2001)

SCALING

As mentioned previously, scaling involves reducing video

to smaller windows For example, this can be

accom-plished by reducing the frame resolution from a full screen

(640 × 480) to a quarter screen (320 × 240) In

addi-tion, frame rate and color depth can also be scaled For

example, the frame rate can be reduced from 30 to 15

frames per second The color depth can be scaled from

24-bit to 16-bit According to Compaq (1998), the process

noted in this example would reduce the video ﬁle size from

216 Mbps to 18 Mbps and the quality of the video would be

reduced However, as can be seen from the available

band-widths shown in Table 1, many delivery methods would

not support a data rate of 18 Mbps Therefore, to further

reduce the data rate, video compression is necessary

COMPRESSING AND ENCODING

The goal of compression is to represent video with as few

bits as possible Compression of video and audio involves

the use of compression algorithms known as codecs The

term codec comes from the combination of the terms

encoder and decoder—cod from encoder and dec from

decoder (RealNetworks, 2000) An encoder converts a

ﬁle into a format that can be streamed This includes

breaking a ﬁle down into data packets that can be sent

and read as they are transmitted through the network A

decoder sorts, decodes, and reads the data packets as they

are received at the destination Files are compressed byencoder/decoder pairs for streaming over a network.Encoders generally accept speciﬁc input ﬁle formatsused in the capture and digitizing process The encodersthen convert the input formats into proprietary streamingformats for storage or transmission to the decoder Somecodecs may be process-intensive on the encode side in or-der to create programs one time that will be played manytimes by the users Other codecs are divided more equallybetween encoding and decoding; these are typically usedfor live broadcasts (Compaq, 1998)

As mentioned above, each of the three major streamingtechnologies has its preferred encoding and compressingformats Many users opt to work with one of these threetechnologies because they are relatively easy to use, andtechnical support is provided by each of the technologies.These technologies provide options to users for selectingvideo quality and data transmission rates during the com-pression and encoding process Depending on the appli-cation and technology used, multiple streaming files mayhave to be produced to match the different bandwidths ofthe networks over which the video is streamed Two of thethree major technologies have advanced options where astreaming file can be produced that has a data transmis-sion rate that will adapt to the varying bandwidths of thenetworks The specifics of these technologies will be dis-cussed in a later section

Even with the dominance of the three major gies, there are some open standards for compression al-gorithms It is important to be aware of these standardsand understand how the compression algorithms work.With this knowledge, the user can make better decisionswhen creating, delivering, and viewing streaming video.The compression algorithms will be discussed in moredetail later However, they all utilize the same basic com-pression techniques to one degree or another Therefore,

technolo-it is essential to review the compression techniques beforediscussing the algorithms

First, compression techniques are either lossless orlossy Lossless compression is a process where data arecompressed without any alteration of the data in the com-pression process There are situations where messagesmust be transmitted without any changes In these cases,lossless compression can be used For example, losslesscompression is typically used on computers to compresslarge ﬁles before emailing them (Vantum Corporation,2001) A number of lossless techniques are available How-ever, for video ﬁles in particular, more compression isneeded than the lossless techniques can provide

Lossy techniques involve altering or removing the datafor efﬁcient transmission With these techniques, the orig-inal video can only be approximately reconstructed fromits compressed representation This is acceptable for videoand audio applications as long the data alteration orremoval is not too great The amount of alternation orremoval that is acceptable depends on the application(Vantum Corporation, 2001)

A number of video compression techniques takeadvantage of the fact that the information from frame toframe is essentially the same For example, a video thatshows a person’s head while that person is talking willhave the same background throughout the video The onlychanges will be in the person’s facial expressions and other

Trang 6

gestures In this situation, the video information can be

represented by a key frame along with delta frames

con-taining the changes between the frames This is known

as interframe compression In addition, individual frames

may be compressed using lossy techniques An example of

this is a technique where the number of bits representing

color information is reduced and some color information

is lost This is known as intraframe compression

Com-bining the interframe and intraframe compression

tech-niques can result in up to a 200:1 compression (Compaq,

1998)

Another compression technique is called quantizing

It is the basis for most lossy compression algorithms

Es-sentially, it is a process where rounding of data is done to

reduce the display precision For the most part, the eye

cannot detect these changes to the ﬁne details (Fischer &

Schroeder, 1996) An example of this type of compression

is the intraframe compression described above Another

example is the conversion from the RGB color format used

in computer monitors to the YcrCb format used in digital

videos that was discussed in the capturing and digitizing

section of this paper

Filtering is a very common technique that involves theremoval of unnecessary data Transforming is another

technique, where a mathematical function is used to

con-vert the data into a code used for transmission The

trans-form can then be inverted to recover the data (Vantum

Corporation, 2001)

For videos that have audio, the actual process used tocompress audio is very different from that used to com-

press video even though the techniques that are used are

very similar to those described above This is because the

eye and ear work very differently The ear has a much

higher dynamic range and resolution The ear can pick

out more details but it is slower than the eye (Filippini,

1997) Sound is recorded as voltage levels and it is

sam-pled by the computer a number of times per second The

higher the sampling rate, the higher the quality and hence,

the greater the need for compression Compressing audio

data involves removing the unneeded and redundant parts

of the signal In addition, the portions of the signal that

cannot be heard are removed

VIDEO COMPRESSION ALGORITHMS

Some algorithms were designed for wide bandwidths and

some for narrow bandwidths Some algorithms were

de-veloped speciﬁcally for CD-ROMs and others for

stream-ing video There are a number of compression algorithms

available for streaming video; this chapter will discuss the

major ones in use today These algorithms are MPEG-1,

MPEG-2, MPEG-4, H.261, H.263, and MJPEG The

video compression algorithms can be separated into two

groups: those that make use of frame-to-frame

redun-dancy and those that do not The algorithms that make

use of this redundancy can achieve signiﬁcantly greater

compression However, more computational power is

re-quired to encode video where frame-to-frame

redundan-cies are utilized

As mentioned in earlier in this paper, MPEG stands forMoving Pictures Experts Group, which is a work group of

the International Standards Organization (ISO) (Compaq,

1998) This group has deﬁned several levels of standards

for video and audio compression The MPEG standardonly speciﬁes a data model for compression and, thus,

it is an open, independent standard MPEG is becomingvery popular with streaming video creators and users.The ﬁrst of these standards, MPEG-1, was made avail-able in 1993 and was aimed primarily at video conferenc-ing, videophones, computer games, and ﬁrst-generationCD-ROMs It was designed for consumer video andCD-ROM audio applications that operate at a data rate ofapproximately 1.5 Mbps and a frame rate of 30 frames persecond It has a resolution of 360× 242 and supports play-back functions such as fast forward, reverse, and randomaccess into the bitstream (Compaq, 1998) It is currentlyused for video CDs and it is a common format for video

on the Internet when good quality is desired and whenits bandwidth requirements can be supported (VantumCorporation, 2001)

MPEG-1 uses interframe compression to removeredundant data between the frames, as discussed in theprevious section on compression techniques It also usesintraframe compression within an individual frame asdescribed in the previous section This compression al-gorithm generates three types of frames: I-frames, P-frames, and B-frames I-frames do not reference otherprevious or future frames They are stand-alone or Inde-pendent frames and they are larger than the other frames.They are compressed only with intraframe compression.They are the entry points for indexing or rewinding thevideo, because they represent complete pictures (Compaq,1998)

On the other hand, P-frames contain predictive mation with respect to the previous I or P frames Theycontain only the pixels that have changed since the lastframe, and they account for motion In addition, theyare smaller than the I-frames, because they are morecompressed I-frames are sent at regular intervals duringtransmission process P-frames are sent at some time in-terval after the I-frames have been sent (this time inter-val will vary based on the transmission of the streamingvideo)

infor-If the video has a lot of motion, the P-frames may notcome fast enough to give the perception of smooth mo-tion Therefore, B-frames are inserted between the I- andP-frames B-frames use data in the previous I- or P-frames

as well as the future I- or P-frames, thus, they are ered bidirectional The data that they contain are an in-terpolation of the data in the previous and future frames,with the assumption that the pixels will not drasticallychange between the two frames As a result, the B-frameshave the most compression and are the smallest of thethree types of frames In order for a decoder to decodethe B-frames, it must have the I- and P-frames that theyare based on; thus the frames may be transmitted out oforder to reduce decoding delays (Comqaq, 1998)

consid-A frame sequence consisting of an I-frame and its lowing B- and P-frames before the next I-frames is called

fol-a group of pictures (GOP) (Compfol-aq, 1998) There fol-are ally around 15 frames in a GOP An example of the MPEGencoding process can be seen in Figure 1 The letters I, P,and B in the ﬁgure represent the I-, P-, and B-frames thatcould possibly be included in a group of pictures The let-ters were sized to indicate the relative size of the frame(as compared to the other frames)

Trang 7

usu-Tuttle WL040/Bidgolio-Vol I WL040-Sample.cls June 20, 2003 17:30 Char Count= 0

Figure 1: MPEG-1 encoding process.

One disadvantage of the MPEG format is that it

can-not easily be edited because video cancan-not be entered at

any point And the quality of the resulting video is

im-pacted by the amount of motion in the video The more

motion in the video, the greater the probability that the

quality will be reduced The MPEG encoding and

de-coding process can require a large amount of

computa-tional resources, which requires the use of specialized

computer hardware or a computer with a powerful

proce-ssor

MPEG-2 was released in 1994 and was designed to be

compatible with MPEG-1 It is used primarily for

deliv-ering digital cable and satellite video to homes It is the

basis of DVD and HDTV MPEG-2 utilizes the same

com-pression techniques as MPEG-1 However, it has been

en-hanced so that it has better compression efﬁciency than

MPEG-1 MPEG-2 supports two encoding schemes

de-pending on the application The ﬁrst scheme has a

vari-able bit rate, which keeps the quality constant The

sec-ond scheme involves varying the quality to keep the bit

rate constant MPEG-2 is not considered an ideal format

for streaming over the Internet because it works best at

transmission rates higher than most networks can handle

(Cunningham & Francis, 2001)

MPEG-4 is one of the most recent video formats and is

geared toward Internet and mobile applications

includ-ing video conferencinclud-ing, video terminals, Internet video

phones, wireless mobile video, and interactive home

shop-ping It was originally designed to support data rates less

than 64 Kbps but has been enhanced to handle data rates

ranging from 8 Kbps to 35 Mbps MPEG-4 is different

from MPEG-1 and -2 in that it has been enhanced to

han-dle the transmission of objects described by shape,

tex-ture, and motion, versus just the transmission of

rectan-gular frames of pixels In fact, it is very similar to H.263,

which is the video conferencing standard (Compaq, 1998)

This feature makes MPEG-4 well suited to handle

multi-media objects, which are used in interactive DVD,

inter-active Web pages, and animations

MPEG-7 is the newest standard It is designed for

mul-timedia data and can be used independent of the other

MPEG standards Work is being done on an extension of

the MPEG-7 standard, called MPEG-21

The H.261 and H.263 standards are designed for video

conferences and video phone applications that are

trans-mitted over an ISDN network H.261 has the ability to

adapt the image quality to the bandwidth of the

trans-mission line The transtrans-mission rate for H.261 is usually

around 64 Kbps (Fischer & Schroeder, 1996) H.263 was

developed as an enhancement to H.261 and was designed

to support lower bit rates than H.261 It has a higher

pre-cision for motion compensation than H.261 H.263 is very

similar to the MPEG standards, particularly MPEG-4, and

uses the same compression techniques (Vantum tion, 2001)

Corpora-MJPEG stands for Motion JPEG, and JPEG stands forJoint Photographic Experts Group JPEG is an interna-tional standard for compressing still frames MJPEG is asequence of JPEG compressed still images that represent

a moving picture Thus, MJPEG is a compression methodthat is applied to each frame without respect to the pre-ceding or following image (Vantum Corporation, 2001).MJPEG can be edited easily but it is not able to handleaudio

AUDIO COMPRESSION ALGORITHMS

Each of the three major streaming technologies has itspreferred algorithms for compressing audio In addition,the MPEG group has deﬁned an audio standard calledMPEG-1 for audio As discussed previously, audio com-pression is different than video, although it uses similartechniques The MPEG audio compression uses psychoa-coustic principles, which deal with the way the humanbrain perceives sound (Filippini, 1997)

The ﬁrst principle utilized in the MPEG audio sion is the masking effect This means that weak soundsare not heard, or they are masked, when they are near astrong sound For example, when audio is digitized, somecompression occurs because data are removed and noise

compres-is added to the audio Thcompres-is nocompres-ise can be heard during silentmoments, or between words or sentences However, thisnoise is not heard during talking or when music is playing.This is because the noise is a weaker sound and is masked

by the louder talking or music MPEG uses this maskingeffect to raise the noise floor around a strong sound be-cause the noise will be masked anyway And, by raising thenoise floor, fewer data bits are used, and the signal (or file)

is compressed MPEG uses an algorithm to divide up thesound spectrum into subbands It then calculates the op-timum masking threshold for each band (Filippini, 1997).The second psychoacoustic principle is that the humanear is less sensitive to high and low frequencies, versusmiddle frequencies In essence, MPEG employs a ﬁlteringtechnique along with the masking effect to remove datafrom the high and low frequencies where the changes willnot be noticed It maintains the data in the middle fre-quencies to keep the audio quality as high as possible

DELIVERING THE VIDEO

Once the video has been compressed and encoded forstreaming, the next step is to serve the video to the users

on the Internet As discussed earlier in this chapter, ering video over the Internet is usually accomplished with

deliv-a stredeliv-aming server, instedeliv-ad of deliv-a Web server A stredeliv-aming

Trang 8

R ECEIVING , D ECODING , AND P LAYING THE V IDEO 561

server has some specialized software that allows it to

manage a data stream as it is being transmitted through

the network It utilizes the streaming protocols (RTP and

RTSP) to transmit the video ﬁle

A Web server can be used to stream video, but it hasbeen designed to transfer text and images over the Internet

and it does not have the means to control a stream (Strom,

2001) When a Web server is used, a user selects a video ﬁle

and it starts to be copied down to the PC using HTTP like

any other data source on the Internet The player takes

control and the video is buffered and played But because

the Web servers are not able to control the stream, the

delivery of the video can be erratic and the user could

experience rebuffering interruptions Thus, it is best to use

a video server to ensure that the user will have a smooth

playback without interruptions

Video servers have capacity limitations, and they canonly deliver a certain number of streams at any one time

The capacity of a server is measured in the number of

si-multaneous streams that it can put out at any given point

in time This can range from 20 to 5,000 or more,

depend-ing on the type of server (DoIt & WISC, 2002) If a user

tries to access a video ﬁle after the server has reached its

maximum capacity, the user will get a message stating that

the server is busy and to try playing the video again after

1 or 2 minutes

It is essential to note that streaming servers require theappropriate hardware, network connections, and techni-

cal expertise to set them up and administer them This can

consume time and resources, so many people and

busi-nesses choose to outsource this task to a host A host is

an agent or department that has the facilities and

techni-cal expertise to serve other people’s streaming videos and

other media content (DoIt & WISC, 2002) Hosts usually

charge a fee for their services There are numerous hosts

that advertise on the Internet When selecting a host, it is

important to ensure that they can support the streaming

technology being used by the client

When using a host, the client will be able to transfermedia ﬁles from his or her local computer to a stream-

ing server This is usually done by using special software

called an FTP client (DoIt & WISC, 2002i) The host will

set the person up with a password-protected account and

a designated amount of server space With this situation,

the person may have text and graphics for a Web site

re-siding on a Web server Then, he or she has streaming

ﬁles on a streaming server This can be managed by using

certain HTML tags on the Web page that will trigger and

control the playback of the media ﬁles from the streaming

server This involves specifying the path of the particular

video ﬁle on the streaming server Each of the three

ma-jor streaming technologies has its own unique embedded

HTML tags for controlling the video ﬁles on servers Many

of the encoding applications can generate these HTML

tags (DoIt & WISC, 2002)

As covered earlier in the discussion on bandwidth, notall networks are suited for the streaming of video Video

works best when the bandwidth of the network is

contin-uously high However, when the bandwidth of the video

exceeds that of the network, delays in the transmission

of the data packets can occur These delays will cause

the picture to ﬂicker and the audio (if present) to start

Server Router Network Router

Client PC’s

Figure 2: Video on demand.

and stop In order to deal with the issues of streamingvideo and media, a new measure of network capabilityhave been developed It is called quality of service (QoS)(Compaq, 1998) Networks that have a good QoS measureprovide a guaranteed bandwidth with few delays The net-works that have the best QoS are those that have dedicatedconnections for streaming

Another network characteristic that needs to be ered for streaming video is the network’s ability to supportvideo-on-demand delivery and webcasting delivery Withvideo on demand (also know as unicasting), a stream isdelivered onetoone to each client, and the user can requestthe video at any time This type of delivery can consume alot of network bandwidth, depending the number of usersrequesting a video According to Compaq (1998), Figure 2shows a simple diagram of how video on demand works.Each line in the exhibit represents a separate stream.Webcasting, is used for live events where there can bepotentially many viewers Webcasting delivers one stream

consid-to many clients simultaneously It does not consume asmuch bandwidth as video on demand But as noted pre-viously, video on demand is much more common because

of the convenience it offers to users Webcasting is uled for speciﬁc times and requires a lot of effort and re-sources to coordinate Networks that support webcastingmust have routers that are multicast capable Figure 3shows how a webcast works, according to Compaq (1998).The lines in the exhibit represent the video stream (notethe single line going across the network)

sched-RECEIVING, DECODING, AND PLAYING THE VIDEO

Finally, at the client desktop, the user accesses the videofile As discussed above, the user clicks on the video filethat he or she wants to view, the request is routed tothe appropriate file on the video server, and the playertechnology on the user’s PC takes control of the data

Server Router Network Multicast

Router

Client PC’s

Figure 3: Webcasting.

