ExOR: Opportunistic Multi-Hop Routing for Wireless Networks

This paper presents ExOR, an integrated network-link layer protocol for the routing of large unicast transfers over multi-hop wireless networks. In traditional routing mechanisms, a single lossy link can trigger a complete retransmission. On the other hand, ExOR chooses the next hop of a route after confirming transmission on that hop. By deferring this choice, the packet transmission has multiple opportunities to make progress and is not restricted to a fixed path of loss-vulnerable points-of-failure. In essence, this dynamic approach to routing enables transmission (without unnecessary retransmissions) over long radio links with high loss rates and provides substantial increases in throughput.

For me, the discussion of interference between potential forwarding nodes was inadequate and I wasn't quite convinced by some of the assumptions that were made. Moreover, the evaluation results were quite conclusive but I would've preferred a better explanation of why the additional inter-node communication required for this scheme is not a major drawback in its adoption. The use of ETX as a path metric in determining hop priority was very appropriate given the papers we've read recently. Finally, I've been thinking about such dynamic routing mechanisms ever since I started learning about networking and wondering why it's rarely mentioned - in that sense, this paper was a very satisfying read. I'd definitely keep it in the syllabus. One last thing - the acronym ExOR (Extremely Opportunistic Routing) is pretty neat. This paper also described a good example of when the end-to-end principle can be compromised.