Distributing jobs in Delay-Tolerant Networks: Is optimization worth it?

Delay-Tolerant Networks (DTNs) refer to mobile wireless networks that lack continuous network connectivity among the nodes. To deal with such intermittent characteristics, DTNs use store-carry-and-forward based routing protocols instead of instantaneous end-to-end path transmissions. Many real-life applications of such networks have been identified and some of them pose complex computational challenges. DTN nodes, however, do not necessarily possess high degree of computational power and are also limited by their energy constraints. Distributed computing has the ability to solve computationally complex and lengthy problems by dividing them into many tasks, which are then spread over the network to be solved simultaneously, resulting in faster execution and significantly lower costs. In this paper, we study the prospect of distributed computing techniques in DTNs. We investigate different job distribution algorithms and their performance across multiple DTN scenarios. We simulate a distributed job processing system on top of a DTN for various scenarios and analyze the effect of node density, job slicing, and mobility models on their performance.