Enabling Relay Selection in Cooperative Networks by Supervised Machine Learning

In fifth generation (5G) networks, cooperative transmission assisted by relays is believed to be an essential element, which significantly improves the system reliability and enhances the network design flexibility. To coordinate multiple relays in cooperative networks and utilize them in an efficient manner, relay selection is required. In this paper, we enable a generic relay selection scenario by supervised machine learning techniques and propose a prototype framework for further investigation. The prototype framework is constructed by a relatively simple artificial neural network consisting of only one hidden layer and the number of neurons in the hidden layer is equal to the number of inputs/outputs. Numerical results show that any relay selection criteria that conform to a certain form can be implemented by such a simple prototype framework, which can reduce the required system complexity for performing complicated processing of relay selection by conventional algorithms. Furthermore, we also point out a number of potential research directions that are worth investigating as future work.