摘要
为了提高稀疏多径信道盲均衡器的收敛速度,提出了一种适用于多进制相移键控(MPSK,M-order phase-shift keying)信号的l0-范数约束的比例系数归一化最小均方常模盲均衡算法。该算法首先利用信号的常模特性和均衡器抽头系数的稀疏性,构造出基于l0-范数约束的稀疏常模盲均衡代价函数,然后依据梯度下降法推导出均衡器抽头系数更新公式,并对迭代步长进行归一化和比例系数化。算法为每个抽头系数分配与其当前时刻的幅度成正比的步长参数,并自适应地对幅度极小系数做向零收缩微调。理论分析和仿真实验表明,与现有稀疏多径信道盲均衡算法相比,该算法在保持较低剩余符号间干扰的同时,能有效提高均衡器的收敛速度。
In order to improve the convergence rate of the blind equalizer for sparse multipath channel, a novel blind equalization approach called l0-norm constraint proportionate normalized least mean square constant algorithm was pro- posed for M-order phase-shift keying (MPSK) signal. Based on the constant modulus characteristics of MPSK signal and the sparse property of equalizer, a new blind equalization cost function with the l0-norm penalty on the equalizer tap coef- ficients was firstly constructed. Then the update formula of the tap coefficients was derived according to the gradient de- scent algorithm. Moreover, the iteration step was updated by drawing upon the normalized proportionate factor. The algo- rithm not only assigned step sizes proportionate to the magnitude of the current individual tap weights, but also attracted the inactive taps to zero adaptively. Theoretical analysis and simulation results show that the proposed algorithm outper- forms the existing blind equalization algorithms for sparse channel in reducing ISI and improving convergence rate.
出处
《通信学报》
EI
CSCD
北大核心
2017年第1期149-157,共9页
Journal on Communications
基金
国家自然科学基金资助项目(No.61401511)~~
关键词
稀疏多径信道
快速盲均衡
l0-范数约束
比例系数因子
sparse multipath channel, fast convergence blind equalization,l0-norm penalty, proportionate factor