Ameliorating cellular traffic peaks through preloading and P2P communications

An increasingly common problem facing cellular data networks is the overloading issue during peak times. We propose two methods to reduce the maximum load of a cellular network caused by interest diffusion over a social network. The first is via preloading, in which social networks can be used to predict user demand and serve users early. The second method is offloading, which uses peer-to-peer communications to migrate content delivery from the cellular network to other networks. We assume that peer-to-peer communications are opportunistic and random as they hinge on users´ mobility. Despite the uncontrollable nature of these inter-meetings between users, offloading can greatly reduce the peak of the cellular traffic. These two methods work especially well in tandem, as preloading creates a base of users that can serve later users via offloading. We focus on the impact of uncertainty by examining various assumptions in detail. We begin by demonstrating that modeling the inter-meeting process as random, as opposed to deterministic, only slightly increases the cellular load peak. On the contrary, replacing the known interest diffusion process with a random process would greatly increase the cellular load peak. This peak is further increased when the commonly used assumption of knowing all users´ identities is relaxed by random identities. We close by highlighting the new challenges when preloading multiple pieces of content and propose a heuristic preloading algorithm.