Distributed subgradient methods for Delay Tolerant Networks

In this paper we apply distributed sub-gradient methods to optimize global performance in Delay Tolerant Networks (DTNs). These methods rely on simple local node operations and consensus algorithms to average neighbours´ information. Existing results for convergence to optimal solutions can only be applied to DTNs in the case of synchronous operation of the nodes and memory-less random meeting processes. In this paper we address both these issues. First, we prove convergence to the optimal solution for a more general class of mobility models. Second, we show that, under asynchronous operations, a direct application of the original sub-gradient method would lead to suboptimal solutions and we propose some adjustments to solve this problem. Further, at the end of the paper, we illustrate a possible DTN application to demonstrate the validity of this optimization approach.