As mentioned above, the basic problem in wireless networks is the fact that wireless links are relatively unreliable. Thus, a number of schemes that work on the physical layer and try to present a relatively ‘clean’ medium to higher layers of the network have been considered.
Some of them have found use in commercial systems. In our previous discussion, we visited one such technique, spread spectrum, which, as mentioned, effectively combats fading as it splits the transmitted signal over a large bandwidth. Due to the fact that different spectral components inside this large bandwidth fade independently, fading will affect only a part of the transmission. Thus, SS achieves resistance to fading. In this section, we describe some other techniques: diversity, coding, equalization and power control.
2.7.1 Diversity Techniques
In an effort to combat the phenomenon of fading in wireless channels, a family of techniques, known as diversity techniques, is used. Many types of diversity techniques exist, such as time, frequency, antenna (also known as space) and polarization diversity. The principle of diver- sity systems is to send copies of the same information signal through several different channels. Performance enhancement is achieved due to the fact that these channels fade
Wireless Communications Principles and Fundamentals 67
independently, thus, fading will affect only a part of the transmission. In this section we describe the fundamentals of the most commonly used type of diversity, antenna diversity and its enhancements, smart antennas. Other types of diversity are considered in chapters that present systems that use them.
2.7.1.1 Antenna Diversity
Antenna diversity, also known as space diversity, is commonly used for performance enhancement in wireless systems. It is essentially a method that calls for a set of array elements (also referred to as branches), mostly two, spaced sufficiently apart from each other, with the spacing usually in the order of the wavelength of the used channel. This is due to the fact that multipath fading is considered independent at distances in the order of the channel’s wavelength.
Antenna diversity can effectively combat multipath fading in NLOS situations. Never- theless, the performance gains are lower in LOS cases. Although applicable in both BSs and mobile stations, antenna diversity poses significant challenges for implementation in the mobile stations, due to limitations relating to power consumption and size of the mobile station antenna. It can be used either for transmission (transmit diversity) or reception (receive diversity) of signals. In both cases, the aim is to increase the quality of the signal at the receiver. In receive diversity, the branches of the antenna system pick up a number of differently fading signals and combine them in order to reconstruct the original transmission at the highest possible quality. When applied at the BS of a cellular system, receive antenna diversity obviously enhances the performance of the uplink. Thus, it has found use in the uplink of a number of wireless systems, such as GSM and IS-136. Transmit diversity, on the other hand, calls for transmission of replicas of the signal by each branch of the antenna and when applied to a the BS of a cellular system will favor the reception quality at the downlink.
However, it has seen limited support in commercial systems.
Figure 2.35 sketches the way a two-branch receive diversity system can combat interfer- ence. A number of algorithms that exploit the signals from the two antenna elements in order to reconstruct the original transmission at the highest possible quality exist. For example, antenna diversity can be used to either select the strongest signal picked up by one of the antenna elements or combine the signals from the two elements in order to reconstruct the original transmission. However, a description of such algorithms is out of the scope of this chapter.
2.7.1.2 Multiantenna Transmission/Reception: Smart Antennas
The term smart (or adaptive) antennas [18] is used to describe antennas that are not fixed but rather change in order to adapt to the conditions of the wireless channel. Smart antennas are considered an enhanced method of antenna diversity. Although applicable to almost all kinds of wireless networks, smart antennas are most commonly considered for use by the BSs of cellular systems. The idea of smart antennas has been around for some years, however, due to the fact that it demands computational power, consideration on its use in commercial systems has started only recently.
Smart antennas combat the deficiencies of conventional omnidirectional antennas. Consid- ering the case of a cellular system, omnidirectional antennas can be regarded as a waste of
power, due to the fact that they radiate power in all directions while the user being serviced by the antenna is only in a certain direction. Smart antennas surpass this inefficiency since they can (a) focus to the radio transmission of the receiver and (b) focus their own transmission towards the receiver, as seen in Figure 2.36 for a cellular system. This technique is also known as beamforming. According to the principle on which beamforming is based, smart antennas can be categorized into [18]:
† Switched lobe, or switched beam.This category of smart antennas is the most simple. It consists of a number of static directive antenna elements and a basic switching function.