Trang 9

562

transmission The player buffers the stream(s), decodes

the data packets, and converts the information back to

analog so the video can be viewed The player usually

has functionality that will allow the user to play, pause,

rewind, and fast forward The player technology must be

compatible with the streaming technology in order for the

user to view the video

With video on demand, the user can control access to

the video He or she can start, stop, rewind, and so forth at

will Although this freedom is desirable to the user, it does

consume bandwidth on the network (as noted above)

With webcasting, the user can only watch the video stream

as it is being transmitted; he or she does not have any

con-trol over the stream Webcasting does not use as much

bandwidth as video on demand

PRODUCING STREAMING VIDEO

Up to this point, this discussion has covered the aspects

of streaming video after it has been created or produced

However, there are some techniques that should be used

when used when producing streaming video that will

make the capturing, editing, compressing, and encoding

processes go much smoother Because streaming video

has to be compressed before delivery on the network, one

of the most important things to remember when

produc-ing the video is to minimize motion and changes in the

ob-jects or people in the video The more motion and change

there is in the video, the more the video will have to be

compressed and thus data (such as ﬁne details and color)

will be altered or removed

Therefore, it is best to use a tripod or other method

to anchor the camera whenever possible If the camera is

held in the video producer’s hands, all of the hand

move-ments will be incorporated into the video The video

pro-ducer should also avoid panning the camera as much as

possible and avoid zooming in and out on a scene Thus,

eliminating the movement of the camera and keeping

zooming in and out to a minimum will prevent changes

from being introduced into the video

The video producer should also try to keep the

back-ground as simple and consistent as possible The producer

should avoid trees, buildings, and so forth that will add

complexity to the video, which will mean more data to

compress In addition, the producer should try to stay as

close to the subject as possible when shooting the video

There may be some temptation to choose a wide shot of

the scene however, will viewed online, the video will seem

fuzzy It is important to remember that the compression

will remove a lot of the ﬁne detail of the wide shot

Last, the video producer should use an external

micro-phone whenever possible With an external micromicro-phone,

the producer can keep it as close to the subject as possible

to get good quality audio With good quality audio, the

audio compression will work much better Audio is just

as important as the images being displayed in the video

VIDEO STREAMING USES

The previous sections have focused on the technical

as-pects of creating, delivering, and playing streaming video

This section will focus on the many uses of streaming

video and the preferences of users First, streaming dia (video along with audio) have grown rapidly over thelast few years The number of Internet sites transmittingstreaming video grew from 30,000 in mid-1998 to 400,000

me-by late 1999 The Net Aid concert in October, 1999, set aworld record for the largest Internet broadcast event for asingle day—2.5 million streams The BBC Online’s Euro-pean solar eclipse site served a million streams in a day inAugust, 1999 The BBC estimated that its streaming audi-ence was growing by 100% every 4 months (Tanaka, 2000)

As can been seen from the above statistics, streamingvideo continues to gain in popularity even with the tech-nical challenges involved in streaming the video over theInternet A Web survey conducted by Tanaka (2000) indi-cates that streaming video appeals to users because theycan select what they want to view when they want to view

it Users like the fact that streaming technology has madespecialized or unique videos or other or media available

to them

The streaming video uses fall into the primary egories of entertainment, news/information, education,training, and business Entertainment was one of the ear-liest uses of streaming video and still remains the pri-mary use of streaming video today Entertainment covers

cat-a wide rcat-ange of medicat-a including movies, music, cat-and TVshows There are numerous Web sites promoting free andpay-per-view movies Many sites feature independent ﬁlmmakers, foreign ﬁlms, and pornography (Bennett, 2002).Pornography video sites are some of the oldest entertain-ment sites on the Internet At this time, pornography may

be the largest online movie market of all on the Web(Bennett, 2002)

Recently, some Web sites have been established thatshow hit movies on the Internet For example, viewerscan watch the blockbuster movies on a Web site for $3.95(Graham, 2002) The Hollywood studios have been slow toutilize the Internet as a medium for showing their moviesbecause they want to be sure that this is a safe way to de-liver their ﬁlms It is interesting to note that many moviesare available on the Internet in unauthorized versions.Many of these movies were copied from DVDs or shot on

a camcorder in a theater and then traded on ﬁle-sharingsites such as Morpheus and Kaaza (Graham, 2002).According to Graham (2002), users need a high-speedInternet access, such as cable or ADSL, in order to watchmovies over the Internet Even with high-speed access,users may experience stutter, or stopping and starting, ifthere is a lot of trafﬁc on the network Users will be able

to view the movie only on a partial or full PC screen sizewindow

Development in the streaming video world has beenthe integration of over-the-air and online entertainmentprograms For example, in November, 1999, ABC.com andWarner Brothers Online hosted a simulcast of an episode

of the Drew Carey Show The television audience watchedDrew’s daily activities, while the Internet audience sawfootage of what was happening in his home when he wasout at work ABC indicated that approximately 650,000streams were served (Tanaka, 2000)

In the news/information category, many users like

to utilize the Internet to view video clips of domesticnews and international news items Other users tend

Trang 10

to gravitate toward sites that provide clips on sporting

events For example, the on-air rating of the JFK Jr

tragedy was only 1.4, while over 2.3 million streams were

delivered from the CNN.com Web site (Tanaka, 2000)

The use of streaming video in the education and ing areas has been rapidly growing for the last few years

train-Universities and colleges, in particular, have been

explor-ing the use of streamexplor-ing video for their distance

learn-ing programs Distance learnlearn-ing has become popular

be-cause many people who have been in the work force for

a few years are returning to school to obtain an advanced

degree, pursue a career change, or upgrade their skills

Many opt for distance learning programs because of their

work or travel schedules, or because the academic

pro-grams they desire are not available locally Because of the

growth of distance learning programs, many colleges and

universities have started to use streaming video as an

al-ternative to mailing out VCR tapes, which can be

cum-bersome With streaming video, these institutions can

ex-pand their distance learning programs to meet the needs

of their students In addition, there are many Web sites

that offer training and tutorial programs on a variety of

subjects

A common presentation method used by educators is alecture that includes static slides These is must easier to

create and will provide good quality sound and images for

those students who have modem connections There are

a number of software tools that can be used to combine

PowerPoint slides with narrations to create streaming

pre-sentations

In view of the above discussion, it should also bepointed out that streaming video is used for teaching ma-

terial that involves motion or dynamic interaction Some

examples of this include medical or laboratory

proce-dures, processes in the physical sciences, interpersonal

skills, and illustrations of real world events or activities

(DoIt & WISC, 2002) In addition, live training or teaching

webcasts are produced using audio, slides, or video The

participants access the Web site from their computers

In-teraction between the instructor and participants occurs

in real time The participants can use a chat window to

type in questions to the presenter during the session (DoIt

& WISC, 2002) These events are very challenging to

coor-dinate and deliver and are not as common as illustrated

audio presentations

Businesses and companies are starting to use ing videos for advertising and communications Some

stream-businesses have started to webcast their products in order

to improve their sales One of the most talked about events

was the Victoria’s Secret fashion show that was

web-cast in February, 1999 (Tanaka, 2000)

Another form of advertising that has become ingly popular is the video banner ad (Tanaka, 2000) This

increas-technology involves using a program that detects whether

or not the client PC has a streaming media player, and

then determines the type if there is one present This is

done before the user clicks on the Web page Once the

user clicks on the Web page, video is played using the

me-dia player on the PC If there is no meme-dia player on the

PC, then a regular GIF banner is displayed

Businesses are also using streaming media to cast presentations, corporate meetings, and in-house

broad-seminars to their employees Many companies are ﬁndingthat this is less expensive than live meetings and seminars,where travel expenses are incurred And it offers oppor-tunities for communication that would not otherwise beavailable For example, a company that uses streamingtechnology may choose to broadcast an industry analysts’meeting or public relations event that, without this tech-nology, would not be feasible to do

THE BIG THREE STREAMING TECHNOLOGIES

As mentioned previously in this chapter, there are threemajor technologies for streaming video: RealOne, Quick-Time, and Windows Media These three players provideall of the tools needed for streaming video, including ap-plications for creating, editing, compression, encoding,serving, and playing Of these three, RealOne is the old-est and still the most widely used (Sauer, 2001) RealOneclaims that they have over 70% of the Internet stream mar-ket with their player being installed in over 90% of homePCs (Cunningham & Francis, 2001) The RealOne technol-ogy supports over 40 media formats and employs the lat-est generation of encoding and compression techniques

They have also developed a technology, called Surestream,

that utilizes an automatic bit-rate technology to adjust thedata stream rate to the bandwidth characteristics of theuser (Cunningham & Francis, 2001)

RealOne has developed some strategic partnershipsthat may give it a competitive advantage for the nearfuture First, RealOne now supports Apple’s QuickTimetechnology And it is working with the National Basket-ball Association and the Major Baseball League on a pay-per-view model (Cunningham & Francis, 2001) However,only the basic player and server versions are free; the moreadvanced server and productions tools available fromRealOne can cost up to several thousand dollars Stream-ing is ReadMedia’s core business and they must chargefees for the use of their applications, whereas theircompetitors can incorporate their streaming technologyinto other products they sell, such as operating systems(Cunningham and Francis, 2001)

Also, RealOne is SMIL compliant SMIL stands for chronized multimedia integration language and it pro-vides a time-based synchronized environment to streamaudio, video, text, images, and animation (Strom, 2001).SMIL is a relatively new language available to streamingusers It is the ofﬁcially recognized standard of the WorldWide Web Consortium (Strom, 2001) SMIL has attracted

syn-a lot of syn-attention becsyn-ause of the fesyn-atures syn-and ﬂexibility itoffers to users

QuickTime was developed by Apple in 1991 and it isone of the oldest formats for videos that are downloaded

It is the one of the recent entrants into the streamingvideo market (Sauer, 2001) One of the advantages thatQuickTime offers is that it can support different com-pression techniques, including those used by RealOne,

as noted above QuickTime also features an open plug-infunction that will allow the utilization of outside compres-sion techniques (Cunningham & Francis, 2001) It is alsoSMIL compliant (as noted above for RealOne) Quicktime

is available in the Apple MAC operating system But it

Trang 11

564

offers a basic player and server tools that are compatible

with other operating systems for free It does charge a fee

for the more advanced systems Those advanced systems

have been used by many in the cinematography ﬁeld for

editing purposes

Windows Media was developed by Microsoft

Corpora-tion and it is a newcomer into the streaming video market

Since its introduction, it has been rapidly gaining ground

on the other two technologies Microsoft includes

Media-Player as part of its Windows operating system, which

can be a convenience for users However, MediaPlayer

is limited in its ﬂexibility in that it has its own

propri-etary compression methods and it does not support many

of the compression techniques utilized or developed by

other companies It does have a MPEG-4 type of

com-pression algorithm and its proprietary comcom-pression

meth-ods are considered to very good Microsoft does have the

player and server tools available as free downloads from

the Internet And it has developed a technology called

Microsoft’s Intelligent Streaming that is like RealOne’s

Surestream It allows the user to put multiple tracks,

each with a different bit rate, into a single streaming

ﬁle This will allow the streaming ﬁle to adjust to

ﬂuctua-tions in the network’s bandwidth (Cunningham & Francis,

2001)

OTHER STREAMING VIDEO SYSTEMS

The previous section covered the major technologies in

the streaming video arena However, a number of other

key players have contributed to the growth of the

stream-ing video ﬁeld The ﬁrst of these is a company called

Sorenson Media, which specializes in compression

tech-nologies They are known as having the highest quality

video compression, particularly for high motion at low

data rates (Segal, 2002a) Sorenson Media also developed

a professional version of their codec and a live

broadcast-ing tool for Quicktime And they have entered the hostbroadcast-ing

and streaming markets (Segal, 2002a) In fact, they were

asked by the Church of Jesus Christ of Latter Day Saints

to host and stream their semiannual conferences live and

then archive the conferences for them (Segal, 2002a)

And they worked with other companies, such as

Macro-media, to build products that will incorporate streaming

media

Video hosting is another area that has several players

There are a number of companies that specialize in video

hosting Many offer services for each of the three major

streaming technologies as well as other independent

tech-nologies However, a potential user of video hosting would

need to do some in-depth research before choosing a host

In addition, there are companies that specialize in

on-line broadcasting for TV news and programming In fact,

some of the major networks have their own online

broad-casting Web sites These include CNN, ABC, and BBC In

addition, some local TV networks in the larger

metropoli-tan areas have their own Web sites There are a few

in-dependent companies that specialize in online

broadcast-ing One that was reviewed was Servecast, which indicated

that they would provide services for sporting events and

other media They also indicated that they could provide

content protection

For streaming video creation, editing, and encoding,

a number of independent technologies are available Forexample, Sonic Foundry has a tool that provides for thecreation of streaming content in RealOne and MicrosoftMedia formats And Terran has a tool called media cleanerthat provides a complete set of tools for preparing videoand audio for the Web It is considered the industryleader in this ﬁeld (Cunningham & Francis, 2001) Addi-tional developments include Heliz “Helix an open sourcedigital-media delivery platform designed to let compa-nies build custom applications that stream any mediaformat on any major operating system to any computingdevice” (RealNetworks, 2003)

In addition to creation, editing, and encoding tools,

a number of companies provide the means to check thebandwidth of a video ﬁle as it is being encoded and com-pressed Terran has a function that will graph the datarate of a video Macromedia Flash also has a tool calledBandwidth Proﬁler that will graph streaming data rates(Kennedy, 2000)

DEVELOPMENTS AND TRENDS

There have been some developments and trends ring in the streaming video ﬁeld Most of these are geared

occur-to providing new technology, increasing network width, improving video quality, and competing againsttelevision In the new streaming technology area, therehave been some recent efforts to develop and introducestreaming technology into the wireless networks Toshibahas developed a chip with an MPEG-4 encoder that al-lows third-generation mobile networks to support two-way videoconferencing (Williams, 2001)

band-Also, another player in this wireless area is Thin media, a company that specializes in wireless multime-dia streaming and video messaging (Segal, 2002b) Theyprovide software tools for encoding, decoding, author-ing, messaging, and streaming that can be installed onsecond-generation mobile networks They are also devel-oping tools for third-generation mobile networks Some

Multi-of the features Multi-of their products include streaming video

on a cell phone and having the capability to tap into alive feed from a Web site (Segal, 2002b) Another applica-tion that Thin Multimedia has is video mail product Withthis product, people can create videos of themselves, while

on a Web site and using a webcam, and then send them

to someone else’s cell phone (Segal, 2002b) The videoscreens that display on the phones are small, 112× 96 pix-els According to Thin Multimedia, certain media forms,such as movies and wide trailers, may not work well But,other forms such as video mail, trafﬁc reports, and news

do work well (Segal, 2002b)

In the area of network bandwidth and performance,the business sector has developed strategies to deal withthe network bandwidth and congestion issues One of thestrategies involves the development of Content DeliveryNetworks (CDNs), which are networks that have the in-frastructure and technology to enable a faster and moreconsistent delivery of streamed media to users The goal

is to reproduce media content and deliver it to the user in

an efﬁcient and straightforward manner (Cunningham &Francis, 2001)

Trang 12

There are several commercial CDNs established that fer their services to Internet users One of the key players is

of-Akamai, with their FreeFlow technology There are other

players that have networks and technology to deliver

me-dia to users Both players claim a tenfold increase in speed

by distributing media content to their world-wide

net-works of servers (Cunningham & Francis, 2001) In

ad-dition to the commercial CDNs, some efforts to

experi-ment with this concept are being undertaken by research

groups

The major streaming technologies are continuouslyworking to extend their products to provide digital rights

management (another phrase for copy protection)

fea-tures, interactivity, e-commerce hooks, and better video

quality (Bennett, 2002) For example, Microsoft has

devel-oped a new streaming format called ActiveMovie

Stream-ing Format (ASF) This format allows multiple data

ob-jects to be combined and stored in a single synchronized

multimedia stream The data objects include audio, video,

still images, events, URLs, HTML pages, and programs

(Bennett, 2002) This format supports digital rights

man-agement and pay per view, and, as mentioned before, it

is SMIL compliant, which will allow Web authors to

cre-ate clickable movies (Bennett, 2002) RealOne is working

to develop partnerships with other companies, including

Apple and Microsoft, to provide greater ﬂexibility in

streaming different media formats

Streaming media companies have turned their tion to the television market Many of the companies

atten-have been developing technology designed to handle

pro-gramming applications And some companies have

al-ready begun offering broadband services to users that

have high-speed Internet access (Tanaka, 2000) For

example, MeTV.com and LikeTelevision.com currently

of-fer opportunities to view TV programs, movies, and so

forth

Some companies are forming alliances or partnerships

to build streaming video networks designed to provide

broadcast-scale streaming media It seems that the trend

for streaming media providers will be to make broadband

content delivery available for high-speed users, and retain

low bit rates for dial-up users (Tanaka, 2000)

CONCLUSION

In reviewing the above materials, it become clear very

quickly that streaming video is a complex, technical

pro-cess Besides the sheer complexity of streaming video,

there are other issues such as copyright usage

However, even with technical complexity and tent information, streaming video is an exciting topic It

inconsis-does provides advantages to user by allowing them to

be-gin playing video without having to completely download

it beforehand And it is surprising to see the number of

applications available and the large number of uses for

the technology

In reviewing the history of streaming video, it becameobvious how closely streaming video and TV are tied to-

gether in their technologies and in new developments

in the information and media areas, and how streaming

media themselves would have not come into being if it

were not for the development of the Internet Interestingly

enough, the data transmission or bandwidth limitations

of the Internet remain one of the biggest challenges tostreaming video over a network The Internet simply wasnot designed for streaming media

With the bandwidth limitations of the Internet in mind,

a lot the streaming media technology is focused on ways

to efﬁciently deliver video over a network First, there isthe raw video that must be captured and digitized into theappropriate input ﬁle format Then the video must be en-coded and compressed into the proper streaming format.Next, the video is delivered over the Internet from a spe-cial server, called a video server The user then receivesand plays the video The process sounds simple but theactual functions are very complex, as can be seen fromthe discussion of compression techniques and algorithmsthat was included in this chapter

In facing these technical challenges, the major ing technologies, RealOne, QuickTime, and WindowsMedia, have implemented some very good tools that make

stream-it possible for a person wstream-ithout streaming media tise to create, encode, and play simple videos Also, thesetechnologies have continued to improve and expand theirproducts and have attracted even more users, as well asbusinesses and educational institutions

exper-With the continued growth of the streaming videoarea, other players have entered the market and pro-vided specialized tools for editing and managing band-width requirements Some players offer services for host-ing streaming videos, and others provide consultingservices for the entire process all the way from creating

to playing videos Many of the players are working to prove the network delivery process, in order to improvethe efﬁciency of streaming video and improve the quality

car-Buffering Compensating for a difference in rate of ﬂow

of data, or time of occurrence of events, when ring data from one device to another

transfer-Digitizing A process of converting any graphic medium

to digital format, so that computerized equipment canread, store, transmit, and recreate it

Encoding Conversion of input formats into proprietarystreaming formats for storage or transmission to a de-coder for streaming media

MPEG Motion Pictures Experts Group, an internationalorganization that developed standards for the encoding

of moving images

PNM Progressive networks media, an older protocol

RTP Real-time protocol, one of the most commonlyused protocols for streaming media on the Internet

Streaming Video A sequence of “moving images” thatare sent in compressed form over the Internet and dis-played by a viewer as they arrive Streaming media arestreaming video with sound

Webcasting Videos of speciﬁc live events are shown at

a predetermined time to many viewers Webcasts are

Trang 13

Bennett, G (2002) Internet video trends Retrieved

May 4, 2002, from The Techno Zone Web site:

http://thetechnozone.com/videobuyersguide/Net Video

Trends.htm

Compaq Computer Corporation (1998) Streaming video

technology Retrieved December 31, 2001, from

White Papers Web site: http://www.itpapers.com/cgi/

psummaryIT.pl?paperid = 4045&scid = 191

Cunningham, D., & Francis, N (2001) An introduction to

streaming video Retrieved April 20, 2002, from

Culti-vate Interactive Web site: http://www.cultiCulti-vate-int.org/

issue4/video/

DoIt & WISC (2002) Streaming media [tutorial on

streaming media developed by DoIt and University of

Wisconsin-Madison] Retrieved April 20, 2002, from

http://streaming.doit.wisc.edu/tutorial/

Filippini, L (1997) MPEG-1 audio Retrieved April 20,

2002, from CRS4 Web site: http://www.crs4.it/∼luigi/

MPEG/mpeg1-a.html

Fischer, B., & Schroeder, U (1999) Part 3—Video formats

and compression methods Retrieved September 24,

1999, from Tom’s Hardware Guide Web site: http://

www6.tomshardware.com/ video / 99q3 / 990924

/video-3–02.html

Fortner, B (2002) Section 1: Points from the past; History

of television technology Retrieved March 25, 2002,

from Communication Using Media Instructor’s Notes:

http://www.rcc.ryerson.ca/schools/rta/brd038/clasmat/

class1/tvhist.htm

Graham, J (2002, March 5) Video on demand’s supply

grows USA Today, Section D, p 6.

