Scalable voip mobility intedration and deployment- P17 pdf

These streams go over the air at the same time, and the receiver uses multiple antennas to pick up this transmission, applies some math to separate back out the streams, and then recomb

180°

90°

-90°

16-QAM

quadrature axis (Q)

x

x

x

0°

180°

90°

-90°

64-QAM

quadrature axis (Q)

x

x

x x

x

x

x

x

x

x

x x

x

in-phase axis (I)

in-phase axis (I)

Figure 5.21: 16-QAM and 64-QAM

source bits and record whether that sum is even or odd as one additional bit If one bit is lost, the sum of the surviving bits can be compared in evenness to the sum bit, therefore

recovering the original data This is called a parity check code, and conceptually introduces

the concept of using arithmetic on some or all of the source bits to produce the extra bits

802.11ag uses a convolutional encoder to expand the source bits, in this case producing twice

the number of bits but, unlike the doubler, doing so more intelligently to avoid some loss

patterns of equal bit losses being worse than others This expansion still produces only one

uses the property that the error-correcting code can tolerate loss, and goes ahead and starts

tossing bits to get to a higher coding rate This process, called puncturing, sounds inefficient

by reducing redundancy that was just added (why not just not add as many extra bits in the first place?), but ends up saving on the complexity of the coding hardware, the radio

The overall picture, then, for 802.11ag encoding is for the data bits to be scrambled (as with 802.11b), then expanded with the error-correcting code, then split among the subcarriers, and then modulated

Because of the larger number of things going on in the signal, there is more risk of losing data if the receiver’s timing goes off a bit from the sender’s To compensate, timing is

maintained by the addition of four pilot subcarriers These carry a known signal pattern—

like the preamble—but do so for the length of the transmission If the receiver’s clock

speeds up or slows down relative to the sender’s, then the constellation would essentially rotate, and as the constellation points are now closer together than with 802.11b, the

rotation would cause the receiver’s bits to jumble The pilot subcarriers’ known pattern lets the receiver adjust as needed

5.5.2.1 Preambles, Slots, and Optimizatoins

The preambles of 802.11ag are at 6Mbps, the lowest 802.11ag data rate They are also

significantly shorter than the short 802.11b preambles: 40 microseconds for 802.11ag

Improvements in signal processing technology since 802.11b came along allowed the

designers to not need to provide as much synchronization time

Furthermore, the radios can stop receiving and start transmitting more quickly for 802.11ag than they were expected to be able to do for 802.11b This allows 802.11ag to use shorter slots than 802.11b, allowing for less wasted time For 802.11g, short slots are an option, and are determined based on the presence of 802.11b clients, to prevent 802.11g clients from using a different contention scheme 802.11a always assumes the faster slots

5.5.2.2 802.11b Protection

802.11a has no legacy clients to deal with However, 802.11g is in the 2.4GHz band, and has to avoid destroying 802.11b performance when the two devices are present together

This destruction would occur because 802.11b radios use carrier detection (see Section 5.4.3) to determine whether the channel is clear before transmitting, and that means that the 802.11b devices are looking for 802.11b transmissions 802.11g transmissions, however, look nothing like 802.11b, and so 802.11b radios would end up only seeing 802.11g as some sort of foreign interference

To prevent this from disrupting any 802.11b device’s traffic, 802.11g introduces the notion

of CTS-to-self protection Because 802.11b clients can only see 802.11b traffic, a way to

stop them from transmitting when an 802.11g transmission will start is for the 802.11g device to send an 802.11b (legacy) CTS message first This CTS message, sent not as a part

itself, sets the virtual carrier sense for all devices that can hear it The CTS frame has a Duration field—or the length of time to quiet the other stations—long enough to let it finish the 802.11g transmission that will follow

CTS-to-self protection is automatically turned on for any AP that has 802.11b legacy clients associated to it, or for access points who overhear neighboring access points that have 802.11b clients assigned to them This CTS-to-self message can be incredibly inefficient, and has the potential to disrupt voice mobility networks, as will be mentioned later in this chapter

