摘要
为解决LMS类算法简单但收敛速度慢,RLS类算法收敛跟踪性能好但计算太复杂的不足,提出了基于牛顿梯度的盲最大似然序列估计算法,即将牛顿梯度算法与Viterbi-MLSE算法相结合。牛顿梯度法是用当前时刻的梯度估计代替前一时刻的梯度估计,并由矩阵求逆定理导出了一种新的相关函数自适应滤波算法。分析了基于牛顿梯度的盲最大似然序列估计算法的基本机理,并通过实验仿真验证了其可靠性。与传统的基于LMS的盲最大似然序列估计算法相比,具有收敛速度快的特点。
The algorithm of blind MLSE based on Newton gradient is an efficient algorithm tracing fast time-varying channel. Because the convergence speed of LMS algorithm is slow, so blind MLSE based on Newton gradient algorithm is proposed. Newton gradient algorithm a new correlation function adaptive filtering algorithm is proposed by replacing the previous gradient estimation with the current gradient estimation and using the matrix inversion lemma. It is analyzed the principle of blind MLSE based on Newton gradient algorithm in this paper. The simulation results and the theoretical analysis show the improved performance of the convergence speed compared with LMS blind MLSE algorithm.
出处
《太原理工大学学报》
CAS
北大核心
2008年第4期394-396,共3页
Journal of Taiyuan University of Technology