Inventors Online Museum (2002) Inventing television

Retrieved March 18, 2002, from Inventors Online

Mu-seum Presents History of the Invention and

Inven-tors of Television: http://www.invenInven-torsmuseum.com/

television.htm

Kennedy, T (2000) Don’t be scared of bandwidth math

Retrieved January 12, 2002, from Streaming

Me-dia World Web Site: http://www.streamingmeMe-diaworld

com/symm/tutor/bandmath/index.html

Kennedy, T (2001) Streaming basics: Editing video for

streaming Retrieved March 31, 2002, from Streaming

Media World Web site: http://smw.internet.com/video/

Jan-Microsoft.com (2001a) A brief history of the Internet.Retrieved January 1, 2002, from Microsoft Insider,:http://www.microsoft.com/insider/internet/articles/history.htm

Microsoft.com (2001b) The digital media revolution.Retrieved December 31, 2001, from Windows Me-dia Technologies: http://www.microsoft.com/windows/windowsmedia/overview/default.asp

RealNetworks (2000) Chapter 6: What is streaming dia and how does it work? Retrieved January 12, 2002,from Real Networks Web site: http://service.real.com/help/player/plus manual.8/htmﬁles/whatisrpp.htmRealNetworks (2002) Announcing Helix on July 22,

me-2002 Retrived April 16, 2002, from http://www.realnetworks.com/solutions/leadership/helix.htmlSegal, N (2002a) Sorenson Media: Video compressionsoftware Retrieved May 4, 2002, from Streaming Me-dia World Web site: http://streamingmediaworld.com/video/docs/sorenson/

Segal, N (2002b) Thin multimedia: Wireless streamingvideo Retrieved May 4, 2002, from Streaming Me-dia World Web site: http://streamingmediaworld.com/videos/docs/thin/index.html

Strom, J (2001) Streaming video: A look behind thescenes Retrieved May 4, 2002, from Cultivate In-teractive Web site: http://www.cultivate-int.org/issue4/scenes/

Tanaka, K (2000) Motion pictures on the Net: ing media industry, technology, and early adopters.Retrieved January 1, 2002, from Internet Society Website: http://www.isoc.org/inet2000/cdproceedings/4c/4c 2.htm

Stream-Vantum Corporation (2001) In-depth comparison ofvideo CODECs [White Paper] Retrieved March 28,

2002, from Vantam Corporation Web site: http://www.vantum.com/pdf/codecs.pdf

Videomaker Magazine (2001) Streaming video primer.Retrieved December 31, 2001, from Chaminade Col-lege Preparatory Web site: http://www.chaminade.org/mis/Articles/StreamingVideo.htm

Williams, M (2001) New chip could bring video to mobilephones Retrieved January 17, 2001, from CNN.comWeb site: http://www.cnn.com/2001/TECH/computing/01/17/mpeg4.chip.idg/index.html

Trang 14

WL040C-192-Cecil WL040/Bidgolio-Vol I WL040-Sample.cls September 14, 2003 18:14 Char Count= 0

Importance of the Emerging VE Model and

Potential Beneﬁts of Adopting the VE Model 570

Problems and Challenges in the Creation of VEs 570Technologies and Frameworks for the

Computer Architectures and Technologies ThatSupport the Realization of VEs 571Industry and University Initiatives Related to

lo-a neighboring stlo-ate From time to time, lo-a dent control computer ﬂashes a signal informingthe human manager of the parts being assembledand other work in progress including those beingpacked and shipped

resi-This scenario provides a snapshot of the future of

manu-facturing in this country and throughout the world With

remarkable advances in information technology,

com-puter networks (especially the Internet), and

manufactur-ing integration, the achievement of a truly global

man-ufacturing enterprise seems to be within reach Smaller

and mid-sized enterprise in remote parts of the world will

increasingly become part of this revolution by forming

partnerships with larger organizations The notion of such

“virtual” partnerships, in which distributed organizations

form “virtual teams” and develop products for a changing

customer-driven market forms the basis of virtual

enter-prises (VE)

This chapter provides an overview of the concepts,techniques, technologies, and issues that need to be un-

derstood, adopted, and studied in the quest to realize a

truly virtual enterprise-oriented approach to product

de-velopment The underlying theme is the role of the

In-ternet in the design and realization of such virtual prises Some of the challenges and hurdles, which may

enter-be encountered by industrial organizations during mentation, are also delineated in this chapter Other sec-tions of the chapter include discussions of some of theInternet-based, computer-based frameworks for VE real-ization, modeling, and communication techniques for VEcollaboration and a summary of various industry and uni-versity projects related to this subject area

imple-Concept of a Virtual Enterprise

Today, the concept of a virtual enterprise is being widelyheralded as a collaborative partnership for the future as itholds distinct advantages and benefits for organizationsworldwide Formally, “a VE can be described as a consor-tium of industrial organizations, which come together toform temporary partnerships to respond quickly to chang-ing customer demand” (NIIIP, 2002) In a VE, the partnerorganizations are geographically distributed, possess di-verse skills and resources, and collaborate virtually to pro-duce a final product (Figure 1) In a traditional (non-VE)enterprise, the team members are co-located physically inone specific site and usually belong to the same organiza-tion

In this decade and beyond, it is predicted that growingproduct complexity and resultant diverse skill require-ments underscore the need for organizations to worktogether as a VE More importantly, such a collaborativeframework will enable the harnessing of remote and far-ﬂung manufacturing facilities (and resources) and createnew opportunities for these remotely located organiza-tions who can becomes partners and pillars of the Ameri-can and international industrial base Small and medium-sized “mom and pop” operations with specialized capabil-ities can link with the industrial mega-giants or with othersimilar-sized enterprises to produce a diverse mix of prod-ucts beginning to typify the evolving global market Forthis reason and several others, American and other inter-national industrial organizations have shown keen inter-est in virtual enterprise-related principles and practices

567

Trang 15

V IRTUAL E NTERPRISES

568

y

Suppliery

Project MgmtProduct Design

Process PlanningManufacturing

Testing

Assemblyy

SupplierSupplier

Figure 1: Intra-nation VE collaboration.

Characteristics of a Virtual Enterprise

The concept of a VE is not new However, very few

partner-ships can truly claim to have functioned as a virtual

enter-prise In this context, it is important to highlight a few

as-pects related to the historical development of VEs There

are various deﬁnitions and connotations of a “virtual

en-terprise.” Numerous reports and articles in newspapers

and magazines tout the “functioning” and “success of long

running implementations” of VE-based practices A

ca-sual reader who peruses the numerous VE-related articles

(that have appeared in the media) might conclude that VE

implementation is commonplace and such a model has

existed for decades The key misconception relates to the

degree of “seamlessness” of the information exchange in

a VE The questions that must be asked by every reader

interested in this topic are as follows:

(1) Is the information merely being exchanged by two

partners or organizations by e-mail or the World Wide

Web?

(2) Is the information being re-processed or used to

per-form some engineering or technical task?

(3) Is the information exchange seamless without manual

re-keying of information?

(4) Is the information sharing mainly restricted to

keep-ing business partners informed of new or occurrkeep-ing

activities?

There are distinctive differences between partnerships

that function as “true VEs” and others that are “quasi-VE.”

The criteria (which can serve as a litmus test) by which

organizations can claim that they have implemented

VE-based practices include the following:

The partners involved must belong to different

organiza-tions and have different core areas of expertise (for

ex-ample, Company P may have manufacturing expertise

while Company F may possess skills in testing

prod-ucts);

The partners must be geographically distributed;

The computer platforms used by the partners must be

het-erogeneous in nature (having different operating

sys-tems such as Windows, UNIX, or Macintosh);

The software systems used in the collaboration must beheterogeneous and be implemented in different pro-gramming languages (such as C and Java ); andThe information exchange must be through electronicmeans and must be seamless

A quasi-VE is a partnership where one or many of theabove mentioned criteria holds true (but not all) Most

“global” enterprises today function in a manner that can

be referred to as ‘quasi-VEs’ The most difﬁcult criterion toadhere is the ability to exchange information seamlessly

“Seamless” (in this VE context) refers to the automatedtranslation of information from one data format to an-other without any manual intervention or re-keying of in-formation Many market-leading organizations exchangedesign and engineering information by e-mail and thenspend substantial time attempting to extract information

by manual means or by using a variety of software in a quence of tasks The major drawback of such approaches

se-is that they negate substantially the advantages these ganizations intended to accrue by adopting a VE-basedmodel of collaboration Valuable time and resources arelost by the inability to exchange information from onedata format to another This is perhaps one of the great-est challenges facing organizations interested in adopt-ing a VE model While the casual reader may concludethat organizations communicating through the Internet

or-in some collaborative manner are functionor-ing as a VE, it

is important to underscore the fact that this is a lar misconception Simply using electronic means (such

popu-as the Internet) to exchange information does not antee greater productivity Possessing the ability to usethe information (obtained from a partner) immediately in

guar-an accessible format to perform a speciﬁc target activity

is the key to realizing the beneﬁts associated with a based approach This ability contributes toward the part-ners being “agile” in a highly competitive environment,where customer needs keep constantly changing Whenthese needs change, a VE partner may decide to team with

VE-a different group of orgVE-anizVE-ations, who mVE-ay be using VE-a ferent set of computer tools and data formats After thevirtual teams and responsibilities have been identiﬁed, amajor issue that must be addressed is the nature of theinformation and data being exchanged If this issue is notadequately addressed, then the distributed partners willlose substantial time in accomplishing their collaborativeactivities quickly Consequently, the role of computer ar-chitectures that support the seamless exchange of infor-mation and the importance of data exchange standardsneeds to be better understood

dif-Quasi-VEs began appearing in the late 1980s andearly 1990s These include companies such as McDonnellDouglas Aerospace and Ford, among others Today, manycompanies (including Boeing) are beginning the migra-tion toward being a true virtual enterprise Most of themare facing challenges in exchanging information seam-lessly (which is a major technical problem) as well as indeveloping mutual trust with new partners (which is acultural problem) In the future, successful VEs would

be those organizations that emphasize the use of tured proven methods to aid in tasks such as virtual

Trang 16

struc-WL040C-192-Cecil WL040/Bidgolio-Vol I WL040-Sample.cls September 14, 2003 18:14 Char Count= 0

Table 1 Typical Characteristics of VE Partners in a Satellite Development Domain

Design Corporation Houston, TX Space system designers Design the propulsion systemASSEMTEC San Diego, CA Fabrication specialists Build various satellite subsystemsProcess Consultants, Inc Tampa, FL Manufacturing engineering Develop plans to manufacture

and assemble satellite and its partsSatellite Design, Inc Houston, TX Space system designers Design the cold gas, command,

and other modulesAgile Integrators Colorado Springs, CO Integration specialists Integration with launch bus and testingProject Management Dayton, OH Project management Manage entire product development,

team formation and are willing to adopt leading-edge

technologies that facilitate collaboration among

dis-tributed teams The trend to adopt common information

exchange standards such as XML (extensible markup

lan-guage) and STEP (standard for the exchange of product

design speciﬁcations) will continue Today, most

manu-facturing organizations have indicated their

disappoint-ment at emerging standards of data exchange While

there are many technical challenges in adopting data

ex-change standards, the major resistance to this adoption

comes from an unwillingness to change The emergence

of open architecture-oriented practices and the success

of organizations today that have embraced such

stan-dards will have a deﬁnite impact on future trends and

practices

Types of VEs

There are broadly two major categories of virtual

en-terprises, inter-nation and intra-nation VEs Inter-nation

VEs (or simply international VEs) are those whose

mem-bers extend beyond national boundaries For example,

consider the electronics-manufacturing domain Project

integrators and design partner organizations may be

lo-cated in California while process engineering team

mem-bers and resources are in Texas; in addition, the actual

assembly and manufacturing activities can occur in

vari-ous countries in Asia (such as Taiwan or Singapore) In an

intra-nation VE, a consortium’s partners are within a

spe-ciﬁc nation’s boundaries Figure 1 illustrates the concept

of an intra-national VE Another example of VE

partner-ships and skills is provided in Table 1 In both categories,

the Internet can serve as the communication backbone

linking the VE members There needs to be a

demarca-tion between companies who merely have subcontractors

in various parts of the world (who may manufacture or

assemble parts of a ﬁnal product) and companies who

use the Internet to exchange and share information that

directly inﬂuences the collaborative activities involved A

manufacturing giant based in California may claim to be

part of a global network and yet not function as a true

VE Globalization does not mean just using the Internet

or any other electronic means to exchange information

Numerous organizations claim being part of a global

net-work and function more as quasi-VEs and in some

situa-tions, mainly subcontract a portion of their activities

be-cause of lower manufacturing and other costs Supplier

chain management is one domain in which adoption ofInternet-based approaches has proven to be successful.Data exchange has been less of a problem in this domainand this has enabled the adoption of Internet-based prac-tices to support activities related to this domain

Importance of the Emerging VE Model and the Role of the Internet

At the onset of this new millennium, manufacturing nizations worldwide are collaborating and functioning as

orga-a virtuorga-al enterprise With revolutionorga-ary orga-advorga-ances in mation technology (IT) and electronic communicationsserving as catalysts, the Internet has emerged as a power-ful integration vehicle for the realization of the global mar-ketplace Private and government organizations have rec-ognized the potential of the Internet as a VE facilitator andhave begun to implement distributed collaboration ap-proaches using the Internet as a backbone In this context,there have been several research and industry initiativesthat have sought to focus on the development of innova-tive integration frameworks to support distributed collab-orative activities The term “distributed design, planning,and manufacturing” refers broadly to a subset of VE activ-ities, where physically distributed design, planning, andmanufacturing resources interact with each other acrossheterogeneous computer systems (or platforms) to ac-complish identiﬁed manufacturing tasks These resourcescan include personnel (such as design or manufacturingengineers), software tools (used to create design, man-ufacturing plans, etc), computer systems (on which thesoftware tools reside), and machines (including robots,assembly and metal cutting equipment, etc)

infor-The development of a product can occur in multiplephases referred to as the product development life cycle.Typically, a life cycle (LC) of a product includes conceptu-alization of a design idea, detailed design and engineeringanalysis, project planning, manufacturing and assemblyplanning, supply chain management, manufacturing (orfabrication), testing, service, delivery and, retirement orrecycling (Figure 2) In today’s global economy, the com-plex life-cycle activities involved in developing a productare being performed in a distributed manner (see Fig-ure 1) The project teams, software tools, analysis mod-els, and manufacturing resources involved in this cycleare also becoming increasingly distributed and are im-plemented on heterogeneous computing systems, which

Trang 17

570

Service Engineering Suppliers/

Subcontractors

Product/Process Development

further compounds the existing complex life-cycle

inte-gration problem (Cecil, 2001)

Potential Beneﬁts of Adopting the VE Model

The major beneﬁts of adopting a VE-oriented model

in-clude (a) the ability to respond rapidly to a newly

iden-tiﬁed customer need for a product and (b) cost savings

A VE framework will allow Company A to partner with

a design specialist (Company B) and a few other

organi-zations to quickly come up with a way to manufacture

a product X After a product demand diminishes or

dis-appears, these VE partners can disband When another

market need (such as a new product) is identiﬁed,

Com-pany A can form another VE partnership with yet another

diverse group of organizations and become involved in

de-veloping a new product Y By becoming temporary

part-ners for a product’s life cycle, Company A becomes more

agile in being able to serve customers in today’s

chang-ing global environment A virtual enterprise-oriented

ap-proach will provide organizations the capability to be

ag-ile in today’s customer-driven global market The term

“agility” is meant to indicate the ability to keep up with

customers’ changing needs For example, a cheetah is an

agile animal When its prey changes direction, the

chee-tah’s agile nature enables it to change direction and still

continue the chase in pursuit The Internet along with

other technologies is enabling numerous industrial

orga-nizations to become more agile When companies adopt

a VE-oriented approach coupled with the power and

scal-ability of the Internet, they will be able to form national

and international partnerships to produce low-cost,

high-quality products

CREATION OF VIRTUAL ENTERPRISES

Problems and Challenges in the

Creation of VEs

The challenges facing VE implementation can be grouped

under technical and cultural One of the major

techni-cal problems is achieving seamless exchange of

informa-tion as well as the integrainforma-tion of the myriad of

activi-ties involved in designing and building products With the

ever-increasing use of the Internet by industrial

organiza-tions today to exchange technical and business

informa-tion, more advanced and sophisticated IT-based

frame-works and approaches that will support accomplishment

of complex life-cycle activities seamlessly are under velopment To improve the productivity of VEs, a variety

de-of issues, especially those dealing with the role de-of the ternet as a VE facilitator, need to be addressed Internet-based computer frameworks and architectures must becarefully evaluated with respect to their ability to sup-port realization of VE goals and objectives Some of thecriteria can include the following: Using computer frame-work A, can VE partners quickly respond to customer de-mands? Does the proposed Internet approach facilitateseamless exchange of information including engineering,planning, and other life-cycle data? Can information andsoftware systems communicate across heterogeneous sys-tems (such as Windows, Macintosh, and UNIX environ-ments)? At any given time, are the VE partners aware ofeach other’s task accomplishments, their work-in-process(WIP), and the rate of progress toward achieving theiroverall target production?

In-The cultural problems relate to the ability to trust newpartners, adopt new ways of collaboration, and interacteffectively with team members who have less face-to-faceinteraction during collaboration

Technologies and Frameworks for the Realization of VEs

The Internet, by far, is the most versatile communicationvehicle that can be used to create and manage VEs It isbeing widely used by business enterprises globally to ex-change information in all phases of a product’s life cycle.The Internet can be viewed as “a network of networks”that is scalable and can connect remote corners of ourworld The two most widely used protocols of the Internetare the transfer control protocol (TCP)/Internet protocol(IP) (commonly referred to as TCP/IP) and the hypertexttransfer protocol (HTTP) (which is better known as theWorld Wide Web) The TCP/IP was developed nearly 30years ago and it’s the backbone for most of the computer-based communications today Communication via e-mail,controlled discussion groups on speciﬁc topics, and Inter-net video-based conversations has become commonplaceand is replacing the more traditional and expensive tele-phone discussions and satellite based video conferencing

as well Architectures such as CORBA (discussed later inthis chapter) have been developed on protocols such asTCP/IP Other developments such as the advent of “In-ternet2” (which will provide substantially more band-width and more effective transmission of video graphicsand virtual reality-based images) will continue to emergeand mature in response to industrial and educationalneeds

Another technology (which has been used widely by alarge number of industry giants including Wal-Mart andFord) is “Electronic Data Interchange” (EDI) Partners in-volved in an EDI transaction can exchange informationfrom one computer to another directly in a secure auto-mated manner (Cecil, 1996) With the help of translatorsand transmission standards (national and international),business and technical information including purchaseorders, invoices, quotes, and design documentation can

be exchanged Also, electronic funds can be exchangedfrom one computer to another

Trang 18

B

1

Buyers

VAN VAN

A

2

3 C

B

1

Buyers

Figure 3: Buying and selling using EDI.

