‘Gh. 6 + Equalization, Dversiy, and Channel Coding 357 Euclidean distance) between the coded signals. In the receiver, the signals are decoded by a soft decision maximum likelihood sequence decoder. Coding gains as large as 6 dB can be obtained without any bandwidth expansion or reduction in the effective information rate. Example 66 Equatizetion ‘The IS-54 USDC standard specifies the use of decision feedback equalizers (DFE). Diversity 1. The US AMPS system makes use of spatial selection diversity. 2. The PACS standard specifies the use of antenna diversity for base stations and portable units. Channel Coding 1. The IS-95 standard as proposed makes use of @ rate 1/3, constraint length L=9 convolutional code with block interleaving, The interleaver used is a 32°18 block interleaver. 2, The AMPS system makes use of a (40,28) BCH code for the forward control channel end a (48,30) BCH code for the reverse control channel Equalization, diversity, and channel coding, as discussed, have the same goal of improving the reliability and quality of the communication service. Each technique has it’s own advantages and disadvantages. The trade-offs to be con- sidered are those of complexity/power/cost versus system performance. Esch technique is capable of improving system performance significantly. 6.18 Problems 6.1 Use identical notation described in Section 63, except now let N dy = waa nus and verify that the MSE is identical for the multiple input linear filter shown in Figure P6.1 (this structure is used for maximal ratio com. bining diversity, RAKE receivers, and adaptive antennas) Figure P6.1: Multiple input adaptive linear combiner. 6.2 Consider the two-tap adaptive equalizer shown in Figure P6.2 (a) Find on expression for MSE in terms of wo, 1, and N. (b) HfN>2, find the minimum MSE.Gh 6 « Equalization, Diversity. and Channel Cocing (©) If wo=0, wy=-2 and N=4 samplesieyel td) For parameters in te}, what ss the MS ) | Figure P6.2: A two-tap adaptive linear equalizer, 6.3 For the equalizer in Figure P6.2, what weight values will produce a rms of e4=2? Assume N'=5, and express your answer in terms of wy, and w 6.4 Ifa digital signal processing chip can perform 1 million multiplications per sec- ond, determine the time required between each iteration for the following adaptive equalizer algorithms (a) LMS () Kalman RLS (c) Square root RLS DFE (d) Gradient lattice DFE 6.5. Suppose a quick rule of thumb is that an RLS algorithm requires 50 iterations to converge, whereas the LMS algorithm requires 1000 iterations. For a DSP chip that performs 25 million multiplications per second determine the maxi- mum symbol rate sind maximum time interval before retraining if a 5 tap equalizer requires 10% transmission overhead, and the following Doppler spreads are found in a 1900 MHz channel. In eases where 25 million multiphi- cations per second is not fast enough, determine the minimum DSP multiplica tions per second required for each equalizer implementation and compare maximum symtol rates for a fixed numher of multiplications per second. (a) 100 Hz (b) 1000 Hz fe) 10,000 Hz (Hint: Consider the eoherence time and its impact un equalizer training) 6.6. Use computer simulation to implement a 2 stage (3 tap» LMS equalizer based ‘on the circuit shown in Figure 62. Assume each delay element offers 10 micro seconds of delay, and the transmitted baseband signal 2 is a rectangular binary pulse train of alternating ('s and 0's, where each pulse has a duration of 10 microseconds. Assume that xt» passes through a time dispersive channel before being applied to the equalizer. Ifthe channel is a stationary 2-ray chan nel, with equal amplitude impulses at ¢=0 and ‘=15 microseconds, use equa- tions (6.35) — 16.37; to verify that the original x1) can be recreated by the equalizer. Use pulses ssith raised cosine rolloff witha = | (a) Provide graphical data illustrating the equalizer converges ‘by Plot the MMSE as a function of the number of steration: (co) How many iterations sre required Lo obtain convergence?Ch. 8 + Equalization, Diversity, and Channel Coding 389 (d) What happens if the second ray is placed at =25 microseconds? (e) What happens if the second ray is set equal to zero? (This effect is known as equalizer noise, and is the result of using an equalizer when one is not required by the channel 6.7 Consider a single branch Rayleigh fading signal has a 20% chance of being 6 4B below some mean SNR threshold (a) Determine the mean of the Rayleigh fading signal as referenced to the threshold. (b) Find the 'kelihood that a two branch selection diversity receiver will he 6 4B below the mean SNR threshold (c) Find the likelihood that a three branch selection diversity receiver will be 6 dB below the mean SNR threshold (d)_ Find the likelihood that a four branch selection diversity receiver will be 6 4B below the mean SNR threshold. (e) Based on your answers above, is there a law of diminishing returns when diversity is used? 68 Using computer simulation, reconstruct Figure 6.11 which illustrates the improvements offered by selection diversity. 69 Prove that the maximal ratio combining results in equations (6.66) — (6.69) are accurate, and plot the probability distributions of SNR = yy as a function of 7/0 for 1,2, 3,and 4-branch diversity. 6.10 Compare 7/1 (selection diversity) with 74,’ T (maximal ratio combining) for to 6 branches. Specifically, compare how the average SNR increases for each diversity scheme as « new branch is added, Does this make sense? What is the average SNR improvement offered by 6-branch maximal ratio combining as compared to 6-branch selection diversity? If y/T = 0.01 , determine the proba- bility that the received signal will be below this thresholé for maximal ratio combining and selection diversity (assume 6 branches are used). How does this compare with a single Rayleigh fading channel with the same threshold? 6.11 Extending the diversity concepts in this chapter and using the flat fading BER analysis of Chapter 5, it is possible to determine the BER for a wide range of modulation techniques when selection diversity is applied. Define ¥, as the required B,/.No to achieve a specific BER =y in a flat Rayleigh fading channel, and let y denote the random SNR due to fading. Furthermore let Pry) denote a function that deseribes the BER for a particular modulation when the SNR = 7. It follows: y = PriPepy>x| = PrlyePixy = te 2 (a) Find an expression that solves 7,, in terms of PYx) and (b) When M uncorrelated fading branches are used for diversity selection, write a new expression for x. (cl Determine the required average E,/No for BPSK in order to sustain a 10." BER ina Rayleigh fading channel (a) When 4 branch diversity is used, determine the required average Ey/No for BPSK in order to sustain a 10° BER in a Rayleigh fading channel