AP Working Mode Selection and Blocklength Optimization for Industrial URLLC in Network-Assisted Full-Duplex Cell-Free Massive MIMO Systems

Network-assisted full-duplex (NAFD) cell-free massive multiple-input–multiple-output (MIMO) systems, which virtually achieve FD transmission by employing half-duplex access points (APs) to simultaneously serve uplink (UL) and downlink (DL) users on the same frequency bands, can decrease latency and enhance reliability since self-interference cancellation is unnecessary and macrodiversity is provided, thus helping industrial ultrareliable low-latency communication applications. In this article, we consider a monitoring scenario of industrial automation in NAFD cell-free systems using finite blocklength codewords and analyze UL and DL signals, end-to-end (E2E) delay, and E2E decoding error probability (DEP). Then, an optimization problem is formulated to minimize the maximum E2E DEP among all actuators by jointly designing AP working mode, UL and DL blocklengths under the E2E delay constraint. We propose the estimation of distribution algorithm-differential evolution (EDA-DE) method with low complexity to obtain a near-optimal solution, where the block coordinate descent is used to divide this problem into two parts. The first AP working mode selection is solved by EDA, and the second UL and DL blocklengths are designed by DE. Simulation results indicate that the performance of our proposed EDA-DE method is close to that of exhaustive search with lower computational complexity.