针对稀疏MIMO信道系统模型线性均衡过程中输入信号，输出信号都含有噪声的情况提出了一种变遗忘因子的稀疏正则化总体最小二乘算法（VFF-SRTLS）。本算法中采用总体最小二乘（TLS）的代价函数即瑞利商加入正则化的l1范数和l0范数作为其代价函数，并利用次梯度下降法产生的迭代式用以更新均衡滤波器系数，使均衡过程中代价函数最小；同时为了使算法能够适应信道快变环境而采用变遗忘因子（VFF），并且根据最速下降法得到遗忘因子的迭代式。仿真结果表明，在信噪比为10 dB的2×2 MIMO线性均衡过程中VFF-l1-RTLS算法的收敛MSE值比RLS算法低约2 dB，VFF-l0-RTLS算法的收敛MSE值比RLS算法低约1.5 dB。
This paper proposes a sparsity regularized total least square algorithm with variable forgetting factor (VFF-SRTLS) for linear equalization (LE) of the MIMO system models in which the input signals and the output signals are both contaminated with noises. In this algorithm, regularized l1-norm and l0-norm penalty functions are added in the cost function of the total least square algorithm (TLS) which is the Rayleigh Quotient and it utilizes subgradient descent method to get the recursive equations for updating the equalizer filter weights which minimizes the cost function in the equalization processing; and for improving the adaptive capacity of rapidly changed environment it adopts a variable forgetting factor (VFF) and gets the update equation of the forgetting factor according to the steepest descent method. The simulation results show that the VFF-l1-RTLS algorithm improved the MSE performance by about 2dB, and the VFF-l0-RTLS algorithm improved the MSE performance by about 1.5dB than the RLS algorithm in the LE processing for the sparse 2×2 MIMO system.
2016,(s1): 152-157 收稿日期：2016-07-31
DOI：10.3404/j.issn.1672-7619.2016.S1.028
分类号：TP301.6
基金项目：国家自然科学基金项目（50909029，61471138）；国际科技合作专项项目（2013DFR20050）；水声技术重点实验室基金项目（201420040）；国防科学技术工业委员会基础研究基金项目（B2420132004）
作者简介：张友文(1974-),男,副教授,研究方向为水声通信及组网技术、水声阵列信号处理技术。
参考文献：
[1] HAYKIN S. Adaptive filter theory[M]. Upper Saddle River, HJ:Prentice Hall, 2002.
[2] LJUNG L, SÖDERSTRÖM T. Theory and practice of recursive identification[M]. Cambridge, MA:MIT Press, 1983.
[3] DAVILA C E. An efficient recursive total least squares algorithm for FIR adaptive filtering[J]. IEEE Transactions on Signal Processing, 1994, 42(2):268-280.
[4] EKSIOGLU E M, TANC A K. RLS algorithm with convex regularization[J]. IEEE Signal Processing Letters, 2011, 18(8):470-473.
[5] ZAKHAROV Y V, NASCIMENTO V H. DCD-RLS adaptive filters with penalties for sparse identification[J]. IEEE Transactions on Signal Processing, 2013, 61(12):3198-3213.
[6] TANC A K. Sparsity regularized recursive total least-squares[J]. Digital Signal Processing, 2015, 40:176-180.
[7] MALEKI-TEHRANI A, HASSIBI B, CIOFFI J M. Adaptive equalization of multiple-input multiple-output (MIMO) frequency selective channels[C]//Proceedings of the 33rd Asilomar Conference on Signals, Systems, and Computers. Pacific Grove, CA:IEEE, 1996:547-551.
[8] FOSCHINI G J, GOLDEN G D, VALENZUELA R A, et al. Simplified processing for high spectral efficiency wireless communication employing multi-element arrays[J]. IEEE Journal on Selected Areas in Communications, 1999, 17(11):1841-1852.
[9] MALEKI-TEHRANI A, HASSIBI B, CIOFFI J M. Adaptive equalization of multiple-input multiple-output (MIMO) channels[C]//Proceedings of 2000 IEEE International Conference on Communications. New Orleans, LA:IEEE, 2000:1607-1674.
[10] ARABLOUEI R, DOĞANÇAY K. Low-complexity adaptive decision-feedback equalization of MIMO channels[J]. Signal Processing, 2011, 92(6):1515-1524.
[11] RONTOGIANNIS A A, KEKATOS V, BERBERIDIS K. An adaptive decision feedback equalizer for time-varying frequency selective MIMO channels[C]//Proceedings of the 7th Workshop on Signal Processing Advances on Wireless Communications. Cannes, France:IEEE, 2006.
[12] FENG D Z, ZHANG X D, CHANG D X, et al. A fast recursive total least squares algorithm for adaptive FIR filtering[J]. IEEE Transactions on Signal Processing, 2004, 52(10):2729-2737.
