Abstract
Supervised machine learning is a promising technique to achieve joint resource allocation and selection of mobile relays in cooperative heterogeneous networks. Joint resource allocation and mobile relay selection are NP-hard problems in the sense that they cannot be solved in polynomial time. To solve this problem, highly complex heuristic solutions based on channel conditions and physical-layer parameters have been considered. Our research explores an artificial neural network (ANN) approach to the joint allocation of resources and choice of a mobile relay with efficient utilization of resources in terms of energy efficiency (EE) and spectral efficiency (SE). To this, there exists a tradeoff between the two, i.e., EE and SE, and to consider this tradeoff, a new metric resource efficiency (RE) has already been introduced in the literature. This article addresses the problem of learning a regression model for determining optimal transmit power and resource allocation and choosing the appropriate relay for cooperative communication to maximize RE. Specifically, a multilayer fully connected ANN-based resource allocation and relay selection (ANN-RARS) algorithm is proposed, and a comparison of ANN-RARS with particle swarm optimization-based resource allocation and relay selection (PSO-RARS) algorithm is performed. Based on the performance analysis, ANN-RARS performs better than the PSO-RARS algorithm.