Probabilistic routing based on fine-grained contact characterization in delay tolerant networks

Most existing history-based routing protocols for delay tolerant networks resort to coarse-grained encounter information for making message forwarding decisions. However, this coarse-grained information can not give precise expression of the contact patterns between nodes in the network, thereby leading to inaccurate forwarding decisions. In this paper, we present FG-PRoPHET, a probabilistic routing scheme based on fine-grained contact characterization. Each node in FG-PRoPHET uses a slotted sliding window mechanism to maintain the information of historical contacts, by which new contact information can be quickly incorporated and outdated history data can be easily removed. The granularity of collected history data can be controlled by adjusting the size of the sliding window to reflect the contact patterns. Based on the fine-grained contact statistics, a greedy forwarding scheme is designed by combining the advantages of contact duration based forwarding and quota-based routing. The performance of FG-PRoPHET was evaluated through extensive simulations, and results show that, compared with existing schemes, FG-PRoPHET can significantly enhance message delivery rate with very low communication overhead.