基于自适应混合能量参数的变步长LMS水声信道均衡算法

Variable step size LMS equalization algorithm based on adaptive mixed-power parameter in underwater acoustic channels

提出了一种新的变步长算法,并将该算法用于水声信道均衡。该算法克服改进归一化最小均方(developed normanized least mean square,XENLMS)算法依赖固定能量参数λ的局限性,遵循变步长算法的步长调整原则在XENLMS算法的基础上引入一个自适应混合能量参数λk,改善算法收敛速度和鲁棒性。首先通过仿真分析变步长算法中的3个固定参数α,β,μ的取值范围及对算法收敛性能的影响;并在两种典型的水声信道环境下,采用两种调制信号对算法的收敛性能进行计算机仿真,结果显示,新算法的收敛速度明显快于XENLMS算法和已有的变步长算法,收敛性能接近递归最小二乘(recursive least square,RLS)算法的最优性能,但计算复杂度远小于RLS算法。最后,木兰湖试验验证了带判决反馈均衡器(decision feedback equalization,DFE)结构的新算法具有较好的克服多径效应和多普勒频移补偿的能力,相比LMS-DFE提高了一个数量级。

An improved novel variable step size least mean square （VSS-XENLMS） adaptive filtering algorithm is proposed and it is applied to underwater acoustic equalization. A variable mixed-power parameter λk is introduced whose the time variation allows the algorithm to follow fast changes in the channel. The proposed al- gorithm overcomes the dependency on the selection of the mixing parameter λ, which has been by developed nor- manized least mean square （XENLMS） algorithm. The selecting about three factors α,β and μ and their influences to convergence ability are analysed. Computer simulations of the proposed algorithm about convergence a- bility are carried out respectively under two underwater acoustic channels, using two modulation signals. Simu- lation results demonstrate that the convergence speed of the proposed algorithm compared with that of XENLMS algorithm and the former variable step-size algorithms has been visibly increased, the convergence performance of the proposed algorithm is compared to that of recursive least square （RLS）, but its computation complexity is far less RLS. Then, Mulan Lake experiment shows that the performance of the decision feedback equalization （DFE）-based the proposed algorithm （VSS-XENLMS-DFE） is better than that of the LMS-DFE algorithm in terms of bit error rate for an order of magnitude, which overcomes the effects of multipath and Doppler shift very well.