一种基于压缩感知的信道缩短滤波器设计

鉴于使用信道缩短滤波器的系统的复杂度会随着滤波器中非零抽头的增加而快速增大, 运用盲自适应子空间追踪(Blind Adaptive Subspace Pursuit, BASP)算法设计了一种稀疏滤波器来减少非零抽头. 首先在最小均方误差(Minimum Mean Square Error, MMSE)准则下, 将信道缩短问题转化为稀疏滤波器的设计问题, 然后通过加入稀疏度盲估计过程, 使滤波器可以自适应地改变稀疏度, 最后在子空间追踪算法(Subspace Pursuit, SP)的框架下实现非零抽头不连续的稀疏滤波器设计. 仿真结果表明, 设计的稀疏滤波器有效地实现了信道缩短, 运用该滤波器的系统可在更低的复杂度下获得更高的精度.

A Design of Channel Shortening Filter Based on Compressed Sensing

The computing complexity of the system grows fast with the number of filter’s nonzero taps. Therefore, a sparse filter is designed in this paper based on the Blind Adaptive Subspace Pursuit algorithm to reduce the number of the nonzero taps. Firstly, the channel shortening problem is transformed into a sparse filter design problem based on the Minimum Mean Square Error criterion. Then, the filter’s sparsity is adaptively changed according to a sparsity blind estimation method. Finally, a sparse filter with noncontiguous nonzeros taps is achieved under the frame of Subspace Pursuit algorithm. The simulation results demonstrate that the designed filter can shorten the channel efficiently. Furthermore, comparing with the existing sparse filters, the system using the designed filter obtains higher accuracy with lower complexity