Just as Internet services today are provided by pendent Internet service providers (ISPs), EDI services

inde-can be obtained through value-added networks (or VANs)

(Figure 3) Companies such as GE, IBM, and AT&T

pro-vide these EDI services, depending on individual and

group needs EDI frameworks allow for the

establish-ment of electronic bulletin boards where various needs

can be posted and subscribers can respond to the

busi-ness opportunities using EDI The U.S Defense Logistics

Agency (DLA) is managing the federal government’s

im-plementation of EDI In recent years, most of the VANs

have offered a hybrid EDI/Internet-based service where

the World Wide Web has been used to exchange

EDI-based documents between customers and businesses

Ad-ditional information on EDI can be obtained from a

num-ber of sources, including books and the Internet (ECRC,

2002)

Computer Architectures and Technologies

That Support the Realization of VEs

When teams and resources are distributed and linked via

the Internet, as mentioned earlier, the major problem is

the ability to communicate across heterogeneous

com-puter platforms The various software modules and

sys-tems used to accomplish engineering and business

func-tions can be implemented on various software paradigms

or frameworks One of the more important (yet basic)

con-cepts in the realm of software computing is the notion of

an “object.” An object-oriented software program (or an

“object”) can be viewed as a discrete software entity that

contains some “data” that can be manipulated using

cer-tain functions or operations In general, such an object has

several advantages over traditional software entities built

using non-object-oriented languages such as Fortran

Ob-ject orientation provides three distinct advantages

includ-ing ease of maintenance, ease of change, and less time to

create Objects can be reused, and in most cases, they

pro-vide a basis that can be extended Objects can be created

from a template The template used to create a group of

objects is termed a “class.” In a manufacturing or any

other enterprise, most objects model real-world entities

Software entities can send messages to objects with ciﬁc requests; the objects, in turn, send their responsesthrough messages

spe-In an spe-Internet-based VE, the various software entitiesthat are distributed can be viewed as engines and tur-bines working together to propel a given enterprise Us-ing distributed object-computing methods, communica-tion among the physically distributed software systems ispossible Distributed computing allows objects to be dis-tributed in an heterogeneous manner across the Internet

by extending object-oriented programming concepts sothat these distributed objects behave as a uniﬁed whole.These objects can reside in their own address space out-side of an application and be distributed on different com-puter platforms linked via the Internet; however, they willbehave as if they were local objects There are severalways to implement a distributed computing environment,which is a key requisite to realize a fully functional VE.Three of the most popular paradigms and approaches arediscussed in the following sections: the common object re-quest broker architecture (CORBA) from the Object Man-agement Group (OMG), the distributed component ob-ject model (DCOM) from Microsoft, and Jini technologyfrom Sun Microsystems Among these three, CORBA andDCOM are architectures and can be compared Jini is built

on top of the Java language and has become popular as itenables systems to function as a federation of services

The Common Object Request Broker Architecture (CORBA)

The Object Management Group is the world’s largest puter industry consortium whose mission is to deﬁne aset of interfaces for software to be interoperable OMG

com-is a nonprofit organization with around 750 members.The OMG provides a structure and a process throughwhich its members can specify technology and then pro-duce commercial software that meets those specifications.CORBA is an industry consensus standard that defines

a higher-level facility for distributed computing The tributed environment is speciﬁed using an object-orientedapproach, which masks the differences relating to objectlocation, type of operating system or computing platform,

Trang 19

dis-WL040C-192-Cecil WL040/Bidgolio-Vol I WL040-Sample.cls September 14, 2003 18:14 Char Count= 0

572

CORBA Services (Object Services)

CORBA facilities (Common Facilities)

Object Request Broker

Application Objects

Figure 4: The object management architecture.

and programming languages (used for implementing

objects) CORBA supports interoperability by using a

well-deﬁned set of interface speciﬁcations (Mowbray &

Zahavi, 1995)

CORBA is a speciﬁcation for an application level

com-munication infrastructure It can be viewed as a

peer-to-peer distributed computing facility capable of

support-ing seamless communication and information exchange,

which is the cornerstone of VE implementation In a

CORBA context, all applications are viewed as objects

These objects are capable of assuming dual roles as a

client and a server A client invokes a service to which

another object (referred to as the server) responds In

gen-eral, CORBA allows more ﬂexibility from a computer

ar-chitecture point of view when compared to pure client–

server frameworks allowed by remote procedure calls

(RPCs)

While CORBA connects objects, the realization of a VE

requires any supporting computer architecture to support

enterprise integration The object management

architec-ture (OMA) seeks to address this VE requirement

(Fig-ure 4) and is based on CORBA The major components

of OMA include the object request broker,

CORBAfacili-ties, CORBAservices, and application objects The object

request broker (ORB) can be viewed as a communication

infrastructure capable of relaying object requests and

re-sponses transparently across distributed computing

envi-ronments such as the Internet The CORBAservices (also

referred to as object services) provide lower level

function-ality such as life-cycle services (such as object creation and

notiﬁcation) and include providing access to online

trans-action processing (OLTP, the widely used application for

business accounting) and a “yellow pages” type of trader

service (where objects can offer their services at speciﬁed

costs)

The CORBA facilities (referred to as common facilities)

provide services for applications and have two major

seg-ments, horizontal and vertical The horizontal facilities

can be widely used in a variety of industrial domains and

markets and include user interface, task management,

in-formation management, and systems management The

vertical facilities deal with standardizing management

of information pertaining to speciﬁc industrial domains

such as healthcare, manufacturing, and ﬁnancial An ample of the horizontal facility is the compound docu-ment management facility, which allows applications astandard way to access the various parts of a document.Using this, a vendor can build tools for manipulation of

ex-a pex-art of the document (for instex-ance, ex-a three-dimensionex-alimage), which can then be marketed without having todevelop the functionality from scratch

The OMA and CORBA both facilitate the creation andmanagement of Internet-based software tools used in atypical VE-oriented product development cycle They al-low engineers and managers in design, manufacturing,and other areas to take advantage of a wide variety ofsoftware tools implemented in various programming lan-guages and on diverse operating systems (from a Macin-tosh to a UNIX system) IT specialists and software pro-grammers especially will beneﬁt from using OMA/CORBAbecause they provide a sophisticated mix of easy com-ponent accessibility and transparent distribution; further,code can be reused in new applications or can be modiﬁedincrementally to suit the increasing scope of applications

in a VE Tools and modules can be developed in a variety

of languages For example, a Java-based program can beused to perform a manufacturing optimization task (inyour VE) while a C++ module can be built to create a userinterface that is accessible to everyone in a VE

The key to object interoperability in CORBA depends

on the deﬁnition of contracts termed “interfaces.” Eachobject has an interface (which is public) and an implemen-tation (which is private) The services of each object areexpressed as a type of contract in this interface, which pro-vides two very important functions: (a) it informs otherclient objects in the VE of the services it provides as well

as specifies how to be “called” (or invoked); and (b) it lows the IT infrastructure to understand the specific man-ner in which it will send and receive messages The latterallows for data translation between client and server ob-jects Each object in the VE also requires a unique identi-fier or handle, which can be used to direct messages Theinterfaces of all the objects can be expressed in a neutrallanguage called the interface definition language (IDL) InCORBA, the IDL definition for all objects is stored in aninterface repository This repository is useful to achieveobject interoperability, which is a key issue in the cre-ation of systems supporting the functions in an Internet-based VE

al-In CORBA, IDL is the means by which a particular ject implementation tells its potential clients what opera-tions are available and how they should be invoked.When

ob-an application object is created in a speciﬁc lob-anguage such

as C or C++ (or any other language that can be mapped

to IDL), each object’s interface can be deﬁned in IDL.There are two ways for distributed objects to be linked

in CORBA: using static IDL interfaces or using the namic invocation interface (DII) When using static in-terfaces (during implementation), the IDL speciﬁcationsare compiled for each interface into header, skeletal, andstub programs for linking the Internet-based distributedapplications When a client object (such as a task man-ager located in VE site 1) invokes an operation on an ob-

dy-ject reference, it links using stubs (which are generated

automatically from an IDL compiler for the language and

Trang 20

Internet

ORB mechanism stub

client

server

Skeleton Object

Adapter

Figure 5: Linking in a VE using stubs and skeletons.

ORB environment the client is implemented in) These

stubs convey this information to the target object via the

ORB (Figure 5) From the client’s perspective, the stub

acts like a local function call After the stub interfaces

with the ORB, the ORB encodes and decodes the

opera-tion’s parameters into formats suitable for transmission

(this is termed “marshaling”) On the server object side

(the server may be a manufacturing analysis object or any

other application object located in another VE site 2), the

information is “unmarshaled” and a skeletal program

in-terfaces with the ORB through an object adapter This

skeleton can map the request back to the implementation

language of the server object (the skeletal program is

ob-tained using an IDL compiler the server is running on)

When the ORB receives a request, the skeleton in essence

performs a call-back to the server object After the server

completes processing the request, the results are returned

to the client via the skeleton/stub route (along with any

re-ported errors or problems encountered)

The static interface approach does not support thedynamic use of newly created objects once they are in-

troduced into the VE’s network The dynamic

invoca-tion interface provides this funcinvoca-tion and enables a client

to discover new objects and their interfaces, retrieve

their interface deﬁnitions, construct and send requests,

and receive the associated response from objects on the

Internet

In CORBA, the encapsulation properties of the ous software objects enable location transparency in the

vari-VE In an encapsulated component, there are two parts:

the public interface (presented to the outside world) and

the private implementation (which is appropriately

hid-den from view) When a client sends a message, the

in-vocation is sent to the ORB (and not the target server

object); the ORB routes the message to the destination

or server object Consequently, the location of the object

within the virtual enterprise does not matter How the

re-sults were obtained is of concern only to the server

com-ponent (or object) and the client need not know how the

server processed its request The interfaces can be viewed

as contracts by an object By using IDL, which allows the

interfaces to be speciﬁed in a neutral language, it is

pos-sible to separate interfaces from the implementation The

mapping from IDL to languages such as C, C++, Java, and

Smalltalk (among others) can be achieved by enabling

var-ious resources in a VE to be implemented in a variety of

programming languages running on heterogeneous

plat-forms ranging from UNIX to Windows By using IDL to

specify interfaces, different VE partners and team

mem-bers can independently implement different parts of a

dis-tributed system, which can be used to accomplish target

Figure 6: Clients and server objects communicating via the

Internet using IIOP

life-cycle activities (ranging from design through testing

of a product) When application objects in a VE are movedfrom one site to another in a VE, the ORBs can use the ob-ject references to locate them In general, the client doesnot know whether a target server object is local or dis-tributed (and linked using the Internet)

In any typical VE-oriented implementation, each Website of a company partner will have an ORB ORBs(whether implemented in the same language or in differ-ent languages at each site) can communicate to each otherusing the Internet Inter-ORB protocol (IIOP) (Figure 6).The IIOP is the general inter-ORB protocol (GIOP) overthe TCP/IP, which is mandatory for CORBA 2.0 compli-ance The IIOP is based on the TCP/IP, which is the mostpopular transport mechanism available today and is theprotocol of the Internet Interoperable object referencesenable invocations to pass from one (language) ORB toanother

Security is an important issue in any distributed proach; use of various computers introduces issues of con-sistency and trust between them In a distributed system,information is in transit and is more vulnerable to out-side “attacks.” CORBAsecurity is part of the CORBAser-vices and is ﬂexible and can be modiﬁed to suit differ-ent security needs (Seigel, 2000) The building blocks

ap-of CORBAsecurity include identiﬁcation and tion of principals, authorization and access control, del-egation, non-repudiation, cryptography and maintainingavailability, and performing security auditing to maintainuser accountability

authentica-Personal computers (PCs) constitute the major ment of desktop computer systems used in industry today.Using OMA/CORBA, PCs can become powerful partici-pants in the design and functioning of VEs linked usingthe Internet Each ORB can be from a different vendor andimplemented in diverse programming languages ORBproducts can differ greatly yet conform to the OMG speci-ﬁcations and guarantee interoperability over the Internet.Some of the ORB products include ObjectBroker (fromthe formerly known Digital Equipment Corporation), theSOM product set (from IBM), the DAIS product set(from ICL), Orbix (from Iona Technologies), DistributedSmalltalk and ORB Plus (from Hewlett–Packard), and theNEO product family (from SunSoft, Inc.) Additional in-formation on CORBA and its speciﬁcations can be ob-tained from the OMG Web site (OMG, 2002)

Trang 21

seg-WL040C-192-Cecil WL040/Bidgolio-Vol I WL040-Sample.cls September 14, 2003 18:14 Char Count= 0

574

The Distributed Component Object Model (DCOM)

The Microsoft distributed component object model

(DCOM) also referred to as “COM on the wire” uses

a protocol called object remote procedure call (ORPC)

(Microsoft, 2002) DCOM is used substantially on the

Win-dows platform Microsoft provides common object model

(COM) implementations for Windows and Solaris

plat-forms while other companies provide implementations

for UNIX, Linux, and other mainframe platforms The

Ac-tive Group, which is a consortium of vendors interested

in the evolution of COM and DCOM, manages both these

speciﬁcations COM required that the provider and user

of an interface both reside on the same computer For

instance, Microsoft Visual Basic could activate and use

Microsoft Excel on the same computer but was not

capa-ble of controlling Excel on another computer located on

the same local area network or on the Internet DCOM

ex-tends the original COM to support communication among

distributed objects on different computers in the

Inter-net (or local area or wide area Inter-networks) An interface

client can make a request for another interface, which

can be provided by an instance of another object, which is

on another computer on the Internet COM’s distribution

mechanism can connect the client to the server so that the

method called from the client is received by the server (or

provider) on another computer where it is executed and

the return values are returned to the client (or consumer)

The distribution is transparent to both clients and servers

One of the major mechanisms in COM is the

activa-tion mechanism, which establishes connecactiva-tions to

com-ponents and creates new instances of comcom-ponents In

COM, object classes possess globally unique identiﬁers

(GUIDs) Class IDs are GUIDs used to refer to speciﬁc

classes of objects In DCOM, the object creations in the

COM libraries are enhanced to allow creation on other

computers linked via the Internet To create a remote or

distributed object on the Internet, the COM libraries must

know the network name of that server (and the class

iden-tiﬁer CLSID) Based on this information, a service control

manager on the client computer links to the service

con-trol manager on the server computer and then requests

creation of a new object DCOM provides several ways to

allow clients to indicate the remote server names when a

new object is created

DCOM uses the object remote procedure call, which is

a layer on top of the distributed computing environment’s

(DCE) remote procedure call and interacts with COM’s

run-time services A DCOM server object is a piece of

soft-ware code capable of offering objects of a speciﬁc category

at a certain time (called run-time) Each server supports

multiple interfaces, which can each represent different

be-haviors A client in DCOM ﬁrst acquires a “pointer” to one

of the server’s (or provider’s) interfaces and then makes a

call to one of its exposed or public methods Using the

in-terface pointer obtained, the client objects can invoke the

public methods of the target server even though it is

lo-cated on another computer available on the Internet The

servers in DCOM can be written in various programming

languages such as Visual Basic, Java, and C++

COM uses the remote procedure call infrastructure to

accomplish marshaling and unmarshaling For this, the

exact method signature including the data types, sizes of

any arrays in the parameter list, and types of structuremembers must be known This information is provided

in an interface deﬁnition language, which is built on top

of the industry standard IDL (described earlier in theCORBA section) The IDL files are compiled using specialcompilers (such as the Microsoft IDL compiler MIDL),which generate C language source files, which contain thecode for marshaling and unmarshaling for the interfacesdescribed in the IDL file The client code is termedthe “proxy” and the server object is called the “stub.”When the proxy/stub for a specific interface is needed inCOM, the interface ID (IID) is identified from the systemregistry Whereas CORBA supports multiple inheritances

at the IDL level, DCOM does not ObjectBroker (which isDigital Equipment Corporation’s implementation of theCORBA speciﬁcation) can work with Microsoft’s objectlinking and embedding (OLE) functions for data objectsstored on non-Microsoft platforms The interfaces toMicrosoft’s OLE and dynamic data exchange (DDE) areavailable in ObjectBroker ObjectBroker’s OLE networkportal can intercept OLE calls on the PC and map them toObjectBroker messages, which can then be routed to theappropriate server In addition, ObjectBroker also allowsinterfacing with Microsoft’s Visual Basic, which enablesgraphical applications to be developed quickly and ex-tends the desktop’s capabilities to access information re-siding on computers linked via any network including theInternet

Java and Jini Technology

Java remote method invocation (RMI) relies on a col called the Java remote method protocol Java relies

proto-on object serializatiproto-on, which allows objects to be mitted as a stream (Raj, 2002) The major drawback isthat both the server and client objects must be written inJava However, Java RMI can be implemented on a va-riety of heterogeneous operating systems (from UNIX toWindows) with one restriction: there should be a Java Vir-tual Machine implementation for that platform Internet-based VEs that have no legacy software and are not con-cerned about the introduction of heterogeneous imple-mentations (in C++ and other languages) can implementJava-based systems Java has many advantages includingbeing object-oriented and offering ease of programming,modularity, and elegance Jini (from Sun Microsystems)seeks to extend the beneﬁts of object-oriented program-ming to the Internet (or any network) Jini is built on top

trans-of Java, object serialization, and RMI and enables puters to communicate to each other through object in-terfaces Jini provides for a more network-centric envi-ronment where computers not possessing a disk drive be-come more commonplace and interact over dynamicallychanging networks (such as the Internet) Jini technologyprovides ways to add, remove, or locate computers as well

com-as services It can help VE partners build and use a

dis-tributed system as a federation of services to accomplish

their target activities The set of all available services able on the Internet to the VE will compose this federationwith no speciﬁc service in charge Jini’s infrastructure pro-vides a way for clients and services to locate each otherusing a lookup service (which is a directory of currentlyavailable services)

Trang 22

avail-WL040C-192-Cecil WL040/Bidgolio-Vol I WL040-Sample.cls September 14, 2003 18:14 Char Count= 0

Agents and Mobile Agents

Agents have also become increasingly popular and are

well suited to support the functioning of Internet-based

VEs Agents can be viewed as software entities, which

are semi-autonomous, proactive, and adaptive and which

have a long life They can collaborate with each other to

work toward common or independent goals (Deshmukh,

Krothapalli, Middlekoop, & Smith, 1999; Krothapalli &

Deshmukh, 1999; Lange, 1999) Using an Internet-based

framework, agents that can help accomplish a number of

tasks in a VE can be designed and created These tasks can

be system oriented (for example, monitoring and

notiﬁca-tion of task complenotiﬁca-tions) or product/process oriented (for

example, generation of a plan to assemble three parts)

