Many-to-many matching games for proactive social-caching in wireless small cell networks

International audience—In this paper, we address the caching problem in small cell networks from a game theoretic point of view. In particular, we formulate the caching problem as a many-to-many matching game between small base stations and service providers' servers. The servers store a set of videos and aim to cache these videos at the small base stations in order to reduce the experienced delay by the end-users. On the other hand, small base stations cache the videos according to their local popularity, so as to reduce the load on the backhaul links. We propose a new matching algorithm for the many-to-many problem and prove that it reaches a pairwise stable outcome. Simulation results show that the number of satisfied requests by the small base stations in the proposed caching algorithm can reach up to three times the satisfaction of a random caching policy. Moreover, the expected download time of all the videos can be reduced significantly. I. INTRODUCTION During the last decade, the rapid proliferation of smart-phones coupled with the rising popularity of Online Social Networks (OSNs) have led to an exponential growth of mobile video traffic [1]. Meanwhile, existing wireless networks have already reached their capacity limits, especially during peak hours [2]. In order to ensure acceptable Quality of Experience (QoE) for the end-users, the next generation of wireless net-works will consist of a very dense deployment of low-cost and low-power small base stations (SBSs) [3]. The SBSs provide a cost-effective way to offload traffic from the main macro-cellular networks. However, the prospective performance gains expected from SBS deployment will be limited by capacity-limited and possibly heterogeneous backhaul links that connect the SBSs to the core network [3]. Indeed, the solution of deploying high speed fiber-optic backhaul is too expensive and thus, the use of DSL backhaul connections need to be considered [4]. Distributed caching at the network edge is considered as a promising solution to deal with the backhaul bottleneck [5], [6]. The basic idea is to duplicate and store the data at the SBSs side. Consequently, users' requests can be served locally, from the closest SBSs without using the backhaul links, when possible. In prior works, the cache placement problem has been mainly addressed for wired networks, especially for Content Delivery Networks (CDNs) [7]–[10]. However, the CDNs topology is different from the small cell networks topology since the CDNs' caching servers are part of the core network while the SBSs are connected to the core network and the User Equipments (UE) via backhaul and radio links, respectively. Hence, it is necessary to explore new cach