802.11g data rates, even though they are identical to 802.11a data rates, may go by the

additional term ERP in product literature The term, short for Extended Rate PHY, is

typical of the language used in the 802.11 standard (worse abuses will come up in the next section), but is good to know for when it occasionally slips into product documents for users When you see ERP, think 802.11g, and when you see non-ERP, think 802.11b-only legacy devices Neither term is correct for 802.11a devices, which are just known as

802.11a

5.5.3 802.11n

54Mbps seemed like a lot at the time, but enterprise wireline networks operate at 100Mbps

or more To allow for even higher data rates, IEEE has embarked upon the 802.11n

standard 802.11n revolutionizes Wi-Fi by adding another radio breakthrough, as well as a long list of additional enhancements and optimizations

The most important addition 802.11n brings is the use of a technology called multiple-in,

multiple-out (MIMO) MIMO does something that seems counterintuitive—almost

magical—to those used to thinking about how two radios transmitting at once cause

collisions and destroy wireless networks MIMO transmits multiple signals at once, on the same channel, at the same time, and at the same power levels However, MIMO is not magic, just math, and is able to greatly increase the speed of the network

MIMO works by requiring each device to have multiple antennas These antennas are not

terribly far apart—a few inches at the most—but they need to be present MIMO then splits

the data across those antennas, sending out multiple spatial streams These streams go over

the air at the same time, and the receiver uses multiple antennas to pick up this transmission, applies some math to separate back out the streams, and then recombines the data

We will go through how this multiple simultaneous transmissions work in a moment

However, because MIMO has some general rules that the products using them need to

follow, let us start with those rules, and the state of the technology

802.11n is a very new standard, and, for 2008 and most of 2009, was not yet finished

However, every major Wi-Fi vendor was selling 802.11n-based products How can this

happen? As it turns out, major parts of 802.11n were complete enough for vendors to build products to from 802.11n Draft 2.0 The features that were complete enough, and were also interesting enough to encourage users to purchase products based on them, were written

down by the Wi-Fi Alliance into an industry certification program, also known as 802.11n Draft 2.0 This program specifies a rigorous set of interoperability tests, to ensure that

vendors that pass it have built their devices to the same specification (as in, they didn’t

make major errors)

The existence of the certification program should bring you comfort in knowing that

802.11n devices will work together For 802.11n Draft 2.0 products, the Wi-Fi Alliance,

which uses its role to ensure that devices interoperate, happened to do their work a bit

earlier than IEEE As it has turned out, however, the Draft 2.0 feature set and program are essentially the same as those for the final standard This makes sense, because the vendors that figure prominently in IEEE and the Wi-Fi Alliance had a tremendous incentive to

ensure that the final standard was only minimally changed from the draft

The main feature is the ability to provide 300Mbps data rates for clients This is achieved

by using two spatial streams and double-wide 40MHz channels (The standard defines up to four spatial streams, and some—but not most—devices can accommodate three streams, yet the overwhelming majority to date can use only two.) Furthermore, the WMM and

aggregation optimizations go a long way towards closing the gap between the data rate and the actual highest application throughput

Ignoring the MIMO aspect, 802.11n is similar to 802.11ag It too is OFDM, with exactly the same subcarrier setup as 802.11ag, except for increasing the number of data subcarriers

by four to 52 (and thus slightly increasing the frequency width) Additionally, the symbol is

usually still 4 microseconds; however, there is an optional mode known as short guard

interval that shaves 400 nanoseconds off the symbol’s length, getting a slightly higher kick

in throughput There are still eight modulation and coding rates that establish the data rates,

except that one BPSK mode is removed and another 64-QAM mode, with an even higher coding rate, is added These eight rates, however, are multiplied into a much higher number

of rates, based on the channel width, the number of spatial streams, and the guard

interval All together, the notion of data rates being signaled in the products by Mbps has

been abandoned, and replacing it is the concept of the modulation and coding scheme