Agents called “mobile agents” hold enormous potential in

revolutionizing the way in which VEs function A mobile

agent is capable of replication and autonomous movement

from one VE site to another and of performing tasks based

on information collected from various sites in a VE These

are in contrast to stationary agents or objects, which

exe-cute only on the system they reside on and when they need

to interact with objects on another system, use methods

such as remote procedure calling (RPC) Mobile agents

are not bound to their host site or system but can travel

(or “roam”) among the various computing hosts Such an

agent can transport its state and code along with it to

another location, where it can resume execution Several

university and industry projects have highlighted the

ben-eﬁt of using mobile agents in distributed environments

(Cecil, 2002a; Lange, 1999) The potential advantages of

a mobile agent approach include overcoming network

la-tency (they can be dispatched from a remote control

medi-ator to act locally and eliminate latency necessary for

real-time response and control), encapsulating protocols, and

offering better performance, increased ﬂexibility, and

sup-port dynamic response to changing scenarios The tactical

advantage of performance is gained by sending a

compo-nent across the network to a VE site, where the work gets

completed The computers in the VE need to be connected

only long enough to send the mobile components and later

to receive it back This same concept comes in useful

espe-cially for monitoring remote activities in manufacturing

and other domains (Cecil, 2002b)

Industry and University Initiatives Related

to Internet-Based VEs

Many industrial organizations with remote locations and

distributed resources and partners have adopted a VE

mode of functioning using the Internet as the

commu-nication backbone The VE model has been adopted in a

wide variety of industries including aerospace

engineer-ing, airline and travel industry, shipbuildengineer-ing, computer

manufacturing, healthcare, IT systems consulting,

bank-ing, electronic commerce, and telecommunications In

general, organizations involved in the service and

consult-ing sectors (who do not produce a physical product for

customers but rather provide services) can use

Internet-based frameworks to exchange information seamlessly in

a distributed collaborative manner Organizations such

as Ford, Boeing, the Sabre Group, Lufthansa,

Schlum-berger, Pratt and Whitney, Cisco Systems, Raytheon,

Harvard University, and NASA Goddard Space Flight ter have adopted CORBA-based architectures to be moreagile and customer responsive (CORBA, 2002)

Cen-Boeing, the world’s largest producer of commercial liners, uses a CORBA-based framework to integrate itsdesign, manufacturing, and resource management activi-ties The manufacturing and assembly activities at Boe-ing involve as many as 3,000,000 individual parts foreach aircraft produced Information integration, inven-tory management, collaborative sharing of design, man-ufacturing, and work-in-process data are extremely com-plex and require a robust distributed IT infrastructure.Boeing’s Internet-based VE computer architecture (which

air-is based on the OMA/CORBA model) can support morethan 45,000 users and 9,000 concurrent users in variousregions across the USA (CORBA, 2001)

The NIIIP project is a national initiative that focuses

on developing, demonstrating, and transferring (to terested organizations) the technology to enable indus-trial virtual enterprises The National Industrial Infor-mation Infrastructure Protocols (NIIIP) consortium is agroup of industry, university, and government organiza-tions deﬁning (and have deﬁned) the NIIIP protocols aswell as demonstrating their use Some of the consor-tium members include CAD Framework Initiative, DigitalEquipment Corporation, IBM, General Dynamics Elec-tric Boat, Lockheed Martin Aeronautical Systems Com-pany, National Institute of Standards and Technology,STEP Tools, Rensselaer Polytechnic Institute, Texas In-struments, and the University of Florida Additional in-formation about the NIIIP reference architecture can beobtained from their Web site (NIIIP, 2002) NIIIP aims

in-to establish standards-based software framework proin-to-cols as well as develop software and toolkits (as part of itseffort to provide a technical foundation) for implement-ing virtual enterprises There are four building blocks ofthe NIIIP reference architecture, and they include com-munication (using the Internet), use of object technology,knowledge and task management, and common informa-tion model speciﬁcation and exchange

proto-Several U.S federal programs have initiated projects todesign collaborative virtual environments Among theseare the Distributed Knowledge Environment of the De-partment of Defense; Intelligent Collaboration and Vi-sualization initiative of the Defense Advanced ResearchProjects Agency (DARPA); System Integration for Manu-facturing Applications; National Advanced Manufactur-ing test bed of NIST; Rapid Design Exploration andOptimization (RaDEO), and Agile Infrastructure for Man-ufacturing Systems (AIMS) A distributed Internet-basedprocess planning system called CHOLA has been devel-oped at New Mexico State University (Cecil, 2001; Cecil,2002a) CAD ﬁles and process planning modules are dis-tributed among heterogeneous environments Dynamicinformation (such as design data, equipment capabilityand availability, and tool availability) from remote loca-tions is used by the VE (which includes ITESM in Monter-rey, Mexico, and Penn State University in State College,PA) to generate a process plan for various part designs(Cecil, 2002a) Each site has an object request broker,which acts as a communication infrastructure linking thedistributed sites via the Internet This combined research

Trang 23

576

and curriculum development initiative was funded by the

National Science Foundation as part of the emphasis to

introduce emerging engineering concepts and practices

to engineering students

Activities and Phases in the Creation

of Internet-Based VEs

There are no structured methods or steps to support the

creation of VEs today However, by addressing the

ma-jor issues involved to establish, sustain, and function as

a VE, industrial organizations can develop their own

ap-proaches to successfully adopt VE practices The major

activities in any VE development include the following:

Development of an understanding of a given VE’s

prod-uct(s) and customers;

Identiﬁcation of the potential VE partners and formation

of the product development teams;

Development of an information-based enterprise model

of the VE’s collaborative activities and tasks;

Design and implementation of an Internet-based

dis-tributed software system that will link all VE team

members (and possibly customers) and be used to

ac-complish the various VE activities;

Initiation of a pilot initiative in which partners function

as a VE using the implemented Internet-based system;

and

Based on performance in pilot initiative, identiﬁcation

and adoption of necessary changes to the overall

ap-proach and Internet-based architecture and software

system

Among these activities, the most important activities

are the creation of an information-oriented enterprise

model and the work associated with the design and

im-plementation of an Internet-based system to support the

functioning of a VE (Cecil, 2002c) The use of

information-oriented enterprise models are becoming increasingly

popular and have their roots in the integrated

deﬁni-tion (IDEF) work that originated in the ICAM Program

at Wright Patterson Air Force Base decades ago Apart

from older methodologies such as the IDEF-0 and IDEF-3

(Cecil, 2002b; Mayer, 1992), more recent development

include the enterprise modeling language (EML,

pro-posed by Virtual Enterprise Technologies) (Xavier & Cecil,

2001) EML was designed with the primary goal of

en-abling companies interested in creating and

participat-ing in VEs to conceptualize and model the way in which

partner organizations would interact with each other in

accomplishing various enterprise activities Further, it

provides a structured way to propose and reﬁne how to

accomplish detailed activities encompassing all or some

activities in a product development cycle (ranging from

development of a product design collaboratively to the

generation of project and manufacturing plans to

sup-plier chain management) EML enables VE team

mem-bers to describe what activities will be performed and how

to accomplish them using available resources and

con-straints Using decompositions, detailed models at

vari-ous levels of abstraction can be built In EML, the top-level

Influencing Criteria (IC)

Focus Unit

Task effector

Decision Objects (DO) Associated Performing Agents (APA)

Manufacture the Satellite Design

Team : VE Team A Physical Resource: Shop Floor # 22 Software Object: Controller

Constraints: Manufacturing Capabilities,

Manufacturing Schedule

Information Inputs: Process Plans, BOM

Information Objects: NI, UI, FI Physical Objects: nanosatellite

&

Figure 7: Functional unit representations using EML.

life-cycle activities are captured as functional units Eachfunctional unit can correspond to an identiﬁed life-cyclefunction such as “create product design X” or “manufac-ture design X using VE partners.” An information-rich de-scription (or model) can be developed using four classes

of attributes including inﬂuencing criteria, task effectors,decision objects, and associated performing agents Fig-ure 7 illustrates the main information attributes captured

in EML, which include focus units and the four attributeclasses EML provides a structured basis to model VE ac-tivities, capture their interrelationships, and specify theiraccomplishment using temporal precedence criteria.Inﬂuencing criteria help VE team members identifymajor constraints as well as enable identiﬁcation of in-formation needed by VE teams or a VE subcontractor toaccomplish a certain task For example, in Figure 7, theinformation inputs needed to accomplish the target ac-tivity “manufacture the satellite design” include processplans and Bill of Materials (BOM) Associated perform-ing agents help model who or which mechanisms will ac-tually help accomplish target activities or subactivities

&

Decision d i

Manufacture Satellite Design

Manufacture Propulsion Module (PM)

Manufacture Cold Gas Module (CGM)

Manufacture Command Module (CM)

Assemble (PM, CGM)

Assemble assembly 1 with CM of Satellite

Trang 24

Decision objects represent outputs after tasks are

com-pleted Task effectors enable capturing the temporal

logic underlying the task accomplishments (for instance,

should Tasks A and B be accomplished concurrently or

sequentially?) Figure 8 shows the decomposition of the

functional unit described in Figure 7 (the boxes

cor-responding to inﬂuencing criteria and associated

per-forming agents have been omitted for easier reading in

Figure 8) In general, enterprise model building of future

activities will enable the IT team composed of VE team

members to design and build effective Internet-based

systems (based on CORBA or any other architecture),

which will be the backbone of the various distributed VE

activities

CONCLUSION

The Internet is a powerful vehicle for VEs to be created

and deployed This chapter discussed the major

Internet-based approaches, tools, and technologies available

to-day to establish virtual enterprises Other supporting

tech-nologies can also be used on the Internet to promote

bet-ter communication among distributed team members in a

VE An example of such a supporting technology is virtual

reality, which can play a major role in the functioning of

VEs (Banerjee, Banerjee, Ye, & Dech, 1999; Brown, 1999;

Cecil 2002b, 2002c; Goldin, Venneri, & Noor, 1999)

Dis-tributed team members can communicate effectively

us-ing this powerful technology from various locations With

the development of Internet2, the use of virtual reality

for VE task accomplishments is expected to become more

widespread A key aspect of collaboration, which must be

embraced in any VE, relates to the notion of “concurrent

engineering” (Mayer, Su, & Cecil, 1997) In such an

con-current approach, the distributed cross-functional teams

must consider both product and process design issues

si-multaneously to ensure reduced costs, shorter

develop-ment lead time, and higher product quality

Internet2 is under development by a partnership volving U.S universities, industry, and government; it is

in-a more in-advin-anced network thin-at will link universities,

gov-ernment, and research laboratories for the purposes of

collaboration, distance learning, research, health services,

and other applications that require high bandwidth

be-tween the distributed sites Internet2 is not intended to

replace the Internet; rather, it will complement the

ca-pabilities of the Internet by providing additional

capa-bilities such as the development and deployment of

ad-vanced network applications and technologies, including

substantial increase in the bandwidth Logistical

network-ing is a new approach for synthesiznetwork-ing networknetwork-ing and

storage to create a communicative infrastructure for

net-work multimedia and distributed applications In May

2002, researchers from Logistical Computing and

Inter-networking (LoCI) Laboratory at the University of

Ten-nessee, where research in logistical networking research

is being pursued, demonstrated “Video IBPster,” an

ap-plication that can deliver video at high performance The

technology used is simpler and less expensive to deploy

than current approaches to streaming video Additional

information is available at the Internet2 Web site

(Inter-net2, 2002)

The adoption of the Internet as the vehicle of munication by industries worldwide will continue togrow As both technology and practices mature, the VEmodel is expected to become more widespread Therewill be more emphasis on the structured design of VEapproaches (using information modeling methods), de-velopment of effective virtual team formation/interactionmethods (Hackman, 1990), and seamless exchange of in-formation Organizations that adopt a quasi-VE approachwill be forced (by increasing competition) to focus onunproductive practices relating to data incompatibilityand information exchange Smaller organizations will beable to form partnerships with larger enterprises and havemore access to more market opportunities worldwide.The Internet has provided a more open approach andincreased business opportunities for industrial organiza-tions worldwide It has transformed the essential man-ner in which new products are made, created a morecustomer-oriented environment and radically changedthe manner in which people communicate with eachother The Internet is the cornerstone of the informationtechnology revolution It has created new opportunitiesand provided groundbreaking avenues to better health,education, and literacy; it has become the de facto com-munication vehicle of choice by millions worldwide andhas allowed us to be closer to each other than ever before.When a group of engineers and programmers began thecreation of the Internet several decades ago, they had lit-tle idea of its far-reaching impact Today, thanks to theirvision and dedication, people all over the world feel closer

com-to each other, even if oceans separate us The clich´e is true:

We are but a cyber-click away from each other Our worldwill never be the same again

GLOSSARY

Computer architecture The functional appearance of acomputer to its immediate users or, as in this chapter,the relationship of the various software elements andthe manner in which they interact with each other andthe processor to accomplish speciﬁc tasks The CORBAand DCOM models discussed in this chapter are exam-ples of two different architectures

Distributed computing A functional task or activitycompleted in a collaborative manner by humans and/orsoftware systems that are not co-located but residing

on several geographically distributed computers linkedvia the Internet or any other network

Flexible manufacturing The ability to manufacture awide variety of parts using reconﬁgurable computer-controlled manufacturing equipment

Mobile agents Software entities that migrate from onecomputer to another on a network such as the Internet

Process planning A task or function that identiﬁes theprocess steps needed to manufacture a given design

CROSS REFERENCES

See Client/Server Computing; E-systems for the Support of

Manufacturing Operations; Intelligent Agents; Virtual ality on the Internet: Collaborative Virtual Reality; Virtual Teams.

Trang 25

Re-WL040C-192-Cecil WL040/Bidgolio-Vol I WL040-Sample.cls September 14, 2003 18:14 Char Count= 0

578

REFERENCES

Banerjee, A., Banerjee, P., Ye, N., & Dech, F (1999)

Assembly planning effectiveness using virtual reality.

Presence—Teleoperators & Virtual Environments, 8(2),

204–217

Brown, A S (1999) Virtual people take on the job of

test-ing complex designs Mechanical Engineertest-ing, 121(7),

44–49

Cecil, J A (1996) Electronic commerce and the

elec-tronic commerce resource centers Marine Safety

Coun-cil Proceedings—The Coast Guard Journal of Safety at

Sea, 53(4), 10–13.

Cecil, J A (2001) Distributed heterogeneous architecture

for manufacturing applications (DHARMA) (Project

Re-port) Las Cruces, NM: Virtual Enterprise Engineering

Laboratory (VEEL), New Mexico State University

Cecil, J A (2002a) A distributed framework for satellite

product development (Project Report) Las Cruces, NM:

Virtual Enterprise Engineering Laboratory (VEEL),

New Mexico State University

Cecil, J A (2002b) A functional model of ﬁxture design

to aid in the design and development of automated

ﬁx-ture design systems Journal of Manufacturing Systems,

21(1), 58–72.

Cecil, J A (2002c) DHARMAM: An university–industry

partnership to enhance manufacturing engineering

education Paper presented at Advanced

Manufactur-ing Institute’s International Conference on University &

Manufacturing Industry Collaboration, August 12–13,

2002, Kansas City, Missouri

Cecil, J A., Kanchanapiboon, A., Kanda, P., & Muthaiyan,

A (2002, July) A virtual prototyping test bed for

elec-tronics assembly In Proceedings of the 2002

Interna-tional Electronics Manufacturing Technology (IEMT)

Symposium (pp 130–136) Piscataway, NJ: IEEE

Ser-vice Center

CORBA (2001) Manufacturing: Boeing Commercial

Airplanes Group Retrieved May 24, 2002, from

http://www.corba.org/industries/mfg/boeing.html

CORBA (2002) CORBA success stories Retrieved May 24,

2002, from http://www.corba.org/success.htm

Deshmukh, A., Krothapalli, A., Middlekoop, T., & Smith,

C A (1999, January) Emergent aerospace designs

using negotiating autonomous agents: Laboratory for

Fundamental and Applied Research in Multi-agent tems report Amherst, MA: University of Massachusetts.

Sys-EDI (2002) Retrieved April 10, 2003, from http://www.edi-information.com/

Goldin, D., Venneri, S., & Noor, A (1999) Ready for the

future? Mechanical Engineering, 121(11), 61–70 Hackman, R (Ed.) (1990) Groups that work and “those

who don’t,” creating conditions for effective teamwork.

San Francisco: Jossey–Bass

Internet2 (2002) Retrieved May 24, 2002, from http://

www.internet2.eduKrothapalli, A., & Deshmukh, A (1999) Design of negotia-tion protocols for multi-agent manufacturing systems

International Journal of Production Research, 37(7),

1601–1624

Lange, D., & Mitsuru, O (1999) Seven good reasons for

mobile agents Communication of the ACM, 42, 88–89 Mayer, R J (Ed.) (1992) Information integration for

concurrent engineering (IICE)—IDEF3 process tion capture method report (AL-TR-1992-0057) College

descrip-Station, TX: Knowledge Based Systems

Mayer, R J., Su, C J., & Cecil, J A (1997) Enabling current engineering in the product analysis environ-

con-ment Institute of Industrial Engineers (IIE)

Transac-tions, 29, 791–797.

Microsoft (2002) Distributed component object model

(DCOM)—Downloads, speciﬁcations, samples, papers and resources Retrieved May 24, 2002, from http://

www.microsoft.com/com/tech/dcom.asp

Mowbray, T., & Zahavi, R (1995) The essential CORBA,

systems integration using distributed objects New York:

Wiley

NIIIP (2002) Welecome to: National Industrial

Informa-tion Infrastructure Protocols and related projects

Re-trieved May 24, 2002, from http://www.niiip.orgObject Management Group (OMG) (2002) RetrievedMay 24, 2002, from http://www.omg.org

Raj, G S (2002) A component engineering cornucopia.

Retrieved May 24, 2002, from http://gsraj.tripod.com/

Seigel, J (2000) The CORBA 3 fundamentals and

program-ming object management group New York: Wiley.

