Decision Feedback Signal Combining for Cell-Free Massive MIMO With Distributed Implementation

Cell-free massive multiple-input multiple-output (MIMO) consisting of a large number of distributed antennas has shown great potential to provide uniformly good wireless services via coherent transmission or reception. The signal combing for decoding uplink data can be performed in a centralized or distributed manner. Centralized combining outperforms distributed solutions on achievable spectral efficiency (SE) via coordinated interference suppression based on network-wide channel state information (CSI) at the central processing unit (CPU). Distributed combining preprocess the received signal at access points (APs) with a lower computational complexity based on local CSI, but its achievable SE is much lower than centralized combining. In this paper, we utilize decision feedback to narrow the performance gap of distributed combining with its centralized counterpart. We employ the detected signal of the distributed combiner as training data to estimate the equivalent channel at the CPU. Based on the estimated CSI, we propose an iterative decision feedback signal combining scheme and derive its achievable spectral efficiency. Our analytical and simulational results reveal that the proposed scheme can effectively suppress the leakage interference of distributed combiners without additional CSI exchange between APs and CPU, thus leading to significant performance improvement. Moreover, the proposed method provides high robustness against pilot contamination. It can achieve a similar or even higher 90%-likely SE than fully centralized combining with less computational complexity at the cost of higher decoding latency.