运用均衡技术的sc-fdma.txt

这段程序清楚了显示了运用均衡技术的SC-FDMA的性能优势

function [SER_ifdma SER_lfdma] = scfdma(SP) 
 
numSymbols = SP.FFTsize; 
Q = numSymbols/SP.inputBlockSize; 
H_channel = fft(SP.channel,SP.FFTsize); 
 
for n = 1:length(SP.SNR), 
    tic; 
    errCount_ifdma = 0; 
    errCount_lfdma = 0; 
 
    for k = 1:SP.numRun, 
        tmp = round(rand(2,SP.inputBlockSize)); 
        tmp = tmp*2 - 1; 
        inputSymbols = (tmp(1,:) + i*tmp(2,:))/sqrt(2); 
     
        inputSymbols_freq = fft(inputSymbols); 
        inputSamples_ifdma = zeros(1,numSymbols); 
        inputSamples_lfdma = zeros(1,numSymbols); 
 
        inputSamples_ifdma(1+SP.subband:Q:numSymbols) = inputSymbols_freq; 
        inputSamples_lfdma([1:SP.inputBlockSize]+SP.inputBlockSize*SP.subband) = inputSymbols_freq; 
        inputSamples_ifdma = ifft(inputSamples_ifdma); 
        inputSamples_lfdma = ifft(inputSamples_lfdma); 
 
        TxSamples_ifdma = [inputSamples_ifdma(numSymbols-SP.CPsize+1:numSymbols) inputSamples_ifdma]; 
        TxSamples_lfdma = [inputSamples_lfdma(numSymbols-SP.CPsize+1:numSymbols) inputSamples_lfdma]; 
         
        RxSamples_ifdma = filter(SP.channel, 1, TxSamples_ifdma); % Multipath Channel 
        RxSamples_lfdma = filter(SP.channel, 1, TxSamples_lfdma); % Multipath Channel 
     
        tmp = randn(2, numSymbols+SP.CPsize); 
        complexNoise = (tmp(1,:) + i*tmp(2,:))/sqrt(2); 
        noisePower = 10^(-SP.SNR(n)/10); 
        RxSamples_ifdma = RxSamples_ifdma + sqrt(noisePower/Q)*complexNoise; 
        RxSamples_lfdma = RxSamples_lfdma + sqrt(noisePower/Q)*complexNoise; 
         
        RxSamples_ifdma = RxSamples_ifdma(SP.CPsize+1:numSymbols+SP.CPsize); 
        RxSamples_lfdma = RxSamples_lfdma(SP.CPsize+1:numSymbols+SP.CPsize); 
        Y_ifdma = fft(RxSamples_ifdma, SP.FFTsize); 
        Y_lfdma = fft(RxSamples_lfdma, SP.FFTsize); 
         
        Y_ifdma = Y_ifdma(1+SP.subband:Q:numSymbols); 
        Y_lfdma = Y_lfdma([1:SP.inputBlockSize]+SP.inputBlockSize*SP.subband); 
         
        H_eff = H_channel(1+SP.subband:Q:numSymbols); 
        if SP.equalizerType == 'ZERO' 
            Y_ifdma = Y_ifdma./H_eff; 
        elseif SP.equalizerType == 'MMSE' 
            C = conj(H_eff)./(conj(H_eff).*H_eff + 10^(-SP.SNR(n)/10)); 
            Y_ifdma = Y_ifdma.*C; 
        end 
         
        H_eff = H_channel([1:SP.inputBlockSize]+SP.inputBlockSize*SP.subband); 
        if SP.equalizerType == 'ZERO' 
            Y_lfdma = Y_lfdma./H_eff; 
        elseif SP.equalizerType == 'MMSE' 
            C = conj(H_eff)./(conj(H_eff).*H_eff + 10^(-SP.SNR(n)/10)); 
            Y_lfdma = Y_lfdma.*C; 
        end 
 
        EstSymbols_ifdma = ifft(Y_ifdma); 
        EstSymbols_lfdma = ifft(Y_lfdma); 
         
        EstSymbols_ifdma = sign(real(EstSymbols_ifdma)) + i*sign(imag(EstSymbols_ifdma)); 
        EstSymbols_ifdma = EstSymbols_ifdma/sqrt(2); 
        EstSymbols_lfdma = sign(real(EstSymbols_lfdma)) + i*sign(imag(EstSymbols_lfdma)); 
        EstSymbols_lfdma = EstSymbols_lfdma/sqrt(2); 
 
        I_ifdma = find((inputSymbols-EstSymbols_ifdma) == 0); 
        errCount_ifdma = errCount_ifdma + (SP.inputBlockSize-length(I_ifdma)); 
        I_lfdma = find((inputSymbols-EstSymbols_lfdma) == 0); 
        errCount_lfdma = errCount_lfdma + (SP.inputBlockSize-length(I_lfdma)); 
    end 
    SER_ifdma(n,:) = errCount_ifdma / (SP.inputBlockSize*SP.numRun); 
    SER_lfdma(n,:) = errCount_lfdma / (SP.inputBlockSize*SP.numRun); 
    [SP.SNR(n) SER_ifdma(n,:) SER_lfdma(n,:)] 
    toc 
end