Training Sequence Design and Optimization for Structured Compressive Sensing Based Channel Estimation in Massive MIMO Systems

Abstract

In this paper, a highly accuracy and spectrally efficient channel estimation scheme based on training sequence (TS) design and optimization is investigated under the framework of structured compressive sensing (SCS). The auto-coherence and cross-coherence of the blocks are proposed and specified as two key merit factors, which is a new perspective to optimize the block coherence of sensing matrix. To optimize these two factors, a TS, obtained from the inverse discrete Fourier transform (IDFT) of a frequency domain binary training sequence, is designed, and a genetic algorithm is adopted afterwards. Simulation results demonstrate that the proposed TS design and optimization method can significantly decrease the block coherence of the measurement matrix. Moreover, by exploiting the proposed optimized TS's, the channel estimation outperforms conventional TS design obtained by brute force searching in mean square error, and can approach the Cramer-Rao lower bound. Therefore, the proposed TS design and optimization scheme can be a promising technology in the future 5G communications.