Optimal Cache Resource Allocation Based on Deep Neural Networks for Fog Radio Access Networks

Abstract

Cache resource allocation is of great significance for the advanced cellular networks, especially for fog radio access networks (F-RANs). Many cache resource allocation schemes have been proposed to increase the performance of F-RANs optimally. However, it is still challenging to apply these schemes and attain live performance in F-RAN systems since most of them need accurate and real-time data which shows radio link information or other network information. This paper presents a cache resource allocation strategy based on deep neural network (DNN) along with the training method required to train the neural networks. Simulation results in terms of DNN accuracy are shown to validate that the performance of proposed method approaches to that of the conventional iterative method in most cases.