Context-aware Caching Scheme For Wireless Networks

With the growth of Internet services and the popularization of smart mobile terminals,the worldwide mobile data traffic is growing at an exponential rate,with video traffic accounting for 70%of the total.It is a huge challenge for operators to ensure performance under such network pressure.Traditional base station-centric techniques,such as increasing base station deployment density and frequency reuse efficiency,can mitigate the problems caused by the surge in mobile traffic to some extent,but they cannot always resolve the contradictions between limited backhaul link capacity and continuously increasing data traffic.With the development of wireless communications and the evolution to the 5G era,people will play an increasingly important role.By exploiting the behavioral patterns of users,such as the asynchronous file request pattern,wireless edge caching technology has emerged.The popular files repeatedly requested by users are cached at the edge networks,such as a cellular network or a D2D(Device-to-Device)network,and users thus do not have to download files from the core network through backhaul links,hence releasing the pressure of the backhaul links and further improving the network performance.Although there have been many studies on wireless edge caching schemes at home and abroad,there are still two major problems.On the one hand,most of the works did not break out of the framework of the research of wired caching schemes.In other words,Context information in wireless access networks,such as the user's mobility patterns,social networks,etc.,were not fully exploited.On the other hand,although there are many researches on related caching strategies in cellular networks or in D2D networks,there are very few works able to answer the question of which is better between caching at base station and caching in D2D networks under heterogeneous networks.In view of the aforementioned problems,this paper studies how to use the context information in wireless networks to improve the performance of edge network caching from the perspective of context-aware at the edge to improve the performance of edge network caching.Furthermore,some fundamental insights into how to balance the advantages of caching at base station and D2D networks is provided.The main work of this paper and its main contributions are the following three aspects:1.Research of mobility-aware caching scheme for cellular network networks:Taking user mobility into account,and integrating both the contextual information of mobility patterns and distributed cache into the classical probabilistic caching scheme,we propose a mobility-aware coded probabilistic caching scheme for small base station networks.The proposed scheme balances the gains of user mobility,content diversity,and channel selection diversity,with the goal of maximizing the throughput of cellular networks.Based on stochastic geometry,we numerically analyze the impacts of user mobility,file popularity,and backhaul capacity on the proposed caching schemes and its performance.It is found that our proposed scheme achieves a higher throughput than other caching schemes under different degree of user mobility,content popularity skewness and backhaul capability,which is more remarkable with intense user mobility,flat popularity profile and poor backhaul capability.2.Research of mobility-aware caching scheme in D2D networks:We first study mobility-aware caching scheme under D2D networks,where user mobility is modeled as queuing behavior.By solving the optimal problem,the cache decisions with the goal of maximizing throughput is obtained,and the rule of the decision changes with the user's mobility intensity,file popularity,user density and other factors.Finally,on the basis of the obtained research results,and taking user mobility,file popularity and user density into account,the question of which is better between caching at base station and caching in D2D networks under heterogeneous networks is answered.3.Research of multi-context aware caching scheme:Different from the above research built on typical user analysis,we shift our focus from a typical user to all the subscribers in wireless networks.Exploiting real-time user information in the wireless network,such as the user's moving path,file access history,traffic usage trace,and social networks,etc.an algorithm is provided to learn the behavior of different types of users online,and establish the mapping relationships between user types and behavior patterns.Finally,a user's behavior pattern is predicted and the file pre-caching is achieved,with the goal of maximizing the amount of the offload backhaul data traffic.