(MCS), based on a small number representing what the actual parameters are in a table This is similar to how simple channel numbers represent much more complicated

frequencies

Table 5.14 contains the common two-stream set for 802.11n devices and encompasses 60 different data rate options, each with its own slightly different SNR requirement, channel width, or robustness

There are additional features that 802.11n has as options, which are not commonly

implemented but could be of great benefit for voice mobility Among these features are

space-time block codes (STBC), transmit beamforming, and extended power save

capabilities However, because these features have not yet become commonplace, their use

is rather limited

5.5.3.1 MIMO

As mentioned before, MIMO lets the devices transmit at, three times, or four times by using

as many spatial streams The general rule for MIMO, as a theory, is that the number of spatial streams usable is the lesser of the number of antennas that can be used

simultaneously on the transmitter and the receiver In theory, a 100-antenna transmitter and

a 50-antenna receiver could allow for a 50 (< 100) spatial stream radio

In practice, there are limits 802.11n defines only four spatial streams maximum The Wi-Fi Alliance Draft 2.0 certification tested for only two spatial streams, and most devices today remain only capable of two streams The reasons are rather simple More antennas require more room for antennas, and it is hard to find room to place them Also, more antennas

Table 5.13: 802.11n Common features

Two spatial streams Doubles performance over non-MIMO

Aggregation Allows for very high efficiency WMM Quality-of-service is mandatory (see Section 6.0.1) WPA2 High-grade encryption support is mandatory (see Section 5.6.1.1 ) 40MHz wide channels Doubles performance over 20MHz

MRC/Receive Beamforming Longer range in some cases

require more radio chains—an 802.11n radio is actually made of multiple copies of the

parts needed to make an 802.11ag radio work—and those are expensive and draw power Finally, practical considerations prevent higher numbers of simultaneous streams from

working well

Now, for the description of why MIMO works For this example, assume that the sender and receiver each have three antennas, but follow the industry norm of using only two

spatial streams The sender divides its signal into two spatial streams, then spreads its two streams across the three antennas Those two streams from three signals bounce around the environment, and end up as three different signals at the receiver, one for each of the

receiver’s antennas Each of those three signals is some different combination of the three signals from the sender, and thus is some different combination of the two spatial streams This is where the math sets in The different combinations are usually very different If you

Table 5.14: The 802.11n data rates

MCS Modulation Bits per

Symbol per Subcarrier

Coding Rate

Spatial Streams

Speed 20MHz

Long GI

20MHz Short GI

40MHz Long GI

40MHz Short GI

write out the two spatial streams, the effects of the sender’s spreading and the channel’s bouncing, and the receiver’s receiving, you can produce a matrix equation, from linear algebra Because the combinations are different for each antenna—linearly independent, in fact—the receiver can undo the effects of the channel using linear algebra and produce the original streams For further explanation, see the appendix at the end of this chapter

Basically, the MIMO receiver uses the preamble of the frame that is sent to discover what the effects of the channel are on the streams, and then uses that to undo those effects

The effect of having multiple antennas when only one spatial stream is used (and a main

effect as a part of MIMO) is for beamforming There are two parts to beamforming: receive

beamforming, and transmit beamforming The term beamforming arose from RADAR, where stationary equipment used electronics and a large number of antennas to set up interference patterns just right to concentrate a signal in a direction or to a point, as if the antennas were mounted on a swivel and were pointed, although they are not

