基于重加权近似消息传递算法的稀疏信道估计

Reweighted approximate message passing for compressive channel sensing

基于近似消息传递(AMP)算法,提出了一种重加权近似消息传递(Rw AMP)算法用于稀疏信道估计,该算法增加了重加权过程与衰减机制。在信号重构时加入权重更新过程,待重构信号中较大的元素几乎不变,较小元素迅速降低为零,从而提高正确重构概率。其次,当稀疏信道估计的观测矩阵为Toeplitz矩阵时,为了提高AMP类算法收敛性,文中加入衰减机制。仿真结果表明,在相同复杂度的条件下,Rw AMP算法的信号重构性能优于基追踪(BP)算法和AMP算法。

In this paper, a new version of approximate message passing with reweighting and damping mechanisms （RwAMP） is designed for compressive Channel sensing （CCS） problem. By the reweighting mechanism, large signal elements will be changed slowly or even unchanged during the iterations, and small signal elements will be diminished to zero quickly. Secondly, unconvergence of AMP is observed under the scenario of Toeplitz measurement matrix in CCS, so damping technique is adopted to improve the algorithm stability. The simulation results show that, on signal reconstruction performance, RwAMP outperforms basis pursuit （BP） with much lower complexity, and is much more superior to AMP with equivalent complexity.