Q-Learning-Based Resource Allocation in Heterogeneous Cellular Networks

Abstract

Resource allocation being essential for wireless networks has been widely modelled by various techniques such as optimization. Due to the computational complexity. Machine learning could be a potential tool for resource allocation in 5G and beyond instead of such a mathematical model. In this article, Q-learning-based resource allocation in a heterogeneous cellular network being the focus of interest is modelled to maximize the network rate. As compared with the round-robin scheduling, all user throughputs increase significantly by means of Q-learning approach resulting in such a network rate maximization. For example, there is approximately 11-percent increase in the peak throughput in the case of Q-learning. Thus, machine learning appears to be one of promising solutions for resource allocation in future wireless networks.