Distributed Signal Detection with Multi-shot Weighted Combining

Recently, the distributed multiple-input multiple-output (MIMO) systems have been widely investigated, where several physically separated access points (APs) connect to a central processing unit (CPU) via fronthaul links to realize joint signal processing. Although appreciable spatial degrees of freedom can be exploited, the huge interaction overhead between APs and CPU is inevitable and prevents this architecture from being practically implemented. To achieve high signal detection accuracy while reducing interaction overhead, we propose a distributed signal detection scheme with multi-shot weighted combining, where the detected signals are iteratively refined via AP detections and subsequent CPU combining. Specifically, the regularized detection is carried out at the APs by penalizing any discrepancies between the local least-square detected signals at the AP and the weighted combined signals fed back from the CPU. The regularized detections are then collected and combined at the CPU for a refined joint detection. Using the operator-valued free probability theory, the combining weights at the CPU only depend on the statistical channel state information between APs and UEs, which alleviates the interaction overhead of fronthaul links. Numerical results demonstrate that the proposed distributed detection with multi-shot combining scheme rapidly converges and achieves improved detection accuracy compared to the detection schemes with one-shot combining.