Capacity-Aware Routing Using Throw-Boxes

Deploying the static wireless devices (throw-boxes) at certain locations can help message dissemination among mobile nodes in delay tolerant networks. The mobile nodes traveling between throw-boxes form network links that carry the temporally stored messages at one box to another. However, the aggregated mobility between pairs of boxes reveals a time-dependent nature on its link delay (travel time) and link loading capacity (buffer size). The state-of-art DTN routing protocols do not address the challenge of combined delay and capacity with time-dependent links. In this paper, we tackle this challenge by introducing a capacity-aware routing protocol that is able to search the shortest path that considers the time-varying delay and capacity of the virtual links. We use a Markov Chain to model the evolution of the real-time link delay and capacity, and use the Markov Chain states to help derive the forwarding decision and routing policy. In evaluating the capacity-aware routing scheme, we use a network graph with the virtual links extracted from the contact trace. The simulations validate the advantages of the proposed routing strategy.