Linear Compression-based Precoding for Massive MIMO Systems with Decentralized Baseband Processing

Massive MIMO using decentralized baseband processing (DBP) architecture has been envisioned as a promising technique to enable the functionality of future wireless systems due to its ability to reduce fronthaul cost and computational complexity. The precoding design under the DBP architecture has been studied recently, but most of the existing algorithms are encountered with excessive interconnection costs, or the algorithms are only tailored for the total power constraints. In view of this, we propose a communication-efficient linear compression-based precoding (LCP) scheme and study the multi-carrier joint compression and precoding design (MCJCPD) problem under the practical per-antenna power constraints. Specifically, in the LCP scheme, the central unit compresses the transmit signals on subcarriers by individual compressors and then sends the low-dimensional compressed signals to the distributed units for precoding. However, the associated MC-JCPD problem is challenging to solve since the compressor and the precoders are complexly coupled. To address the problem, we propose two efficient algorithms based on the penalty dual decomposition-based method and the matrix factorization-based method. Simulation results show that the proposed LCP scheme can approach the centralized scheme but with a significantly reduced interconnection cost.