A Deep Neural Network Based Optimization Approach for Wireless Resource Management

This paper demonstrates the feasibility of emerging disruptive deep learning technology to solve NP-hard transmit power control problem in future wireless networks. Existing approaches solve this optimization problem by converging to an optimal power allocation solution and require high computation time. This research proposes a deep neural network (DNN) based strategy for solving this high computational convergence time problem while maintaining sufficient sum-rate. The proposed scheme offers a low computational time to the optimization problem of radio resource allocation in real-time applications. The results reveal that our proposed DNN based approach can approximate the behavior of the optimization algorithm and speeds up the operation up to 288 times compared to an iterative based approach.