Receive beamforming isn’t necessarily thought of the same way as transmit beamforming, but is the major reason why 802.11n has higher range than 802.11abg There are a number

of techniques for doing what could be called receive beamforming One term used

surprisingly often by vendors in describing their products in data sheets is maximum ratio

combining (MRC) To understand it at a high level, the receiver is twiddling with how it combines the signals received on each of its antennas to maximize the power of the final signal it received Because of the way interference patterns and combinations work, it turns

out that the twiddling it does is a unique pattern (H, the channel matrix, if you read the

math briefing) based on the client’s location But because receive beamforming learns that pattern when it sees the preamble (same as MIMO), this is not a problem, and the end result

is an apparent amplification of the signal, thus increasing range for reception 802.11n clients with MIMO have longer range on legacy access points than legacy clients do, for that reason Therefore, if you need to extend the range of a couple of clients, your best bet

is to upgrade the clients to 802.11n (Upgrading the access points without upgrading the clients may not increase range at all in many cases.)

Transmit beamforming is also possible 802.11n defines two types of beamforming, known

as explicit and implicit beamforming Explicit beamforming uses the cooperation of the

receiver to determine the best way of combining signals across the transmitter’s antennas for forming the signal to the receiver This cooperation requires features on the receivers that are not commonly implemented Implicit beamforming simply requires the transmitter, assuming that the channel it sees from the receiver when that device transmits is the same

as what the receiver sees from the transmitter By assuming this reciprocity, the transmitter does not need to involve the receiver However, it is forced into its guess, which may not be correct, and requires that the transmitter always keep tabs on the receiver’s channel

conditions by either sending an RTS to it before every packet, thus eliciting a responding

CTS that will help uncover the channel conditions, or by winging it and hoping the receiver doesn’t move much For this reason, some vendors are limiting their transmit beamforming support to only legacy, non-802.11n clients

Transmit beamforming is an interesting concept for voice mobility networks based on the microcell approach of reducing transmit power levels to begin with

5.5.3.2 Legacy Support

Legacy, in the context of 802.11n, means 802.11ag as well as 802.11b To avoid the same problem of 802.11g possibly being interfered with by 802.11b here, with 802.11ag stepping

on 802.11n, 802.11n uses a different form of protection Instead of the wasteful CTS

frames, and the protocol necessary to decide whether a CTS frame is needed, 802.11n uses

a special preamble The preamble first starts off as an 802.11ag preamble, not only at

6Mbps but signaling a 6Mbps data rate for the following data, and including a length that will encompass the entire 802.11n frame However, as soon as the preamble is over, the

802.11n radio stops transmitting in 802.11ag, and switches over to 802.11n, where it

continues with more preamble fields, including the real data rate 802.11n devices see the entire frame 802.11ag devices see the preamble only, but defer just as if the frame were all 802.11ag but out of range Thus, the technologies do not interfere

For 802.11b clients, 802.11n still uses an 802.11b CTS frame

5.5.3.3 Aggregation

802.11 frames have to have the preamble, but there was no particular reason that one

preamble couldn’t cover multiple frames 802.11n introduces this concept with frame

aggregation, or A-MPDUs

A-MPDUs are a special type of jumbo frame that contains multiple 802.11 data frames sent from and to the same wireless device Up to four milliseconds or roughly 64,000 bytes can be packed into one of these aggregates In almost every sense, an A-MPDU can just be thought

of as a concatenation of data: every byte is sent at the same data rate that the preamble calls out, and the A-MPDU is retransmitted in full if the expected acknowledgment does not come back immediately from the receiver However, unlike a simply larger data frame, if some of the data frames within the aggregate are received and others are not, the receiver can indicate this by using a special Block Acknowledgment This block acknowledgment specifies

precisely which of the senders’ frames were received and which had errors in them For those frames that were not received, the sender can choose to add those to later aggregates, thus not wasting time resending frames that already arrived

The main benefit of aggregation is to reduce the overhead of the preamble and backoff for 802.11n

5.5.3.4 Double-Wide Channels (40MHz)

Double-wide channels work by bonding together two adjacent 20MHz channels into one larger 40MHz channel This 40MHz channel acts just like a wider 20MHz channel, but offering data rates slightly higher than twice that of the 20MHz channel (The slight

increase over twice comes from using up the gap between channels as usable bandwidth.) There is some inconsistency in the naming of this feature in the industry Some devices call

it “double-wide,” others call it “channel bonding,” and others call it just “40MHz.” We’ll use “double-wide” and “40MHz” for this description

