应用基扩展模型的混合信号单通道盲分离算法
A Single Channel Blind Separation Algorithm for Mixed Signals Applying the Base Expansion Model

针对传统最小均方误差逐幸存路径处理(LMS？？PSP)单通道盲分离算法在时变信道下性能差的问题,提出一种基于基扩展模型逐幸存路径处理(BEM？？PSP)的单通道盲分离算法。首先对接收到的部分混合信号进行LMS？？PSP单通道盲分离,得到部分准确的信道冲激响应(CIR);然后结合时变信道下基于基扩展模型进行信道估计的思想,完成整个时间周期CIR的估计;最后采用Viterbi算法对混合信号进行序列估计,从而实现时变信道下混合信号的单通道盲分离。仿真结果表明,对于2路混合QPSK信号,在相同仿真条件下,BEM？？PSP算法较LMS？？PSP算法能降低50％的复杂度且能获得更好的性能,在20 dB处的误码率可达4×10-2,而LMS？？PSP单通道盲分离算法的误码率只能达到1×10-1,并且在同等过采样倍数下,该算法能获得更高的性能提升。
A new algorithm based on basis expansion model per？？survivor processing (BEM？？PSP) is proposed to overcome the poor performance of the traditional least mean square per？？survivor processing (LMS？？PSP) algorithm for single channel blind separation in time？？varying channels. First, a portion of accurate channel impulse response (CIR) is obtained through processing a portion of received mixed signals. Then, estimations of the CIR during the whole time period is accomplished by combining the channel estimation using the basis expansion model. Finally, the Viterbi algorithm is applied to estimate sequences to the mixed signals, and the single channel blind separation of the mixed signals is accomplished. Simulation results and a comparison with the LMS？？PSP single channel blind separation algorithm in same simulation conditions show that the complexity of the proposed algorithm reduces by 50% with better performance in processing mixed QPSK signals, and that the proposed algorithm achieves a bit error rate of 4×10-2 while the LMS？？PSP algorithm only achieves a bit error rate of 1×10-1 when the signal to noise ratio is 20 dB. It is also observed that the proposed algorithm obtains a higher performance improvement with the same oversampling ratio