We introduce TAROT (Trajectory-Assisted ROuTing), a DTN routing framework that detects and extracts structure in node movement in real-time. TAROT is motivated by the postulate that mobility, in particular human mobility such as vehicles, is seldom random and thus exhibits recognizable patterns. TAROT's mobility pattern extraction capabilities transcends current solutions that rely on abbreviated (in some cases, instantaneous) snapshots of mobility history. TAROT is therefore able to predict future mobility with increased accuracy. Routing decisions are guided by node mobility patterns, ultimately resulting in more efficient routing and forwarding of messages. Our approach is capable of accommodating conditions where the best node may be one that is currently moving away from the destination. In its current implementation, TAROT uses a "controlled epidemic" approach to route messages in which nodes will only be "infected" with a message if their mobility pattern takes them closer to the destination. We evaluate TAROT's performance through simulations using the QualNet network simulator. A side-by-side comparison against Epidemic Routing under a variety of mobility and workload scenarios show that TAROT is able to match Epidemic's high data delivery guarantees at substantially reduced overhead (over 60% in some of our experiments). TAROT's efficiency comes at the price of a slight increase in delivery delay (around 20% in our experiments). We argue that applications that use intermittently-connected networked environments are inherently tolerant of delay, and therefore favor slight increases in delay for increased efficiency and reduced resource consumption.
Search
Reviews
[Write a review of this article]
Find related articles from these CiteULike users
