多模盲均衡算法的一种最速下降优化实现方法

盲均衡算法的传统实现方式大都基于随机梯度下降法,随机梯度法实现简单的代价是较慢的收敛速度和较大的稳态误差.文章基于实用性的考虑,采用批数据处理方式,给出了多模盲均衡算法的一种最速下降实现方法,该方法在每次迭代过程中不需要产生均衡器输出,而是直接利用接收数据的统计量和当前均衡器系数来估计代价函数的最速下降方向,具有收敛速度快,稳态误差小的特点;可以实现流水线式的实时处理,适合包长固定或可变的信号传输场合.

Steepest Descent Algorithm Implementation for Multimodulus Blind Equalization

Blind equalization algorithms are generally optimized via stochastic gradient descent (SGD) method. SGD has low complexity at the cost of slower convergence and greater steady state error. From practical considerations, this paper presents a novel steepest descent batch implementation method for multimodulus blind equalization algorithm, which does not generate intermediate receiver output in each iteration. The steepest descent direction of the cost function can be determined based on equalizer parameters and pre computed channel output statistics, which results in higher convergence speed and less steady state errors. The novel method will only introduce a fixed processing delay pipeline, making it well suited for both fixed and variable length packet transmissions.