Xavier, B., & Cecil, J (2001, August) Design of an

en-terprise modeling language (VETI EML Report) Las

Cruces, NM: Virtual Enterprise Technologies, Inc.(VETI)

Trang 26

Virtual˙Private WL040/Bidgolio-Vol I WL040-Sample.cls August 14, 2003 17:53 Char Count= 0

Virtual Private Networks:

Internet Protocol (IP) Based

Virtual Private Networks:

Internet Protocol (IP) Based

David E McDysan, WorldCom

Introduction to IP-Based Virtual Private

Applications of IP Virtual Private Networks 579Drivers for IP-Based Virtual Private Networks 579Introduction to Virtual Private Networks

A Taxonomy of IP-Based Virtual Private

Customer-Edge-Based Virtual Private Networks 584

CE Virtual Private Networks Over Virtual

VIRTUAL PRIVATE NETWORKS

Applications of IP Virtual Private Networks

The public Internet plays an important role in many

enter-prises (McDysan, 2000) Users can exchange information

with individuals anywhere in the world via e-mail, Web

sites, transaction systems, ﬁle sharing, and ﬁle transfer

Furthermore, the Internet is a rapidly growing means of

conducting business for commercial enterprises It also

provides a means for companies to advertise their goods

and services The Internet can help reduce administrative

costs by placing the data entry, veriﬁcation, and

think-time aspects of order entry and service parameter

selec-tion in the hands of the end user This replaces the older,

less-efﬁcient paradigm of people in enterprises interacting

over the postal system and/or the telephone and

facsim-ile to place orders, update records, and complete business

transactions The Web provides the automated means for

the end user to peruse the choices at his or her own speed,

requiring the expenditure of energy and time of only one

person Furthermore, careful design of the Web site by

experts allows many more people access to the best set of

information In the classic telephone or facsimile method,

the level of expertise depended on the particular agent the

caller reached

The tremendous volume of such information on publicWeb sites continues to grow and increase in quality, based

upon real-world experience and user feedback When the

Web site contains enterprise-speciﬁc information that, for

one reason or another, is sensitive, we call the application

an intranet One level of security is that of user

identi-ﬁcations (IDs) and passwords This is the same level of

security used on many public domain Web sites The next

level of security is that of encryption and ﬁrewalls,

top-ics covered in the next section A more challenging

activ-ity is the use, by multiple enterprises, of the Internet in a

virtual private fashion in an application called an extranet.The premier example to date is probably that of the Au-tomotive Network eXchange, which connects major auto-motive manufacturers and their suppliers, as described atthe end of this article

In addition to control over who may communicatewith whom, as described above, virtual private networks(VPNs) have a number of additional important require-ments Of course, providing verifiable authentication thatspecific sites and users are part of a specific intranet orextranet VPN is an important requirement Also, keepingthe administrative cost of VPNs under control requiresautomation of membership discovery in conjunction withthis authentication Furthermore, customer networks willmake use of private IP addresses or nonunique IP address(e.g., unregistered addresses) This implies that there is

no guarantee that the IP addresses used in the customerVPN are globally unique

Drivers for IP-Based Virtual Private Networks

Progress marches ever onward, and the world of working is no different (McDysan, 2000) Similarly to theway enterprises constructed private data networks overthe telecommunications infrastructure developed for tele-phony, the industry is developing a new wave of technolo-gies, overlaying the basic suite of Internet protocols, toconstruct VPNs When the public network infrastructure

net-of a VPN matches that net-of the enterprise equipment, thensigniﬁcant savings can occur This is a recurring theme inthe history of communication networks, with the Internetsimply the latest frontier

Successful enterprises are cost conscious Even largegovernment programs are subject to public scrutiny Inthe highly competitive world of commercial enterprises,those that are not cost conscious fail on a predictable and

579

Trang 27

V IRTUAL P RIVATE N ETWORKS : I NTERNET P ROTOCOL (IP) B ASED

580

regular basis Standing still is simply not good enough

The maturation of computing hardware and the

support-ing software has ushered in the postindustrial

informa-tion age Now, enterprises need to interconnect

employ-ees, databases, servers, afﬁliates, and suppliers in a rapidly

changing business environment Flexibility becomes an

overarching requirement Those enterprises that do not

adapt will not survive

Increased competition breeds the need for innovation

In traditional services and products, new, smaller

com-panies grab market share by offering new and innovative

services more rapidly, or by offering traditional services or

products at a lower cost The incumbents sometimes cry

foul, claiming that the newcomers are “cream skimming”

the lucrative market segments The newcomers counter

that the incumbents are the “fat cats,” who have all the

cream Although some monopolies do exist, either

reg-ulated or de facto, the pace of change is ever

accelerat-ing

The worldwide adoption of the Web is a great

equal-izer Even a small enterprise can have a large impact and

presence via the electronic Web that never sleeps The

user-friendly Web browser with downloadable plug-ins

empowers distribution of new paradigm-shifting

applica-tions within days to weeks The rapid adoption of

elec-tronic commerce will forever change the way business

operates and government administrates Enterprises are

rapidly deploying Web-based intranet and extranet

tech-nology to reduce internal costs, in many cases replacing

legacy mainframe-based systems

Communication networks continue to shrink the

dis-tances between nations, cultures, and time zones The

introduction of each new type of communication

tech-nology empowers the nearly instantaneous dissemination

of new media types around the globe Beginning with

the ﬁrst transatlantic telephone cable in 1956, the speed

of transfer of news and breaking information fell from

days to minutes Communications satellites ushered in

the era of video and multimedia distribution in the 1960s,

on the heels of the space age In the late 20th century,

high-capacity ﬁber optic transoceanic and

transcontinen-tal cables connected the planet, bringing the beneﬁts of

digital transmission to the corridors used by most

en-terprises This increase in high-performance connectivity

enables enterprises to scale beyond national boundaries,

particularly in the commercial and nonproﬁt sector, and it

also has an impact on governmental enterprises Witness

the lowering of national barriers in the European Union,

as an example

Most enterprises have some sensitive information that

would be of value to competitors or other parties

En-terprises trust the implicit security in private leased-line

networks In fact, a major impediment to the adoption

of VPNs is ensuring that this new technology delivers the

level of privacy and security that enterprises have come to

expect from private lines Toward this end, the

fundamen-tal security requirements of any VPN are the following

(Kosiur, 1998; Schneier, 1995; McDysan, 2000):

authen-tication, validating that originators are indeed who they

claim to be; access control, the act of allowing only

au-thorized users admission to the network; conﬁdentiality,

ensuring that no one can read or copy data transmitted

across the network; and integrity, guaranteeing that no

one can alter data transferred by the network

VPN approaches employ different methods to meetthese requirements These methods are sometimes im-plicit and sometimes explicit Security is a fundamentalrequirement for customer-edge (CE)-based VPNs operat-ing over the shared Internet infrastructure Of course,good security begins with secure practices For example,

if the employees of an enterprise leave their user IDs, words, or encryption keys lying around, then all the secu-rity technology in the world won’t protect sensitive infor-mation

pass-Most enterprises believe that quality of service (QoS),trafﬁc management, and prioritized or differentiated ser-vice will become an increasingly important driver in theirevolving communications needs Some applications, such

as voice and video, require rigid amounts of capacity andminimum levels of quality to operate acceptably Otherapplications, such as Web browsing, ﬁle transfers, ande-mail, are elastic and can adapt to available capacity to acertain extent However, even elastic applications result inlowered productivity and increase effective cost to the en-terprise if certain minimum-capacity and -quality guide-lines are not met Normally, an enterprise may also need

to prioritize or differentiate between these categories ofapplications to handle intervals of congestion

The primary QoS measures are loss, delay, jitter, andavailability Voice and video applications have the moststringent delay, jitter, and loss requirements Interactivedata applications such as Web browsing and electroniccollaboration have less-stringent delay and loss require-ments Non-real-time applications, such as ﬁle transfer,e-mail, and data backup, work acceptably across a widerange of loss rates and delay Availability requirementsvary across enterprises

Capacity, also referred to as bandwidth, is tal to the trafﬁc engineering of a VPN, which is necessary

fundamen-to deliver the required QoS Some applications require aminimum amount of capacity to work at all, for exam-ple voice and video The performance of elastic protocolsthat adaptively change their transmission rate in response

to congestion in the network improves as the capacityallocated to them increases The Internet’s transmissioncontrol protocol (TCP), which carries Web trafﬁc and ﬁletransfers, is an example of an elastic protocol Other ap-plications are elastic up to a certain point, after whichadding capacity does not improve performance

Many network providers guarantee speciﬁc QoS andcapacity levels via service level agreements (SLAs) AnSLA, which is a contract between the enterprise user andthe network provider, spells out the capacity provided be-tween points in the network that should be delivered with

a speciﬁed QoS If the network provider fails to meet theterms of the SLA, then the user may be entitled to a re-fund These have become popular capabilities offered atadditional cost by network providers for the private line,frame relay (FR), asynchronous transfer mode (ATM), orInternet infrastructures employed by enterprises to con-struct VPNs

Several approaches have been standardized for ing one or more of the above aspects of QoS The oldest isthe integrated services (Intserv) architecture (RFC 1633,

Trang 28

deliver-Virtual˙Private WL040/Bidgolio-Vol I WL040-Sample.cls August 14, 2003 17:53 Char Count= 0

I NTRODUCTION TO IP-B ASED V IRTUAL P RIVATE N ETWORKS 581

Braden, Clark, & Shenker, 1994) that uses the resource

reservation protocol (RSVP) (RFC 2210, Wroclawski,

1997) Intserv/RSVP allows a host to request one of

sev-eral levels of QoS at a speciﬁed level of capacity for a ﬂow

of packets speciﬁed by the IP address, transport protocol

port numbers, and/or protocol type The RSVP messages

normally follow the same hop-by-hop routed path as other

packets, and if the reservation is successful, then the

net-work provides the requested QoS for the level of capacity

reserved However, because RSVP signaling occurs at the

individual ﬂow, there is a signiﬁcant scalability issue in a

provider’s backbone network due to the signaling load for

a large number of ﬂows For this reason, Intserv/RSVP is

not supported in service provider networks and has seen

only limited use in enterprise networks

In responses to these issues, the IETF deﬁned other approach, which addresses the scalability issues of

an-Intserv/RSVP by treating only aggregates of ﬂows using a

convention called differentiated services (Diffserv) (RFC

2475, Blake et al., 1998) Diffserv redeﬁnes the

type-of-service (TOS) byte in the IP packet header in terms of a

small number of Diffserv code points (DSCPs), which

in-dicate the type of QoS the packet should receive Capacity

reservation at the individual ﬂow level of Intserv/RSVP is

avoided altogether and replaced by classiﬁcation and

traf-ﬁc conditioning (e.g., policing) performed only at the edge

of a DiffServ domain, for example a customer network or

a provider network Furthermore, because Diffserv

oper-ates only on ﬁelds within the IP packet header, it can

coex-ist with IP security protocols whereas Intserv/RSVP may

not, because it may rely on higher-layer protocol ﬁelds

(e.g., transport protocol port numbers) to identify an

in-dividual ﬂow

Most backbone IP networks will likely use DiffServ,possibly using a so-called bandwidth broker, which incor-

porates policy server functions and also deals with

cus-tomer trafﬁc contract and network resource allocation

A bandwidth broker maps service level speciﬁcations to

concrete conﬁgurations of edge routers of a DiffServ

do-main However, Intserv/RSVP or next-generation

reserva-tion signaling protocols still might have a role to play in

signaling reservations in enterprise networks and at the

edge of a service provider network, especially for such plications as digital audio and video, which would beneﬁtfrom reservations for relative long-lived, high-bandwidthﬂows (Braun, 2001)

ap-Introduction to Virtual Private Networks Technologies

A VPN attempts to draw from the best of both the publicand the private networking worlds Such a network is pri-vate in the sense that the data an enterprise transfers overthe VPN is separated and/or secure from that of other en-terprises or the public It is virtual in the sense that the un-derlying public infrastructure is partitioned to have somelevel of service for each enterprise A VPN is communica-tion between a set of sites making use of a shared networkinfrastructure, in contrast to a private network, which hasdedicated facilities connecting the set of sites in an en-terprise To a great extent, the intent is that the logicalstructure of the VPN, such as topology, addressing, con-nectivity, reachability, and access control, is equivalent topart or all of a conventional private network

A good VPN has the low-cost structure of a tous public network but retains the capacity guarantees,quality, control, and security of a private network Howcan a network design achieve these apparently contradic-tory goals? The answer lies in software-deﬁned network-ing technology, sophisticated communications protocols,and good old-fashioned capitalism

ubiqui-FR, ATM, multiprotocol label switching (MPLS), andthe Ethernet are all forms of layer 2 (L2) label-switchingprotocols (McDysan, 2000) A label is the header ﬁeld of

a packet, frame, or cell Labels are unique only to an terface on a device, such as enterprise user equipment or

in-a network switch Figure 1 illustrin-ates in-a simple exin-ample

of the operation of a simple two-port label switch A labelswitch uses the label header from the packet received on

an interface (left side of figure) as an index into a lookuptable in the column marked “In,” which identifies a spe-cific row From this row, the lookup table returns the out-going label from the column marked “Out” and the out-going physical interface from the column marked “Port.”

Trang 29

582

Figure 2: Example of two connection-oriented VPNs in a shared public network.

The switch routes the packet, frame, or cell to the outgoing

physical interface using an internal switching fabric and

“switches” the label to the outgoing label retrieved from

the lookup table The example in the ﬁgure uses patterns

for the packets to trace the result of the label-switching

operation implemented by the lookup tables on the input

side of each port Of course, contention may occur for the

output port in a label switch if multiple packets are

des-tined for the same output Typically, label switches must

implement some form of queuing to handle this

situa-tion

An L2 network consists of a number of label switches

implementing the basic function described above

Typi-cally, these switches also implement a number of other

features related to connection establishment, trafﬁc

con-trol, QoS, congestion concon-trol, and the like Some form of

routing, signaling, and/or network management protocol

establishes a consistent sequence of label-switching

map-pings in the lookup tables to form a logical connection that

can traverse multiple nodes When the network is

connec-tion oriented, for example in FR, ATM, and MPLS, we call

the allowed pairwise communication a virtual circuit or

connection (VC) For a connectionless L2 network, such

as the Ethernet, we call the set of sites that are allowed to

communicate a virtual local area network

Figure 2 illustrates a public connection-oriented

net-work supporting two disjoint VPNs Shaded boxes

repre-sent equipment from different enterprises at various sites

connected to triangles that represent provider-edge (PE)

label switches The label-switched connection-oriented

network implements disjoint virtual connections (either

permanent or switched) between different enterprise

nodes, as indicated by dashed lines of different styles in

the ﬁgure A connection-oriented label-switched network

operates very much like a private line network, but it uses

virtual connections instead of real ones The important

difference is that the service provider switches utilize label

switching instead of Time Division Multiplexing (TDM)

cross-connects to logically share trunk circuits between

multiple enterprise VPNs Thus, a connection-oriented

VPN can be a plug-compatible replacement for a

private-line-based network This has a number of advantages.First, the granularity of capacity allocation is much finerwith a label switch than with that implemented in the rigidTDM hierarchy Second, if the traffic offered by the enter-prises is bursty in nature, the service provider network canefficiently multiplex many traffic streams together Finally,the shared public network achieves economies of scale byutilizing high-speed trunk circuits that have a markedlylower cost per bit per second (bps) than lower-speed linksdo

X.25 was the ﬁrst connection-oriented data VPN, but

it is now being phased out X.25 pioneered a VPN cept called a closed user group (CUG), which is similar tothat of an intranet or extranet In the late 1980s, FR fol-lowed X.25, simplifying the protocol and, hence, improv-ing the price-performance ratio FR pioneered the impor-tant VPN concept of per-connection traffic managementand some simple responses to congestion ATM was thesuccessor to FR, in the mid-1990s, focusing on a fixedcell size to ease hardware implementation and achievehigh performance ATM borrows heavily from the sig-naling protocols of the narrowband integrated servicesdigital network (ISDN), the traffic management concepts

con-of FR, and automatic topology discovery from IP ATMstandards significantly extended the concept of QoS andmore precisely defined traffic management, these beingthe hallmarks of ATM In some ways, MPLS is an en-hancement of ATM: It provides most of the same capabili-ties but also adds some useful extensions and refinementstailored to the support of IP MPLS overcomes the inef-ficiency caused by the partial fill of the last fixed-lengthATM cell when carrying variable-length packets in AAL5.MPLS also supports a more flexible hierarchical aggrega-tion of connections and supports loop detection as well.The design of MPLS also allows tighter integration thandid ATM of connection-oriented traffic engineering with

IP routing protocols in service provider backbones sions of these capabilities are also quite useful in support

Exten-of network-based VPNs

A connectionless protocol like IP does not require asignaling protocol because it does not use connections

Trang 30

I NTRODUCTION TO IP-B ASED V IRTUAL P RIVATE N ETWORKS 583

B A

D

Figure 3: Example of two connectionless VPNs.

to forward user trafﬁc Instead, a routing protocol

dis-tributes topology information such that each node can

make an independent, yet coordinated, decision about the

next hop on which to forward packets that have a

partic-ular destination address preﬁx in the header Unlike label

switching, the addresses in packet headers must be unique

throughout a set of interconnected networks, such as the

Internet Therefore, the forwarding lookup table is

iden-tical in every node in a simple connectionless network

Because each address must be unique, the forwarding

ta-ble could become quite large The Internet scales to large

sizes by carefully administering address assignments so

that their forwarding tables need only process the

high-order preﬁx bits of the address

In a connectionless network, a VPN is a logical lay on a shared IP network of a different type A shared IP

over-network may be the public Internet or a over-network that

sup-ports IP routing protocols implemented speciﬁcally for

use by enterprise customers A secure IP VPN utilizes the

concept of an encrypted tunnel implemented at the

en-terprise equipment connected to the IP network A

tun-nel may exist at the link layer or the network layer as

an association between two endpoints attached to a

pub-lic network, therefore making it virtual Encryption is a

technique that scrambles information such that only the

intended receiver can decode it, thereby achieving privacy

Because an IP network is connectionless, the packets

be-tween enterprise nodes may take different paths,

depend-ing on such conditions as link failures or the conﬁguration

of routing parameters IP routing protocols synchronize

the forwarding tables in all the nodes whenever the state

of the network changes This fundamental difference in

paradigms is what has allowed the Internet to scale the

way it has in response to the tremendous demand that

arose in the latter half of the 1990s

Figure 3 illustrates a connectionless IP-based VPN fortwo enterprises The enterprise nodes are shaded boxes,

each with an IP address that has a preﬁx (e.g., A.1 and B.5)

associated with a triangle indicating the network router to

which the access line attaches (e.g., A and B) For example,

the gray-shaded enterprise node has an address prefix A.2connected to the network router with address prefix A.The figure illustrates the forwarding tables next to eachnetwork router Each table contains an entry labeled “In”for the incoming packet address prefix, which is used tolook up the next-hop outgoing port For example, at router

A, a packet received with destination address prefix B issent out on port 1 Note how these tables contain onlythe address prefix and the next-hop link number, and notthe enterprise node address prefixes Therefore, the enter-prise equipment at the edge of the network implementsthe IP VPN functions This architecture has a number offundamental advantages First, configuration changes tothe enterprise VPN do not require changes in the coreInternet Second, because the Internet is a global pub-lic network, a tunneled enterprise VPN can be implem-ented across multiple Internet service provider (ISP) net-works

Now we look at a categorization of logical VPN typesand the terminology used to describe them

A Taxonomy of IP-Based Virtual Private Networks

The taxonomy of VPN types is primarily determined bywhether the tunnels that provide the service terminate on

CE or PE devices (Carugi et al., 2002; Callon et al., 2002).Figure 4 illustrates the case where the tunnels terminate

on the CE A CE-based VPN is one in which knowledge ofthe service aspects of the customer network is limited to

CE devices Customer sites are interconnected via tunnels

or hierarchical tunnels, as deﬁned in the glossary The vice provider network is unaware of the existence of theVPN because it operates exclusively on the headers of thetunneled packets Speciﬁcally, a CE-based L2 VPN is alink layer (i.e., L2) service provided by CE equipment atthe customer sites, for example the Ethernet In a similarmanner, a CE-based L3 VPN is a network layer (i.e., L3)service provided by CE devices at customer sites, for ex-ample the IP

Trang 31

ser-Virtual˙Private WL040/Bidgolio-Vol I WL040-Sample.cls August 14, 2003 17:53 Char Count= 0

584

CEVPNA

CEVPNB

AccessNetworkPE

PE

Tunnel

CEVPNA

CEVPNB

AccessNetwork

Tunnel

Provider Network(s)

Figure 4: Generic customer edge (CE)-based VPN.

