Chen, Ti YuTi YuChenYu, Chia HaoChia HaoYuTZI-DAR CHIUEH2024-04-302024-04-302023-01-019798350370218https://scholars.lib.ntu.edu.tw/handle/123456789/642196Cell-free (CF) multiple-input multiple-output (MIMO) technology is garnering significant attention as a promising solution to meet the rising demand for a large number of users with high quality of service (QoS) in the upcoming 5G and 6G wireless communication systems. In CF MIMO networks, all user equipments (UEs) share a single subcarrier, resulting in a non-orthogonal multiple access (NOMA) scheme. Effective detection of the uplink (UL) signals received at multiple sites can significantly improve the system transmission efficiency. In this paper, we proposed a dynamic message-passing algorithm (DMPA) for cell-free MIMO detection. DMPA dynamically allocates detection resources on UEs that are difficult to detect by adjusting the factor graph (FG) of the message-passing algorithm in a timely fashion. Simulation results show that the bit error rate (BER) by DMPA outperforms the existing detectors, indicating that DMPA is very competitive. We further designed three low-complexity techniques and proposed DMPA with constellation truncation (DMPACT) that has 95% lower computational complexity than DMPA with literally no BER loss. The low-complexity techniques include initial constellation truncation according to minimum mean-squared error (MMSE) combining outcomes, truncation of less-likely constellation points based on the Log-Likelihood Ratios (LLR) during DMPA, and refined FG adjustment. Simulation results show that DMPACT can decrease the SNR required for 10-3 BER by 2dB with only doubled computation complexity compared to the widely-used MMSE-SIC detector.cell-free | detection | low complexity | message passing algorithm | MIMO | NOMAconference paper10.1109/GCWkshps58843.2023.104650082-s2.0-85190289554https://api.elsevier.com/content/abstract/scopus_id/85190289554