Deep Learning and Social Relationship Based Cooperative Caching Strategy for D2D Communications

Abstract

Device-to-device (D2D)caching has recently been proposed as a promising solution to alleviate the traffic burden for cellular networks. In such D2D caching-enabled networks, the cache strategy including where to cache and how to cache has a substantial impact on the system performance in terms of traffic offloading. In this paper, we propose a comprehensive cooperative caching strategy based on deep learning and social relationship, aiming at effectively reducing the traffic load of D2D cellular networks. An optimization problem is formulated to maximize the data offloading ratio by finding appropriate caching nodes and designing content placement strategies. To find the solution, a heuristic algorithm including content request prediction and content placement is developed. For content request prediction, we propose a method based on the neural network collaborative filtering (NCF)algorithm, which can improve the prediction accuracy by fusing recommendation algorithm and deep learning. For content placement, we develop a multi-user cooperative caching mechanism based on the social relationship and user activity level. Simulation results demonstrate that the proposed caching policy can achieve effective data offloading.