Figure 5 illustrates the case where the tunnels

termi-nate on the PE A PE-based VPN is one in which the service

provider network maintains state information for each

customer VPN such that packets are forwarded between

customer sites in an intranet or extranet context using

the customer’s address space Often, a hierarchical tunnel

is used between PEs, with the outermost tunnel being

im-plemented by a provider (P) router, which provides PE–PE

connectivity (Note that the P and PE functions are logical

and that a single router may implement both functions.)

These tunnels may be dedicated to separate VPNs or they

may be shared between multiple VPNs by the PEs, which

use label stacking to isolate trafﬁc between VPNs These

inner tunnels interconnect an L3 virtual forwarding (or

L2 switching) instance (VFI/VSI) for each VPN instance

in a PE switching router A PE-based L2 VPN provides an

L2 service that switches link-layer packets between

cus-tomer sites using the cuscus-tomer’s link-layer identiﬁers, for

example the Ethernet A PE-based L3 VPN provides an L3

service that routes packets between customer sites using

the customer network’s address space, for example the IP

The CE-based approach is the simplest from the

ser-vice provider backbone perspective, but it requires a fair

amount of conﬁguration and management of the CE On

the other hand, the network-based approach provides

greater control of trafﬁc engineering and performance,

but it incurs additional complexity in the backbone

net-work to achieve these beneﬁts The L3 PPVPN

frame-work document (Callon et al., 2002) further describes

these concepts in the context of a reference model that

deﬁnes layered service relationships between devices andone or more levels of tunnels The next sections coversome speciﬁcs of CE- and PE-based VPNs as they relate

to IP intranets and extranets

CUSTOMER-EDGE-BASED VIRTUAL PRIVATE NETWORKS

As deﬁned earlier, CE-based VPNs are partitioned by nels established between CE devices Routing inside thecustomer network often treats the tunnels as simple point-to-point links, or sometimes as broadcast local area net-works For customer-provisioned CE-based VPNs, pro-visioning and management of the tunnels is up to thecustomer network administration, which is also respon-sible for operation of the routing protocol between CEdevices In provider-provisioned CE-based VPNs, the ser-vice provider(s) perform provisioning and management ofthe tunnels and may also conﬁgure and operate routingprotocols on the CE devices Of course, routing within asite is always under control of the customer

tun-There are two primary types of IP CE-based VPNs, tinguished by the type of tunnel employed The ﬁrst isolder and is used primarily to construct intranets by us-ing CE routers connected via FR or ATM virtual connec-tions The second is newer and is based upon tunnels im-plemented using cryptographic methods over the publicInternet using either dedicated or dial-up access We nowdescribe each of these approaches

PETunnels

CEVPNA

CEVPNB

Access

Provider Network(s)

CEVPNA

CEVPNB

AccessNetwork

Figure 5: Generic PE-based (also called network based) VPN.

Trang 32

C USTOMER -E DGE -B ASED V IRTUAL P RIVATE N ETWORKS 585

FrameRelay or ATMVC’s

Branchsites

CE

Figure 6: CE-based VPN over a partial mesh of L2 hub-and-spoke VCs.

CE Virtual Private Networks Over Virtual

Connection Networks

The FR and ATM connection-oriented VPN alternatives

largely apply to a single service provider In order to

con-nect each site to every other site in a fully meshed network

of N number of sites, the service provider must provision

on the order of N squared virtual connections (VCs) Note

that each VC must be provisioned at every intermediate

FR or ATM switch in the service provider network As the

number of sites becomes large, service providers often

interconnect the sites, creating what is called

hub-and-spoke architecture, as shown in Figure 6 Often, the hub

sites are connected in a full mesh with branch sites

dual-homed to a primary and secondary hub site, as shown in

the ﬁgure Another motivation for the hub-and-spoke

de-sign is that with a full mesh of sites, addition of a new site

requires conﬁguration not only of the new site but of each

of the other VPN sites as well

The trafﬁc forwarded between the sites in a VPN is lated from all others by the logical separation provided

iso-by the virtual connections, which perform label

switch-ing as conﬁgured by a provisionswitch-ing system What results

is, for all practical purposes, a private network Such a

connection-oriented VPN is a good approach for intranets

because of the isolation and site-to-site trafﬁc engineering,provided by the approach is good

On the other hand, conﬁguring such a network for tranets can be complex and inﬂexible For these reasons,e-commerce applications tend to use IP security proto-cols as the foundation for CE-based VPNs that are used

ex-by many intranet and extranet applications

IP Security-Based Customer-Edge Virtual Private Networks

An analogous IP-based VPN network has the same ber of hub-and-spoke sites but requires the addition ofoverlay IP security (IPsec) tunneling and/or encryptionfunctions in the CE devices There is no explicit connec-tion through the devices in the service provider network.Instead, all the tunnel functions are implemented in the

num-CE devices Scaling issues similar to those in num-CE devicesoverlaid on virtual connections arise in IPsec CE-basedVPNs, but here the limits are the number of IPsec tunnelsand the number of routing adjacencies a CE router cansupport Therefore, large IPsec CE-based VPNs also have

a hub-and-spoke architecture, as described previously.Figure 7 illustrates the same hub-and-spoke network ex-ample, with circles showing the hub–spoke tunnels and

CE

CECE

Hubsites

Branchsites

TunnelEndpoints

Trang 33

586

SG R

Web Server

Extranet Data

Remote, Dial-in Users

Figure 8: Pure IP-distributed VPN design.

squares showing the hub–hub tunnels As with the VC

overlay approach, adding a new site to a full mesh

re-quires conﬁguration of a tunnel to every other site

Fur-thermore, if the enterprise does not use globally unique,

routable IP addresses, the CE devices may also include

network address translation functions When a single ISP

provides the network for an IP-based VPN, then

guaran-tees on quality and performance are feasible Beware of

an IP-based VPN built on top of the public Internet using

services provided by several ISPs: It may not provide the

quality necessary for telephone-grade voice or multimedia

applications

The IETF designed the IPsec protocol suite to address

the known issues involved with achieving secure

commu-nications over the Internet (McDysan, 2000) It reduces

the threat of attacks based on IP address spooﬁng and

provides a standardized means for ensuring data integrity,

authenticating a data source, and guaranteeing

conﬁden-tiality of information Furthermore, it tackles the complex

problem of key management head on When a public key

management infrastructure is used, the Internet can be

trusted based upon this set of standards IPsec will play

an important role not only in enterprise VPNs, but also

in electronic commerce and in secure individual end user

communication

IPsec refers to a suite of three interrelated security

pro-tocols implemented by modiﬁcation to, or augmentation

of, an IP packet in conjunction with an infrastructure that

supports key distribution and management An

interre-lated set of Request for Comments (RFCs) published by

the IETF speciﬁes the details of IPsec RFC 2401 (Kent

and Atkinson, 1998) describes the overall IP security

ar-chitecture, whereas RFC 2411 (Thayer et al, 1998) gives

an overview of the IPsec protocol suite and the

docu-ments that describe it Three protocols make up IPsec,

with the names identifying the function performed The

two primary protocols involved in the transfer of data are

called the authentication header (AH) and the

encapsu-lating security payload (ESP) The AH protocol provides

source authentication and data integrity veriﬁcation

us-ing a header ﬁeld, but it does not provide conﬁdentiality

AH also supports an optional mechanism to prevent

re-play attacks The ESP protocol uses both a header and a

trailer ﬁeld to provide conﬁdentiality via encryption ESP

may also provide data integrity veriﬁcation, source

au-thentication, and an antireplay service Because both the

AH and the ESP protocols utilize cryptographic methods,secure distribution and management of keys is a funda-mental requirement IPsec speciﬁes that key managementmay be manual or automatic The automatic key man-agement protocol speciﬁed for IPsec is called Internet keyexchange and involves the mechanism for creating a secu-rity association (SA) between a source and a destinationfor the AH and ESP protocols

The AH and ESP protocols operate in either transport

or tunnel mode, as deﬁned by the parameters of an SA

In transport mode, they provide security by creating

com-ponents of the IPsec header at the same time the sourcegenerates other IP header information This means thattransport mode can operate only between host systems

In tunnel mode, IPsec creates a new IP packet, which

con-tains the IPsec components and encapsulates the originalunsecured packet Because tunnel mode does not modifythe original packet contents, it can be implemented usinghardware or software located at an intermediate securitygateway (SG) between the source or destination system.Figure 8 illustrates a pure IP-based VPN design thathas a cost structure essentially independent of the traf-

fic pattern Here, every site has a firewall and securitygateway, so any site may directly access the Internet orany other site In addition, we show a network accessserver (NAS), remote authentication dial-in user service(RADIUS) server, Web server, and extranet database lo-cated at three separate sites Dial-in users are secured us-ing the RADIUS server and the SG This design also re-duces access costs because traffic for the Internet neednot traverse a firewall at a headquarters site, as shown inthe hierarchical example above Sites may also be dual-homed to different ISPs or to different sites within thesame ISP for resiliency purposes, as necessary This de-sign is better suited to extranet applications and electroniccommerce because communication via the public Inter-net is more interoperable and rapidly deployable than anyother communication service

PROVIDER-EDGE-BASED LAYER 3 VIRTUAL PRIVATE NETWORKS

A PE-based VPN is one in which PE devices in the serviceprovider network provide the partitioning of forwardingand routing information to only those (parts of) sites thatare members of a speciﬁc intranet or extranet This allows

Trang 34

P ROVIDER -E DGE -B ASED L AYER 3 V IRTUAL P RIVATE N ETWORKS 587

PE

P

CE A

CE B

PE

CE C

CE A

PE P

CE C

CE A

PE

CE A

CE B

P

PE

CE C

CE B

Virtual Forwarding Tables

PE

CE B

CE A

Shared Tunnels

A

B

A C

Figure 9: Aggregated routing and shared tunnel network-based L3 VPN.

the existence of the VPN to be hidden from the CE

de-vices, which can operate as though they were part of a

normal customer network As described earlier, PE-based

VPNs use tunnels set up between PE devices These

tun-nels may use one of a number of encapsulations to send

trafﬁc over the provider network(s), for example MPLS,

generic routing encapsulation, IPsec, or IP-in-IP As sites

for new VPNs are added or removed, PE-based VPN

solu-tions provide a means of distributing membership

infor-mation automatically There are two principal methods

deﬁned in the IETF (Callon et al., 2002) for implementing

these types of PE-based VPNs, namely aggregated routing

and virtual routers, which we now describe

Aggregated Routing Virtual

Private Networks

The aggregated routing approach is one in which a

sepa-rate forwarding table exists for each VPN on every PE that

connects to a site in that VPN but where the exchange

of routing information between the PEs is multiplexed,

or aggregated together The BGP/MPLS VPN (RFC 2547,

Rosen & Rekhter, 1999) approach uses extensions to the

border gateway protocol (BGP) to implement this generic

architecture Figure 9 illustrates an example of this

ap-proach, connecting sites from three VPNs, A, B, and C,

in an extranet Each PE has a separate virtual

forward-ing table for each VPN site that it serves, but the

for-warded trafﬁc and exchanged routing information uses a

set of shared tunnels, as shown in the center of the ﬁgure

Often these types of solutions are implemented on a

sin-gle service provider network However, there are some

implementations across more than one provider

net-work

This approach alleviates some of the scaling issues volved with the connection- or tunnel-oriented CE-based

in-approaches described earlier when full communication

between a set of sites is desired Speciﬁcally, when adding

or removing a site, only the PE involved with that siteneed be reconfigured—the BGP/MPLS protocols automa-tically take care of the rest Furthermore, the protocolshave the capability of advertising to their peers more thanone route for the same destination address This can beuseful in an extranet to force traffic exchanged betweendifferent enterprises through additional devices, such asfirewalls or filters

Virtual Router Virtual Private Networks

Although the virtual router (VR)-based approach (RFC

2917, Muthukrishnan & Malis, 2000) also uses PE and Prouters, there are several important differences, as illus-trated in Figure 10 This example uses the same CE sitesfrom the three VPNs discussed in the aggregated routingexample above In a VR VPN, a VR is dedicated to eachVPN in every PE that supports a site for that VPN Thismeans that each enterprise can manage its own routing

on the VR in the PE This works very well in cases wherethe enterprise network has other forms of connectivity be-tween its sites: The VRs look like just another (well con-nected) router to the enterprise network Usually, a sepa-rate set of tunnels is allocated in a full mesh between theVRs, as shown by different line styles in the center of theﬁgure This allows excellent control of capacity allocationand control of QoS between the VPN sites

The VR PE-based VPN is best suited for intranets It isnot frequently used in an extranet because one enterprisewould have to exchange routing information with another.This could lead to undesirable security holes, instability ofthe routing, and, hence, a greater likelihood of an outage,

as well as more difﬁcult coordination in the event of theinevitable moves, adds, and changes It could be used,however, as a backbone network provided by one partnerfor connecting a number of other enterprises together, forexample using CE-based VPNs overlaid on a managed VRPE-based network

Trang 35

588

PE

P

CE A

CE B

PE

CE C CE A

PE P

CE C CE A

PE

CE A

CE B

P

PE CE C

CE B

Virtual Router Instances

PE

CE B

CE A

Per VPN Tunnels

A

B

A C

This section summarizes some important considerations

when choosing a VPN approach and gives an example of

a CE-based IPsec VPN used for electronic commerce

Considerations When Choosing a Virtual

Private Networks Approach

Establishing a set of goals and establishing a plan to meet

them is critical to success in most human endeavors,

and virtual private networking is no exception (McDysan,

2000) The steps here are similar to that of any

large-scale project First, researching requirements, drivers,

and needs is necessary to establish goals Next,

develop-ing several candidate designs and analyzdevelop-ing them in the

harsh light of commercial business reality is a crucial step

A VPN may not be right for the enterprise under

consid-eration at this time, and timing is important Finally, a

decision to implement a new type of VPN or to migrate

existing private network applications to a VPN, is but the

ﬁrst step of many Detailed planning and a well thought

out migration strategy are essential for an enterprise to

achieve its goals identiﬁed in the ﬁrst step above

A number of enterprises have already implemented

VPNs of the types described in this chapter A good

start-ing point is to look at an enterprise that is similar to yours

in some way and to read case studies, papers, and books

about what worked and what did not However, be aware

that the needs of each enterprise are unique, and

there-fore basing a decision upon another’s experiences, while

helpful, cannot guarantee that goals will be met

An important area of requirements research is

analy-ses of potential security threats and essential performance

metrics Formulating a threat model and considering

what would happen if important information were stolen,made public, or corrupted is an essential step Deter-mining the performance required by applications is alsoimportant Consider what would happen if a site weredisconnected for a long period of time Assess what theimpact of network congestion would be Discriminate be-tween what would be nice to have and what is absolutelynecessary in the way of performance—this can make quite

a difference in qualifying network designs and their tual cost

evAlthough a generic framework may not apply to all terprises, there are some helpful points to consider whencategorizing types of requirements One way to analyzeVPN requirements is to consider the community of in-terest and the access methods: cost-effective remote andmobile user access; an infrastructure for intranets thatkeeps resources secure within a single enterprise; an in-frastructure for extranets for controlling resource sharingbetween two or more enterprises

en-The economic crossover point regarding enterprisedial-in versus ISP-provided access services centers aroundthe number of users that require dial-in access and thetype as well as amount of activities these users conduct

In general, a remote user population that generates burstyactivity during relatively long duration sessions is a goodcandidate for ISP access As described earlier, most VPNtechniques differ in the degree of trafﬁc separation andcontrol that an enterprise can have in an intranet context

On the other hand, if a driving requirement for the prise is extranet connectivity, then an IPsec-based solution

enter-is one of the few choices available (for more information,see VPN Consortium, 2003)

Because this is such an important case in the world ofelectronic commerce, we now look at an example where afew large enterprises worked with a number of small-to-medium-size enterprises to create a successful model forextranet deployment

Trang 36

ANX TP

ANX TP IPsec

CSP

Public Internet

CSP

CSP CEPO

ANX Overseer

IPsec

IPsec CASP

ANX Extranet

Figure 11: ANX extranet architecture.

