Block-sparse hybrid precoding and limited feedback for millimeter wave massive MIMO systems

Millimeter wave (mmWave) communication using massive multiple input multiple output (MIMO) techniques has been regarded as a key enabling technology for 5G wireless system, as it can offer gigabit-per-second data rates. Due to the high cost and power consumption of mixed-signal devices in mmWave systems, the hybrid analog and digital precoding transceiver architecture has recently received considerable attention. In this paper, we fully deploy spatial structure of mmWave channel to find hybrid precoders with near-optimal performance and reduce the number of feedback bits. MmWave MIMO channel exhibits the block sparsity structure because of the limited number of scattering clusters with low angular spreads. Consequently, we formulate hybrid precoders design as a block-sparse reconstruction problem and a lower complexity algorithm for finding precoders is proposed based on the greedy sequence clustering. To assure the performance of this algorithm, a estimation method for the number of the blocks and radio frequence (RF) chains is proposed. Under this framework, for the analog precoder the number of the phases needed to be quantized is only two times minus one of that of all blocks, which remains unchanged even the number of RF chains increases. The simulation results demonstrate that the proposed algorithm can achieve the near-optimal performance and save the feedback overhead by utilizing the block sparsity information.