Resilient DNN for joint sub-band allocation and power control in mobile factory subnetworks

In industrial wireless communication environments, the complexity and rapid dynamics of mobile factory subnetworks pose substantial challenges for radio resource management. This paper presents a resilient multi-task deep neural network solution tailored for simultaneous sub-band allocation and power control. Our model adopts an unsupervised training approach, leveraging intermittent channel state information (CSI) updates to navigate the challenges posed by temporal variability and uncertainty. Simulation results show that compared to iterative approaches, which often involve exponential computational complexity and require frequent CSI updates across all subnetworks, the proposed model provides a rapid inference mechanism essential for real-time operations and delivers robust performance even with outdated CSI.