Sparse Bayesian Learning -压缩感知

function [ decodSyms ] = sblDecode( rcvSig,nullToneList,symDemod,nEmItr ) % Performs Sparse Bayesian Learning null-tone based OFDM decoding based on [1]. % % rcvSig: received signal time x nOfdmframes % nullToneList: list of null tones in each ofdm symbol % symDemod: symbol demodulator (see communication toolbox) % nEmItr: number of EM iteration in the SBL % decodSyms: a stream of decoded symbols % % [1] J. Lin, M. Nassar and B. L. Evans, "Non-Parametric Impulsive Noise % Mitigation in OFDM Systems Using Sparse Bayesian Learning", % Proc. IEEE Int. Global Communications Conf., Dec. 5-9, 2011, % Houston, TX USA. % Arranging data into the System Model Format fftSize = size(rcvSig,1); F = 1/realsqrt(fftSize) * dftmtx(fftSize); % normalized DFT mtx % Switching to paper notation (See Reference) y = 1/realsqrt(fftSize)*fft(rcvSig,fftSize,1); yNull = y(nullToneList,:); % y_I FNull = F(nullToneList,:); % F_I % SBL by EM noiseEst = arrayfun(@(indx) sblBasisSel(yNull(:,indx),FNull,nEmItr),1:size(yNull,2),'UniformOutput',false); noiseEst = [noiseEst{:}]; % Denoise received signal densdRcvSig = rcvSig - noiseEst; % Detection demodSymbols = 1/realsqrt(fftSize)*fft(densdRcvSig,fftSize,1); signalNorm = 1/modnorm(symDemod.constellation, 'avpow',1); dataToneList = setdiff(1:fftSize,nullToneList); [dataTonesSignal] = demodSymbols(dataToneList,:); decodSyms = reshape(demodulate(symDemod, signalNorm*dataTonesSignal),[],1); end