Method for Dynamic Subchannel Selection in Heterogeneous Internet of Things Environments

This paper presents a novel method for dynamically selecting subchannels in heterogeneous Internet of Things networks, where fluctuating network conditions and limited device computing power pose significant challenges. By integrating a reinforcement learning framework that leverages both historical transmission data and real-time network states with a tug-of-war algorithm for resource allocation, our approach adaptively balances the probability of successful data delivery, the frequency of subchannel use, and failure rates. Simulation results demonstrate that this technique maximizes transmission efficiency while mini- mizing computational overhead, thereby improving overall network performance and reliability under high-load scenarios and stringent resource constraints.