Wireless Communications Principles and Fundamentals 69
Figure 2.35 A two-branch diversity system
Figure 2.36 Use of smart antennas in cellular systems
For communication with a receiver, this switching function selects the element that maxi- mizes performance, usually measured by received power. This structure is the most simple but also has the lowest performance gains compared to conventional antennas.
† Dynamically phased array.This category utilizes information regarding the direction of arrival of the transmitter’s signal. By monitoring this value, antenna elements can be used to track the user as it moves. This category, which can be seen as an enhancement of the switched-beam concept also maximizes performance in terms of received power.
† Adaptive array.This category utilizes the direction of arrival value of users nearby the entity, which transmits to the antenna. By using this value the radiation pattern can be adjusted to cancel interference from these users. Furthermore, adaptive arrays can be used to combine the different echoes of a user’s transmission and reconstruct the original signal.
Adaptive arrays maximize performance by maximizing the received Signal to Interference noise Ratio (SIR).
The above methods describe tracking of the reception signal by the BS which implements antenna diversity. As far as transmission to a mobile is concerned, the BS may utilize the value of direction of arrival of the mobile node transmission at the BS in order to focus its transmission on the mobile receiver. Figure 2.37 shows possible structures of array elements that form smart antennas. The first two structures are used for beamforming on a horizontal plane, which is enough for large cells, typically found in rural areas. For densely populated cells, the third and fourth structures can be used for two-dimensional beamforming. For transmission, the radiation pattern is produced by splitting the signal to be transmitted into a number of other signals. These signals are directed to different array elements and their amplitudes and relative phases of these signals are adjusted in order to maximize perfor- mance. The opposite happens in reception: The signals from each element produce a combined signal that enters the decoding circuits of the receiver.
Although realizable today, smart antennas are not likely to start being applicable both in Figure 2.37 Possible placements of array elements in smart antennas
the uplink and downlink of mobile systems. Rather it is envisioned that smart antennas will first be introduced in base stations, thus benefiting uplink (mobile to BS) transmissions. The use of directive antennas to the BSs will occur later on. Finally, when smart antennas find full application in mobile systems, these will be able to accommodate traffic from many users in the same frequency at the same time and separate users by spatial information. This will create a new form of multiple access, Space Division Multiple Access (SDMA), where users will be separated based on the angle of their transmission to the base station.
Due to their capability for directive transmission, smart antennas have the obvious advan- tage of interference reduction for users nearby the receiver. Furthermore, the capacity of a system is increased since it is possible that the same spectrum can be used at the same time by more than one user. This can be done by exploiting information regarding their position and using smart antennas to direct BS transmissions to these users. This idea gave rise to the concept of geolocation applications which will be used in next generation wireless networks which will be able to extract spatial information of users more precisely than current wireless networks. Geolocation applications are revisited in Chapter 5. Smart antennas are also likely to have an increased range compared to conventional ones and offer an increased level of security, since eavesdropping of a transmission now requires the eavesdropper to be present in the imaginable line formed between the transmitter and the receiver.
However, the use of smart antennas entails some problems too. The first is the increased implementation complexity and the cost of the method. This is incurred by the need for real- time tracking of the receiver position and real-time updating of the corresponding transmis- sion. For a BS this is difficult since BSs in cellular systems are likely to serve many users at the same time. Furthermore, the fact that SDMA separates users based on their angle to the receiver means that BSs will have to switch users to another SDMA channel when angular collisions occur. In densely populated areas, many such collisions are likely to take place thus requiring increased computational efficiency (which means cost) from the BS. However, the advancements of computer technology have driven down costs and systems having the necessary processing power for this task are available. Nevertheless, the cost of a smart antenna system will be larger than a system with a conventional antenna.
2.7.2 Coding
In all kinds of networks, there is a certain possibility that reception of a bit stream is altered by errors. Coding techniques aim to provide resistance to such errors by adding redundant bits to the transmitted bit stream so that the receiver can either detect and ask for a retransmission, or correct the faulty reception. Thus, we have error detection and error correction coding schemes, respectively. The process of adding this redundant information is known as channel coding, or Forward Error Correction (FEC). By recalling the fact that the BER experienced over a wireless channel can be as high as 1023whereas typical BERs of wired channels are around 10210, one can easily realize the usefulness of such techniques in wireless systems. A vast number of coding techniques exist in the scientific literature; thus, in this section we present the basic techniques used for FEC.
