Design and Optimization on Training Sequence for mmWave Communications: A New Approach for Sparse Channel Estimation in Massive MIMO

In the next generation of cellular networks, millimeter wave (mmWave) communications will play an important role. With the utilization of mmWave communications, the massive multiple-input multiple-output (MIMO) technique can be effectively employed, which will significantly improve system capacity. However, an effective channel estimation scheme is the prerequisite of system stability and in great need for improvement in massive MIMO systems. In this paper, a channel estimation scheme based on training sequence (TS) design and optimization with high accuracy and spectral efficiency is investigated in the framework of structured compressive sensing. As a new perspective to optimize the block coherence of the sensing matrix, the auto-coherence and cross-coherence of the blocks are proposed and specified as two kinds of key merit factors. In order to optimize the two factors, specific TS is designed and obtained from the inverse discrete Fourier transform of a frequency domain binary training sequence, and a genetic algorithm is adopted afterwards to optimize the merit factors of the TS. It is demonstrated by the simulation results that the block coherence of the sensing matrix can be significantly reduced by the proposed TS design and optimization method. Moreover, by using the proposed optimized TS's, the channel estimation outperforms the conventional TS design obtained by the brute force search in terms of the correct recovery probability, mean square error, and bit error rate, and can also approach the Cramer-Rao lower bound.