Residual-Impairments-Aware Feedforward-Feedback Detection for Massive MIMO Systems With Decentralized Baseband Processing

Abstract

Decentralized baseband processing (DBP) has emerged as a promising solution to address the high interconnection costs and computational complexity challenges of centralized processing in massive multiple-input multiple-output (M-MIMO) systems. Despite its promise, as the number of antenna clusters increases, conventional decentralized detectors experience significant performance degradation due to constrained information sharing between clusters. Furthermore, most existing works assume ideal hardware, overlooking the negative impact of hardware impairments. In this letter, we propose a decentralized feedforward-feedback (FF) detector based on vector expectation propagation (VEP) algorithm for DBP architectures. A residual-impairments-aware (RIA) approach with empirically determined transceiver noise is introduced to alleviate the detrimental effect of residual transceiver impairments. Simulation results demonstrate that the proposed FF-VEP detector outperforms conventional EP/VEP detectors, achieving performance comparable to centralized implementations, especially in scenarios with lower BS antenna-to-user ratios and spatially correlated channels. Moreover, the RIA approach also provides notable performance gain over its residual-impairments-agnostic counterpart designed under ideal transceivers.