Routing protocols generally fall into two categories. Proactive protocols disseminate routes throughout a network regardless of the demand for such routes. While these protocols periodically incur route dissemination costs, they can provide routes quickly in response to client requests. Reactive protocols, on the other hand, discover routes only as they are needed. While this on-demand operation may increase route discovery latency, the protocol overhead is low.

The optimal routing protocol depends on network characteristics and may change dynamically. A static network where routes to all nodes are equally in demand all the time calls for a proactive protocol; a dynamic network where a few mobile nodes are the target of most route queries may benefit from a reactive protocol. In the past, the tradeoff between these protocols was made at network deployment time, fixing the protocol for the whole duration of the network lifetime.

SHARP is a hybrid routing protocol that finds the optimal mix of proactive route dissemination and reactive route discovery. An analytical model is used to make the tradeoff optimally. Each SHARP node can direct the routing layer to optimize for a different metric, such as overhead, latency or jitter, for routes targeting that node. Overall, SHARP provides an informed,anaytically-driven mechanism to find the optimal mix of proactive and reactive routing within a network.



Computer Science Department
Cornell University