Distributed Caching in Small Cell Networks

The dense deployment of small cells in indoor and outdoor areas contributes mainly in increasing the capacity of cellular networks. On the other hand, the high number of deployed base stations coupled with the increasing growth of data traffic have prompted the apparition of base stations fi tted with storage capacity to avoid network saturation. The storage devices are used as caching units to overcome the limited backhaul capacity in small cells networks (SCNs). Extending the concept of storage to SCNs, gives rise to many new challenges related to the specific characteristics of these networks such as the heterogeneity of the base stations. Formulating the caching problem while taking into account all these specific characteristics with the aim to satisfy the users expectations result in combinatorial optimization problems. However, classical optimization tools do not ensure the optimality of the provided solutions or often the proposed algorithms have an exponential complexity. While most of the existing works are based on the classical optimization tools, in this thesis, we explore another approach to provide a practical solution for the caching problem. In particular, we focus on matching theory which is a game theoretic approach that provides mathematical tools to formulate, analyze and understand scenarios between sets of players. We model the caching problem as a one-to-one matching game between a set of files and a set of base stations and then, we propose an iterative extension of the deferred acceptance algorithm that needs a stable and optimal matching between the two sets. The experimental results show that the proposed algorithm reduces the backhaul load by 10-15 % compared to a random caching algorithm