基于BP网络的稀疏信道估计算法的改进

An Improved Estimation Algorithm of Sparse Channel in BP Network

为了在稀疏信道下有效地降低传统信道估计算法的复杂度,提出了一种基于反向传播(BP,back propagation)网络的正交匹配追踪后接消除(OMPC,orthogonal matching pursuit cancellation)信道估计改进算法。首先通过OMPC算法确定各个非零抽头的位置,其次设置BP网络的权值与阈值,进一步实现对非零抽头估计值的逼近,最后通过自适应调整来对信道估计的准确度进行修正。通过与基于BP和OMPC信号稀疏分解的比较发现,本方法较好地解决了信号稀疏分解计算量大、收敛速度慢、不能利用信道稀疏特性等难题。

In order to effectively reduce complexity of the traditional channel estimation algorithm under sparse channel,an improved channel estimation algorithm of orthogonal matching pursuit cancellation based on back propagation network is presented in this paper.First the location of channel nonzero-taps are determined by orthogonal matching pursuit cancellation algorithm,and then by setting the initial weight and threshold value of BP neural network,the estimate of nonzero-taps is further approximated,finally through adaptive adjustments the accuracy of the channel estimation are corrected.By comparing the sparse decomposition of the BP and OMPC,the method presented in this paper solves the problems in the signal sparse decomposition such as heavy calculation and slow convergence speed,and the method also can use channel characteristics of sparse.