Cooperative Detection for Cell-Free Massive MIMO With Multiple Central Processing Units Under Backhaul Capacity Limitation

Cell-free massive multiple-input multiple-output (CF-mMIMO) systems offer enhanced spectral efficiency (SE) and uniform coverage through cooperative baseband signal processing across geographically distributed, interconnected central processing units (CPUs). However, signal distortions due to backhaul link capacity limitations can significantly hinder system performance. To address this challenge, we propose distortion-aware clustering (DAC) and distortion-aware adaptive clustering (DAAC) methods to enhance scalability and robustness against distortions. We evaluate their performance against the baseline proximity-based clustering (PC) method under two cooperative data detection approaches: centralized detection (CD) and two-stage detection based on local combining (TDLC). Simulation results demonstrate that (i) DAC and DAAC outperform PC in terms of SE, fairness, and robustness against distortions with DAC improving SE over PC by approximately 7.5, 1.2, and 0.02 dB, while DAAC achieves gains of 10.7, 1.7, and 0.2 dB under low, medium, and sufficient backhauling capacities respectively; (ii) in contrast to existing literature, TDLC can outperform CD depending on the system’s backhauling capacity; (iii) TDLC provides a more favorable tradeoff between backhaul capacity requirements and achievable SE. These findings offer valuable insights for designing scalable, distortion-resilient CF-mMIMO systems and lay the groundwork for efficient resource allocation and network deployment strategies for future wireless communication systems.