Example of Deployment of a

Customer-Edge-Based Virtual Private

Networks in E-commerce

Unless your enterprise is the ﬁrst to try a new

techno-logy, protocol, or architecture, there will likely be case

studies available for review A frequently documented

ex-tranet case study is the Automotive Network eXchange

(ANX) (McDysan, 2000) This extranet VPN involves a

few large enterprises (automotive manufacturers) and a

signiﬁcant number of small-to-medium-size enterprises

(their suppliers) Initiated by the Automotive Industry

Ac-tion Group (AIAG) in 1994, the IPsec-based ANX network

had Chrysler, Ford, and General Motors as the founding

network participants These companies and other major

automotive manufacturers utilize parts and services from

a large number of common original equipment

manufac-turers, such as Bosch, Delta, Fisher, ITT, and TRW

Follow-ing the completion of successful trials in 1997 and 1998,

ANX launched production in November 1998 By the end

of 1999, ANX had nearly 500 registered trading partners

As an example of a quantiﬁable goal achievable by an

ex-tranet, the AIAG estimates that a collaborative planning,

forecasting, and replacement tool running over the ANX

network may save up to $1,200 per vehicle This savings

results from a reduction of the delivery cycle of parts and

supplies and the associated inventory levels

The ANX architecture is based upon a set of terconnected certiﬁed service providers (CSPs), certiﬁed

in-exchange point operators (CEPOs), and certiﬁcate

author-ity service providers to which ANX trading partners

sub-scribe, as illustrated in Figure 11 Telcordia (formerly

Bell-core) has been chosen as the ANX overseer, which awards

certiﬁcation to CSPs and CEPOs The ANX service

qual-ity certiﬁcation categories are network service features,

interoperability, performance, reliability, business

conti-nuity and disaster recovery, security, customer care, and

trouble handling ANX has also speciﬁed that the

Interna-tional Computer Security Association (ICSA) will certify

whether equipment is IPsec compliant

Finding companies with equipment that has the ICSAstamp of approval is a good place to start when looking

for IPsec-compliant vendors

This network is effectively a partitioned set of faces running on top of the public Internet infrastructureoffered by the selected set of certified commercial ISPs Itreplaces the prior complex arrangement of physical andlogical connections between trading partners with onelogically administered, cryptographically secured connec-tion to the ANX extranet Choice of the TCP/IP protocolsuite provides access to a broad range of file transfer,electronic document interchange, e-mail, and other ap-plication software This is especially important in the au-tomotive industry, where computer-based techniques arenow used in almost every stage of the design, manufac-turing, delivery, and maintenance aspects of the business.Although the benefits of ANX apply primarily to medium-to-large-size enterprises in the automotive industry, thedrive toward interoperability will benefit other industrysegments in the longer term (for more information, seewww.anx.com)

inter-GLOSSARY

Customer-edge (CE) device Provides access for users

at a site and has an access connection to a PE device Itallows users at a site to communicate over the accessnetwork with other sites in the VPN

Enterprise A single organization, corporation, or ernment agency that administratively controls and setspolicy for communication among a set of sites

gov-Extranet Allows communication between a set of sitesthat belong to different enterprises, as controlled bythe enterprise administrators and/or a third party.These enterprises have access to a speciﬁed subset

of each other’s sites Examples of extranets include(a) companies performing joint software develop-ment, (b) a group of suppliers and their customersexchanging orders and delivery tracking informa-tion, and (c) different organizations participating in

a consortium that has access to important tion

informa-Generic routing encapsulation (GRE) A general tocol for encapsulating a network layer protocol overanother network layer protocol (RFC 2784, Farinacci,

pro-Li, Hanks, Meyer, & Traina, 2000)

Trang 37

590

Intranet Restricts communication to a set of sites that

belong to one enterprise and via policy may further

restrict communication between groups within these

sites For example, communication between marketing

and engineering may be limited

IP security protocol (IPsec) A set of IETF standards

that deﬁnes a suite of security protocols that provide

conﬁdentiality, integrity, and authentication services

(RFC 2401, Kent & Atkinson, 1998)

Layer 2 tunneling protocol (L2TP) An IETF

standard-ized protocol deﬁned initially for support of dial-in

con-nections (RFC 2661, Townsley, et al., 1999) A

succes-sor to the proprietary Microsoft PPTP and Cisco L2F

protocols, L2TP gives mobile users the appearance of

being on an enterprise LAN

Multiprotocol label switching (MPLS) A switching

technique that forwards packets based upon a

ﬁxed-length label inserted between the link and network

layer or that uses a native layer 2 label, such as FR or

ATM (RFC 3031, Rosen, Viswanathan, & Callon, 2001)

Similar to frame relay and ATM in function, MPLS

dif-fers from these protocols by virtue of its tight coupling

to IP routing protocols

Provider-edge (PE) device A PE device faces the

ser-vice provider core network on one side and interfaces

via an access network to one or more CE devices

Site A set of users who have connectivity without use

of a service provider network, for example users who

are part of the same enterprise in a building or on a

campus

Tunnel Formed by encapsulating packets with a header

used to forward the encapsulated payload to the

tun-nel end point In VPN applications, tuntun-nel end points

may be a CE or a PE device Encapsulating one tunnel

within another forms a hierarchical tunnel, which is

useful for reducing the number of tunnels in the core

of networks Examples of protocols commonly used for

forming a tunnel are MPLS, L2TP, GRE, IPsec, and

IP-in-IP tunnels

User Someone or something that has been authorized

to use a VPN service, for example a human being using

a host or a server

Virtual private network (VPN) A speciﬁc set of sites

conﬁgured as either an intranet or an extranet to

al-low communication A set of users at a site may be a

member of one or many VPNs

CROSS REFERENCES

See Circuit, Message, and Packet Switching; Electronic

Commerce and Electronic Business; Extranets; Internet

Ar-chitecture; Internet Literacy; Internet Security Standards;

Intranets; Public Networks; TCP/IP Suite.

REFERENCES

ANX Network (2003) Retrieved February 10, 2003, from

http://www.anx.com/

Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z.,

& Weiss, W (1998) An architecture for differentiated

services Retrieved February 20, 2003, from http://ietf.

org/rfc/rfc2475.txt

Braden, R., Clark, D., & Shenker, S (1994) Integrated

ser-vices in the Internet architecture: An overview Retrieved

February 20, 2003, from http://ietf.org/rfc/rfc1633.txtBraun, T., Guenter, M., & Khalil, I (2001, May) Manage-

ment of quality of service enabled VPNs IEEE

Com-munications Magazine.

Callon, R., Suzuki, M., DeClerq, J., Gleeson, B., Malis,A., Muthukrishnan, K., RosenE., Sargor, C., & Yu, J

(2002) A framework for layer 3 provider provisioned

vir-tual private networks Unpublished manuscript.

Carugi, M., McDysan, D., Fang, L., Nagarajan, A.,

Sum-imoto, J., & Wilder, R (2002) Service requirements

for provider provisioned virtual private networks

Manu-script in preparation

E-mail list logs, presentations, related ITU-T drafts(2003) Retrieved February 20, 2003, from http://ppvpn.francetelecom.com

Farinacci, D., Li, T., Hanks, S., Meyer, D., & Traina, P

(2000) Generic routing encapsulation (GRE) Retrieved

February 20, 2003, from http://ietf.org/rfc/rfc2784.txtIETF working group charter page, list of RFCs and cur-rent drafts (2003) Retrieved February 20, 2003, fromhttp://ietf.org/html.charters/ppvpn-charter.html

Kent, S., & Atkinson, R (1998) Security architecture for

the Internet protocol Retrieved February 20, 2003, from

http://ietf.org/rfc/rfc2401.txt

Kosiur, D (1998) Building and managing virtual private

networks New York: Wiley.

McDysan, D (2000) VPN applications guide New York:

Wiley

Muthukrishnan, K., & Malis, A (2000) A Core MPLS IP

VPN architecture Retrieved February 20, 2003, from

Rosen, E., & Rekhter, Y (1999) BGP/MPLS VPNs

Re-trieved February 20, 2003, from http://ietf.org/rfc/rfc2547.txt

Rosen, E., Viswanathan, A., & Callon, R (2001)

Multipro-tocol label switching architecture Retrieved February

20, 2003, from http://ietf.org/rfc/rfc3031.txt

Schneier, B (1995) Applied cryptography: Protocols,

algo-rithms, and source code in C New York: Wiley.

Thayer, W., Doraswamy N., & Glenn, R (1998) IP

Secu-rity Document Roadmap Retrieved February 20, 2003,

from http://ietf.org/rfc/rfc2411.txtTownsley, W., Valencia, A., Rubens, A., Pall, G., Zorn, G.,

& Palter, B (1999) Layer two tunneling protocol

L2TP Retrieved February 20, 2003, from http://ietf.org/

rfc/rfc2661.txt

Wroclawski, J (1997) The use of RSVP with IETF

in-tegrated services Retrieved February 20, 2003, from

Virtual Private Network Consortium (2003) Retrieved

February 20, 2003, from http://www.vpnc.org/

Trang 38

WL040A-22 WL040/Bidgolio-Vol I WL040-Sample.cls June 20, 2003 17:34 Char Count= 0

Virtual Reality on the Internet: Collaborative

Synchronous and Asynchronous Work 596

Heterogeneous Views and Abilities 597

Collaborative virtual reality—sharing immersive

compu-ter-generated environments over the high-speed

net-works—is a next-generation interface that will allow

col-laborators on different continents to share a space where

they can interact with each other and with the focus of

their collaboration This text describes ongoing work in

this area at the Electronic Visualization Laboratory at the

University of Illinois at Chicago We ﬁrst discuss what

we mean by the term virtual reality and what the focus

is of our work in collaborative virtual environments We

then discuss the types on information that must be sent

through the networks to maintain these collaborations

Finally, we describe current research in the areas of

asyn-chronous collaboration and heterogeneous perspectives

and conclude with a discussion of what we see as the

future of collaborative virtual environments

VIRTUAL REALITY

Before we discuss collaborative virtual reality, we should

deﬁne what we mean by virtual reality Different

disci-plines have different deﬁnitions for what virtual reality is

and what hardware is required A good novel is a form

of virtual reality that requires no special hardware to be

experienced For our purposes, virtual reality requires

computer-generated stereo visuals, viewer-centered

per-spective, and an ability to interact with the virtual world

Computer-generated stereo visuals allow the user to seethe computer-generated world in three dimensions (3D),

which is how most (but not all) people see the real world

Each eye sees the world from a slightly different position,

allowing us to perceive depth As with the viewing of stereo

photographs or the watching of a 3D movie from the 1950s

or 1980s, the trick is to give each eye its own view of the

material

Viewer-centered perspective allows the user to movehis body or turn his head and the see the appropriate view

of the virtual world from this new position Combined

with stereo visuals, this allows the user to not only see a

3D object in the virtual world but to walk around it or look

under it by moving in exactly the same way as a personwould move around a real 3D object In a 3D movie orphotograph, the viewing position is static—the viewersees only what the camera saw With viewer-centeredperspective, the viewer is the camera and always has thecorrect view of the scene For this to work, the computergenerating the visuals needs to know where the viewer’stwo eyes are

There are several different ways to do stereo visualsand head tracking, which lead to different virtual realitydisplay hardware With a head-mounted display (HMD),the user wears a headset, which isolates her from the realworld, with a small cathode ray tube (CRT) or liquid crys-tal display (LCD) devoted to each eye This allows the user

to turn and tilt her head in any direction and still see thevirtual world A tracker attached to the HMD tells the com-puter the position and orientation of the user’s head Withthat information, the computer can determine where theuser’s eyes are and then draw the graphics appropriately

A ﬁsh tank virtual reality system makes use of a

com-puter monitor and a special pair of tracked LCD shutterglasses The computer monitor displays an image for theuser’s left eye, at the same time telling the glasses to blockout the user’s right eye The computer then does the re-verse, showing an image for the right eye while telling theglasses to block out the left eye By doing this quickly,the user can see objects ﬂoating in front of the monitor.The LCD shutter glasses are lighter than a HMD and don’tisolate the user from the real world A tracker attached tothe LCD shutter glasses gives the position and orientation

of the user’s head

This same technique can be used on a larger scale tocreate a single, large-drafting-table-size display, such asthe ImmersaDesk With a larger back-projected display,Rseveral people can stand in front of the display at the sametime and see the virtual world in stereo, but only one per-son is head tracked

Moving from a single large screen to several largescreens in a system like the CAVE allows the user toR

physically walk around virtual objects A CAVE typicallyhas three 10-ft2 walls and a 10-ft2 ﬂoor, although some

591

Trang 39

V IRTUAL R EALITY ON THE I NTERNET : C OLLABORATIVE V IRTUAL R EALITY

592

Figure 1: The CAVE and ImmersaDesk (Left) A person in the CAVE wearing tracked shutter glasses to see the virtual world in

stereo and carrying the wand (Right) A user sitting in front of the ImmersaDesk wearing the same tracked glasses and carryingthe same wand as in the CAVE The CAVE and ImmersaDesk users can interact with the same virtual worlds from different pers-pectives

CAVEs have four walls, a ceiling, and a ﬂoor to completely

surround the viewers This larger space allows ﬁve people

to comfortably view the virtual world together, although

again only one person is head tracked (Figure 1)

Another approach is to take multiple screens and,

in-stead of wrapping them around the user, use them to give

the viewer a higher resolution wall made up of several

screens A single screen, whether in a ﬁsh-tank virtual

reality setup or a CAVE, typically has a resolution of 1280

pixels by 1024 pixels By combining several screens

to-gether, much higher resolutions are possible

There are many different ways of interacting with the

virtual world and many different devices to allow that

in-teraction The user may want to navigate a large space

with a joystick or use a set of buttons to change the

proper-ties of the virtual world Just as the user’s head is tracked,

other parts of the user’s body can be tracked, so the user’s

body can itself be the interface It’s typical to track the

user’s hand, or the controller the user is holding, to allow

the computer to see where the user is pointing

Although visuals are the most obvious element of

virtual reality, audio is also important, to give the users

additional feedback Haptics, the feeling of touch, is also

important in certain virtual reality applications Often a

lack of feedback to one sense is compensated for by

feed-back to another sense For example, if you don’t have

hap-tic feedback, you may get visual or audible feedback

In order to keep up the illusion, the imagery of the

vir-tual world must be drawn at a rate of at least 15 frames

per second per eye Otherwise, the world will seem to

stut-ter In a movie theatre, we watch ﬁlms composed of still

images moving at 24 frames per second and see smooth

motion; it’s the same in virtual reality This is why

vir-tual reality requires very powerful computers and

graph-ics cards

Throughout the 1990s, this required very expensive

computers, but now it is possible to do single-screen

vir-tual reality using high-end personal computers There is

also current research going on in autostereoscopic

dis-plays, where the user will not need to use special glasses

to see computer-generated stereo imagery For a more

thorough discussion of virtual reality, see Sherman andCraig (2002)

COLLABORATIVE VIRTUAL REALITY

In the 1990s, more and more groups around the worldgained access to virtual reality equipment, making col-laborative virtual reality possible Again, there are sev-eral deﬁnitions of collaborative virtual reality—every daymany people play collaborative or competitive games onthe Internet, which can be considered collaborative virtualreality, and sometimes share environments with hundreds

of other players Since the 1970s, text-based multiuser tual worlds such as MUDS and MOOS have been popular,evolving from their origins as collaborative adventuregames and allowing people to communicate and interactover very-low-bandwidth connections In the mid-1990s,with advances in both computing power and networkspeeds, users could explore 3D worlds over the Inter-net through VRML (virtual reality modeling language)browsers

vir-For our purposes, collaborative virtual reality requiresconnecting up the devices described in the previoussection, allowing people in several places to share a 3Denvironment Some research groups focus on support-ing existing low-bandwidth Internet infrastructures ormassive connectivity involving thousands of participants

at the same time, as in military simulations or based computer games (Singhal & Zyda, 1999) Ourfocus on the use of virtual reality for manufacturing, forscientiﬁc purposes, and for information visualization has

Internet-a different set of requirements We Internet-are building systemsfor small working groups, typically no more than sevencollaborators at a time but with large data distributionrequirements, to share high-ﬁdelity audio and videocommunications and large engineering and scientiﬁcdata stores over high-speed national and internationalnetworks

We want to provide high-quality interaction betweensmall groups of participants involved in design, train-ing, education, scientiﬁc visualization, or computational

Trang 40

steering The ultimate goal is not to reproduce a

face-to-face meeting in every detail, but to provide the

next-generation interface for collaborators, worldwide, to work

together in a virtual environment that is seamlessly

en-hanced by computation and access to large databases

Al-though the goal of audio and video teleconferencing is to

allow distributed participants to interact as though they

are in the same physical location, collaborative virtual

reality allows them to interact as though they are the same

immersive virtual environment This way they can

inter-act with each other as well as the objects in their shared

environment

This shared environment may be for designing a newcar, visualizating climatological data, or visiting other 3D

space that either does not exist physically or cannot be

physically accessed The participants are not talking about

a thunderstorm, they are standing inside one; they are

not looking at a scale model of a new car design, they

are standing inside the full-size engine block We believe

that by transmitting gestures as well as audio and video

between collaborators, these shared virtual environments

give their users a greater sense of presence in the shared

space than do other collaborative mediums By

encourag-ing collaboration and conversation within the data, these

environments may become the preferred place to work

and interact even when traditional face-to-face meetings

are possible However, collaborative virtual reality is not

going to replace e-mail, phone calls, or existing

telecon-ferencing systems They each have their strengths and

uses Just as word processing documents, spreadsheets,

and white boards shared across the Internet put

discus-sions in their appropriate contexts, so does sharing a

vir-tual space as well as the 3D design being considered or the

simulation being visualized A more thorough discussion

can be found elsewhere (Leigh, Johnson, Brown, Sandin,

& DeFanti, 1999)

For example, General Motors uses collaborative virtualreality to allow design and manufacturing teams based in

several sites around the world to import 3D

computer-aided design (CAD) models into the CAVE for quick

vi-sual inspection and design reviews at 1:1 scale The goal

is to allow designers to both synchronously and

asyn-chronously access a design that persists and evolves over

time from locations scattered around the world rather

than forcing collaborators to meet physically at the 1:1

scale clay model A typical working scenario involves a

designer making modiﬁcations on a workstation in a 3D

modeling package and having those changes propagate

automatically to the networked virtual environment,

al-lowing all collaborating participants to see the changes

simultaneously They are then able to critique the design

and suggest changes to the designer who can do so

imme-diately at the CAD work station

The Virtual Reality in Medicine Laboratory at the versity of Illinois at Chicago uses collaborative virtual

Uni-reality to allow a remotely located physician to teach

med-ical students about the 3D structure and function of the

inner ear In this environment, the students and

instruc-tor may point at and rotate the ear to view it from various

perspectives They may also strip away the surrounding

outer ear and temporal bone to more clearly view the

in-ner anatomy Audio from the voice conference is used to

modify the flapping of the eardrum to illustrate its tion This application is effective because it leverages thestereoscopic capabilities of virtual reality to disambiguatethe spatial layout of the various structures in the innerear—something difficult to do on standard flat images inmedical textbooks

func-College undergraduates at Central Missouri State versity and other universities use Virtual Harlem, a vir-tual reality reconstruction of Harlem, New York, duringthe 1920s, in their English classes Virtual Harlem wasdesigned to immerse students of the Harlem Renaissancedirectly in the historical context of the literature of thatperiod to reinforce active learning The goal is to developrich, interactive, and narrative learning experiences toaugment classroom activities for students in the human-ities Collaborative virtual reality allows classes at differ-ent universities to meet and share their views within itsspace, as well as allowing remote expert tour guides totake classes through Virtual Harlem and discuss impor-tant issues that the space brings up This is discussed fur-ther elsewhere (Sosnoski & Carter, 2001)

Uni-Some virtual environments will only exist while peopleare inside it; others will be maintained by a computer sim-ulation that is constantly left running This space existsand evolves over time Users enter the space to check onthe state of the simulated world, discuss the current situ-ation with other collaborators in the space, make adjust-ments to the simulation, or leave messages for collabora-tors who are currently asleep on the far side of the planet.For example, in a computational steering application, asupercomputer may be running a large simulation thattakes several days to complete At regular intervals, the su-percomputer produces a 3D snapshot of the current data,perhaps a visualization of cosmic strings A scientist canthen step into a CAVE and look at the 3D data that has beenproduced to see whether the simulation is progressing cor-rectly or whether it needs be tuned, to focus on particulardetails rather than wait for the simulation to complete

Avatars

Presence in the virtual world is typically maintained

us-ing an avatar, or a computer-generated representation of

a person These avatars may be as simple as a pointerthat depicts the position and orientation of the wand inthe virtual world However, having representations of thephysical bodies of the collaborators can be helpful in aid-ing conversation and understanding in the virtual space,

as you can see where your collaborators are and whatthey are looking at or pointing at Tracking the user’s headand hand position and orientation allows the computer

to draw computer-generated characters representing each

of the remote collaborators These articulated charactersmove along with the remote user and are able to transmit areasonable amount of body language, such as pointing atobjects and nodding or tilting the head This style of avatar

is useful in task-oriented situations, but do not work aswell in negotiations

Seeing high-quality live video of a person’s face can prove negotiations Video avatars, full-motion full-bodyvideos of users, are realistic looking, which improvesrecognition of collaborators but require much higher

Tiêu đề	The Internet Encyclopedia 1 Volume 3 Phần 7
Trường học	University of Example
Chuyên ngành	Information Technology
Thể loại	Tài liệu
Năm xuất bản	2023
Thành phố	Example City

Định dạng
Số trang	98
Dung lượng	2,31 MB