2.7.2.1 Parity Check
The simplest technique, which is known to almost anyone dealing with computer technology,
Wireless Communications Principles and Fundamentals 71
is the parity bit technique that can detect single-bit errors. According to this technique, the transmitter and the receiver agree whether the number of binary 1s that will be contained in the messages they exchange will be odd or even. Thus, the parity schemes are known as odd parity and even parity, respectively. For the sake of presentation assume that the transmitted message is 10010 and that the agreement is on odd number of binary 1s. After the agreement for odd parity is made, the transmitter adds either a binary 1 or a binary 0 to the end of the original message, so that the number of binary 1s in the resulting bit stream is odd and sends the message. Thus, the actual transmitted message is now 100101. If the bit stream arrives intact at the receiver (100101), then the number of binary 1s will be odd, whereas if a single bit error occurs (e.g. 101101), the receiver will detect that the number of binary 1s is even.
Although very simple to implement, the parity scheme has the disadvantages of (a) not being able to detect a multiple of two bit errors in the same message (e.g. in the above case, it would see the reception of 001101 (two bit errors) as correct) and (b) being able only to detect and not to correct a faulty reception. The problem is that the parity scheme has a Hamming distance of 2. The Hamming distance of a set of binary streams defines the least number of bit inversions that, when applied, can lead from a stream of the set to another stream of the set.
By increasing this distance, the scheme can be made more robust. Returning to an example, consider that the receiver and the transmitter agree to exchange only the messages 00000 and 00111, which have a distance of 3. Thus, for the reception of 00000 as 00011 the receiver can detect the double-bit error since the received message is not valid. Furthermore, for the reception of 00000 as 00001, the receiver can correct the single-bit error (as 00001 is closer to 00000 than to 00111) and thus recover the transmitted message.
2.7.2.2 Hamming Code
The above example does not consider adding extra bits to the original message for coding purposes. However, the addition of extra bits for coding should produce a set of valid bit streams that has the maximum possible distance. It holds that if coding leads to a set of valid bit streams of distanced, then it can either detectDerrors or correct Terrors, where d$ D11 andd[ẵ2T11…2T12, respectively.
The Hamming code is a very popular error correcting code with distance 3; thusT ẳ1. In the Hamming code, the number of the bits of the coded message,n, the number of the bits of the message to be coded, k, and the number of coding bits, r, are related according to the following equations:
nẳ2r21; kẳ2r212r ð2:20ị The Hamming code works by placing check bits in those positions of the coded bit stream that are powers of two (1,2,4,8,…), whereas the remaining positions are filled with data bits.
These check bits contribute to the calculation of parity of some, but not all, of the data bits. In order to determine the check bits that are concerned with the integrity of a given data bit, the position of the data bit is rewritten as a sum of numbers that are powers of two and these numbers indicate the coding bits related to the parity of this data bit. This is also the way decoding is done: For every code bit in positionp, the receiver checks the parity of the set of bits related to this coding bit. If an error is found, the number of the coding bit is added to a counter p,initialized to 0. At the end of the decoding procedure, if p–0, it contains the position of the incorrect bit.
A slight modification of the Hamming code permits it to correct not only one bit error but also a burst error (like the ones appearing in wireless links) of length sin bits. For coded messages of lengthjthis is done by:
† coding the messages to be transmitted according to the Hamming code;
† gathering at leastssuch messages into the rows of ans£jmatrixA;
† calculation of thej£smatrixBhaving thejcolumns of matrix A as its rows (retrograde matrix);
† transmission of thejmessages of lengthsas those appear in the lines of matrixB.
At the receiver, the inverse function is performed and an s £ j matrix is obtained that contains the results of transmission of the s j-bit messages. Assuming that the medium suffered a long error burst up to sbit times and consequently up tosbit errors arose, this interleaving scheme leads to the reception of up tosmessages that suffer a bit error at the same position. Thus, this scheme leads up tosreceived messages with a single bit error (thus correctable) at the receiver and not to some totally destroyed bit streams.
