Distributed and Centralized Mobility-Aware Caching Schemes for Collaborative Service Provision in User-Deployed Networks

Opening up to public access, a collection of user- deployed APs has the potential to reduce service costs and improve user service perception, especially for delay tolerant services. However, since the individual APs composing such a network are typically owned and operated by different users, the achievable performances are likely to be limited by the lack of coordination. By adopting an Inter AP communication Protocol (IAPP) that allows APs to collaborate in exchanging messages about the information requested by potentially incoming users, the APs are capable to start caching information, even before the users reach their communication ranges. In a multi-user scenario, in which the backbone capacity limits the performances of the APs, a set of distributed backbone resource management schemes, differing in their degrees of collaboration between APs and their mobility-awareness, are proposed and evaluated in respect to both a centralized caching server solution and a non cooperative scheme. The results show that, at the cost of increased memory requirements, collaborative caching brings significant improvements to both backbone capacity utilization and average user rates, as compared to the non cooperating scheme. At the same time, we show that for a centralized server solution to outperform our collaborative caching scheme, the central server needs to have a backbone capacity comparable to sum of the capacities of the individual APs deployed in the service area.