Centrality-Based Caching for Mobile Wireless Networks

International audienceThe introduction of anticipatory, proactive caching mech-anisms that store popular content at user terminals and base stations is emerging as an integral component of the next-generation, 5G, cellular networks [1], [2]. In essence, endow-ing cellular networks with the ability to cache users' content at the network's edge provides an effective means for offloading traffic from the resource-constrained network infrastructure while reducing access delays to requested content. However, developing efficient caching mechanisms for cellular systems raises important questions such as what, where and when to cache under given perfect or imperfect "statistics" are still under investigation (see [3]–[5] for a brief literature). In recent years, the social networks such as Facebook, Twitter, LinkedIn have become one of the major contributors to mobile data traffic with their network share is expected to increase in upcoming years. Social networking data and social content dissemination provide a key opportunity that can be exploited for optimizing the use of content caching in wireless cellular networks and reducing the traffic load over the network's backhaul infrastructure. However, privacy and security concerns may present an important barrier that prevents network operators from efficiently exploring social data in their resource allocation mechanisms. To overcome this barrier, in this short abstract, we propose a novel caching approach that exploits the fact that, although the information on content dissemination in social networks (evolution of con-tent popularity profile) may not be known, the social network structure can be readily explored for optimized caching at small base stations (SBSs). One key measure of influence in a social network is the so-called notion of centrality which allows to pinpoint important and influential nodes within a social structure [6]. Although the idea of using centrality measures for caching decision in the context of information centric networks has been explored in some existing works such as [7], our scenario is indeed different. In other words, the main contribution of this work is to explore the benefits of centrality-based caching methods in SBSs scenario, where a simplified content dissemination process in social networks is detailed additionally