Optimal Forwarding Criterion of Utility-Based Routing under Sequential Encounters for Delay Tolerant Networks

We propose a new utility-based routing protocol for Delay Tolerant Networks which maximizes the expected number of selected nodes being likely to finally encounter to destination nodes based on the optimal stopping theory. Our previous investigation by analyzing real experimental data showed that frequencies of human contacts generally followed a power-law distribution \cite{FM2011}. Since the power law could not been explained by usual mobility models, we proposed a novel mobility model, \text it{i.e.}, Homesick L\'evy Walk \cite{FM2011} for the explanation. We use this walk model for the performance evaluation of our protocol and some performance measures are compared to those of other protocols, such as Epidemic Routing, Spray and Wait, PRoPHET, and MaxProp. As numerical results, we find that our routing protocol achieves comparable performance in arrival rates of relaying messages to destination nodes despite a small number of message forwarding in comparison to PRoPHET and MaxProp.