Double-wide channels are named by the 20MHz channels that they occupy There are a few nomenclatures in active use, but all are just slight variations of the basic concept There is a

primary channel and an extended channel in a 40MHz channel name 40MHz channels can

operate with 20MHz devices, but only on one 20MHz half of the double-wide channel This one half is the primary channel The other half is used only for the rest of the 40MHz transmissions, and is thus the extended channel For example, a 40MHz channel selection that has a primary channel of 36 and an extended channel of 40 can be written as 36+1, (36, +1), 36U, 36H, and so on The same 40MHz channel, but with the primary being on the other half, can be written as 40−1, (40, −1), or 40L

It is useful to keep in mind, however, that a 40MHz frame only sees one 40MHz channel, and is not split itself into two separate 20MHz frames for each half, unlike what the

terminology suggests

There are a few considerations for double-wide channels The first is that both 20MHz channels that are within the 40MHz should be empty This is especially true for the

extension channel, which should not have any other access points using that channel, unless

a layered or virtualized over-the-air architecture is used that can support that Certainly, no access point should be deployed such that its primary (or only) channel is 40 if 36+1 is being used anywhere around that access point

The second consideration is that some devices cannot support 40MHz, and so will use only the primary half of the channel This is obviously true for legacy (non-802.11n) devices, but

is also true for some 802.11n devices Some 802.11n devices are not designed to take advantage of double-wide channels, and a few early 802.11n devices had support for

double-wide channels in the 2.4GHz band turned off, though newer devices from those manufacturers have solved that problem

The third consideration is that there are only half as many 40MHz channels as there are 20MHz channels, and the direction a 40MHz channel may extend is limited This is to avoid the overlap mentioned earlier, but ends up leading to channel waste unless channels

are carefully planned For example, channels 36 and 40 can be bonded, but channels 40 and

44 can never be, by rule If channel 36 is being used by an existing 20MHz network, then channel 40 can never be used with a 40MHz channel in that case

The 2.4GHz band does not have the limitation of which direction a 40MHz channel can

extend, but because the bandwidth is so limited, only one 40MHz channel can be created successfully The network or administrator can usually choose which Keep in mind that

dynamic architecture vendors recommend against using double-wide channels in the

2.4GHz band, because using it on that architecture eliminates a channel that is necessary for the alternating channel plan to work with Layered and virtualized architectures do not have this limitation

5.5.3.5 Coming Down the Road

802.11 is still growing As of this writing, there is a push to expand the technology to

increasing bands—such as the spectrum opened up by the end of analog television

broadcasts in 2009 These technologies will not show up for a few years, however, and

should not affect voice mobility networks that are being considered for deployment in the near future

5.6 Security for 802.11

Security is a broad subject, and there is an entire chapter dedicated to the unique challenges with security for voice mobility later But any component of voice mobility over Wi-Fi will require some use of 802.11’s built-in encryption Keep in mind that securing the wireless link is not only critical, but may be the only encryption used to prevent eavesdroppers from listening in on sensitive voice calls for many networks

802.11 security has both a rich and somewhat checkered past Because of the initial

application of 802.11 to the home, and some critical mistakes by some of the original

designers, 802.11 started out with inadequate protection for traffic But thankfully, all Wi-Fi-certified devices today are required to support strong security mechanisms

Nevertheless, administrators today do still need to keep in mind some of the older, less

secure technologies—often because the mobile handset might not correctly support the

latest security, and it may fall to you to figure out how to make an old handset work without compromising the security of the rest of the network

A secure wireless network provides at least the following (borrowed from Chapter 8):

• Confidentiality: No wireless device other than the intended recipient can decrypt the

message