一种适用于水声信道盲均衡算法及仿真

A Fast but Accurate Algorithm for Blind Equalization of High-Order QAM Signals

对于高阶 QAM信号 ,常数模算法的收敛误差不为零 ,而多模算法的收敛速度慢。提出了一种混合常数模算法 ,并将该算法应用于水声信道盲均衡。该算法根据均衡器输出信号模值与其判决模值之间的距离 ,设置一判决域。根据该判决域 ,算法的迭代过程在常数模和多模算法之间自动切换 ,且收敛后自动切换到多模算法。该算法收敛速度快 ,且均方误差小 ,特别适合于高阶 QAM信号的盲均衡。通过对水声信道盲均衡的计算机仿真 ,验证了该算法的优点及对多途干扰抑制的有效性。

For high order QAM (Quadrature Amplitude Modulation) signals, the residual square error of CMA (Constant Modulus Algorithm) is unacceptably large and the convergence rate of MMA(Multi Modulus Algorithm) is too slow. We propose a new algorithm, which we call NCMA, to achieve both fast convergence rate and acceptably small residual square error. The key to our success is inequality (6), which defines a decision zone. When inequality (6) is not satisfied, NCMA uses CMA to utilize its fast convergence rate. When inequality (6) is satisfied, NCMA uses MMA to achieve acceptably small residual square error. ' d ' in inequality (6) is some quantity related to the distance between the modulus of the output signal and the decision signal. The iteration process of NCMA alternates automatically between CMA and MMA until it converges to MMA in the end. Fig.3 shows a typical multipath disturbance in deep Underwater Sound Channel (UWSC). The results of the computer simulation of NCMA for the deep UWSC blind equalization prove preliminarily its advantages and its ability to resist multi path disturbance.