Joint Uplink-Downlink Resource Allocation for Industrial Wireless Control Networks with Asymmetric Traffic

Industrial wireless networked control systems (WNCSs) require ultra-reliable, low-latency communication for safety-critical and time-sensitive applications. However, the inherent asymmetry between uplink (UL) data reporting and downlink (DL) control feedback, combined with channel variability, creates significant challenges for traditional resource management strategies. In this article, we propose a novel Asynchronous Dual-actor with Hybrid Critic reinforcement-learning scheme (ADHC), which jointly optimizes UL channel scheduling and DL power allocation. The scheme models UL and DL as temporally asynchronous and functionally heterogeneous actors, and these two actors learn on different time scales, while a hybrid critic blends local and global rewards to coordinate their policies. Simulation results demonstrate that ADHC reduces delay by 30% and cuts the violation ratio to just 1.2% compared to existing centralized and independent learning approaches across multiple reliability and efficiency metrics. This confirms its effectiveness in industrial control networks with asymmetric traffic.