Linear Complexity ZF-based linear precoder for massive-MIMO Systems

The precoding of the downlink-transmit symbol vector in a massive multiple input multiple output (m-MIMO) system has seen numerous linear precoding schemes. When the base station (BS) has ideal channel state information (CSI), techniques such as maximum ratio transmission (MRT) and zero-forcing (ZF) are used. For Rayleigh fading, the sumrate of the ZF precoding scheme is significantly higher than that of the MRT precoding technique due to interference mitigation. However, the ZF precoding scheme requires computing the inverse of the transpose of the channel's Gram matrix, which has significant complexity compared to the MRT precoding scheme. To reduce this complexity and simultaneously achieve the best sumrate, we propose a novel technique that modifies the ZF precoding using Decentralized pre-processing and recently proposed Cayley-Hamilton Theorem with Unit Eigenvalue (CHTUE) and Cayley-Hamilton Theorem with maximum Eigenvalue (CHTME) based inverse approximation. Simulations show that the proposed method's sumrate is much better than the MRT precoder and near ZF precoder, while the computational complexity is significantly low.