Performance of opportunistic content distribution under different levels of cooperation

In this paper, we study the performance of opportunistic content distribution in a delay-tolerant network (DTN). Our primary interest is the benefit of node cooperation on the content dissemination and we investigate thoroughly how different levels of cooperation affect the content delivery delay. The evaluation is based on two complementary approaches, analytical stochastic models (Markov chains) and simulations that use real-world contact traces with high time resolution. These two independent approaches show agreeing results on the qualitative behavior of the system. Our major finding is that node collaboration - even if limited - dramatically improves performance compared to non-cooperation. In particular, we find that limiting the number of times a node shares a content item gives only a slightly worse performance than under unlimited cooperation. Besides, we find that a critical number of nodes is necessary to disseminate content efficiently. Eventually, based on the analytic model we also find that the details of the mobility show little effect on the delivery time when the nodes cooperate.