2.7.2.3 Cyclic Redundancy Check (CRC)
CRC is a widely used error detecting code. For the coding of anm-bit messageM withn coding bits, the transmitter and receiver agree on a common (n11)-bit streamP,withn,m.
CRC codes this message by appending an n-bit sequence F, known as the Frame Check Sequence (FCS) to the end of them-bit message. By shifting M n bits left and modulo-2 dividing the resultEwithP,the FCSFis defined as the remainder of this modulo-2 division.
It can be proven that then1mmessage produced by appendingFto the end ofEis exactly divisible by the predetermined numberP. After reception of the coded messageT ẳE1F, the receiver modulo-2 divides it with P.If there is a nonzero remainder, the message was received with an error, otherwise it is assumed correct. There exists a finite possibility that an error has occurred and the division is still exact, however, this is an unlikely event that can be handled by higher layer protocols.
2.7.2.4 Convolutional Coding
Convolutional codes have found use in several wireless systems, such as the IS-95 cellular standard covered in Chapter 4. Convolutional codes are usually referred to based on the code’s raterẳk/n and constraint lengthK. The code rate of a convolutional code shows the ratio of the number of bitsnthat are output of the convolutional encoder to the number of bitskthat were fed into the encoder. Convolutional coding of a bit stream will produce a larger bit stream. Thus, if the resulting bit stream is to be transmitted to the receiver over the same time period as the source stream, a bandwidth increase is necessary. The constraint length parameter,K, denotes the ‘length’ of the convolutional encoder; stating how manyk- bit stages are available to feed the structure produces the output symbols. In general, the larger the value ofK, the less the probability of a bit suffering an error. Specifically, the value ofKand this probability are exponentially related.
In order to gain an insight into the operation of a convolutional coder, consider the example of Figure 2.38 which shows a convolutional coder withKẳ4 andrẳ1=3. Assume that the bit stream 1001 is to be coded. The first bit of the stream is fed into the first stage of the coder
Wireless Communications Principles and Fundamentals 73
and three bits are produced as a result of coding this bit. This procedure continues until the last bit of the stream reaches the last stage of the coder. Using this structure, the bit stream 1011 would be coded as 111 010 100 110 001 000 011 000. One can notice that the number of bits in the output of the decoder is not 12, as one would expect with what we defined the rate to be, but rather it is higher due to the fact that the operation ends when the last bit of the input sequence exits the encoder. Generally, for anrcode and ak-bit input stream, the number of output bits isðk1Kị=r. Since, in practice, the number of stagesKis a very small number compared to the input stream sizek,ðk1Kị=rứk=r, which is consistent with the definition thatrẳk=n.The operation of a convolutional coder depends on the selection for a value ofK, the number of XOR adders and the way these are connected to stage outputsui.
Two categories of convolutional decoding algorithms exist. The first is sequential decod- ing. It has the advantage of performing very well with convolutional codes of largeK, but it has a variable decoding time. The second category, Viterbi decoding, removes this disadvan- tage by having a fixed decoding time; however, it has an increased computational complexity which is exponentially related to the value ofK.The operation of convolutional decoding is out of the scope of this chapter and the interested reader can seek information in the scientific journals.
2.7.3 Equalization
Equalization techniques have found wide use in wireless systems for combating the effect of ISI. The general idea of equalization is to predict the ISI that will be encountered by a transmission and accordingly modify the signal to be transmitted so that the signal reaching the receiver will represent the information the transmitter wants to send. Two categories of equalization techniques exist: linear and nonlinear. Linear equalization techniques are not preferred for wireless communication systems, whereas nonlinear techniques, such as deci- sion feedback equalization (DFE), data directed estimation (DDE) and maximum likelihood sequence estimation (MLSE) are commonly used for wireless systems. Of the nonlinear techniques, the choice for use in wireless systems is usually DFE since MLSE requires an increased computational complexity and knowledge of the channel characteristics.
DFE employs a set of coefficients that are used for modeling the behavior of the wireless Figure 2.38 A convolutional coder withrẳ1=3 andKẳ4