Energy-Efficient Clustering and Power Allocation in RSMA-Enabled IoT Networks with Finite Blocklength Coding and Hardware Impairments

As the number of internet of things (IoT) devices grows rapidly, ensuring energy efficiency (EE) in IoT networks is becoming a crucial concern. In dense IoT networks, scheduling multiple IoT devices over the same radio resource blocks (RRBs) simultaneously causes interference that can severely degrade a network’s EE. This paper investigates a resource optimization problem for IoT networks that considers both finite blocklength (FBL)-coded transmission and signal distortions caused by practical low-complexity radio frequency front ends that lead to hardware impairments (HWIs). We exploit ratesplitting multiple access (RSMA) to schedule multiple IoT devices over the same RRB. We propose to jointly optimize device clustering and transmit power allocation in order to enhance the network’s EE while managing interference. A twostep framework is devised to solve the optimization problem in a distributed and computationally efficient manner. First, IoT devices are clustered into non-overlapping groups and the RSMA strategy is implemented in each cluster. Second, a deep-Q network-based multi-agent reinforcement learning framework is employed to near-optimally allocate transmit power among the devices in each cluster. Our simulation results shown that our proposed framework enhances the EE of a downlink RSMA IoT network notably more than state-of-the-art transmit PA schemes do in the presence of FBL coding and HWI-induced distortions.