Dynamic Multipath Routing of Multi-Priority Traffic in Wireless Sensor Networks

We present a dynamic multipath routing protocol in which packets from different applications dynamically choose their paths by taking into account the price to be paid for taking each path and their ability to pay. We propose a mechanism in which the prices reflect congestion on routers and thus the waiting time for passing the router by a packet and ability to pay is defined by the application priority and the packet waiting time. These prices increase as usage of shorter routes increases. As a result, low priority applications tend to avoid paths with high prices. Instead, they go via low price routes which may be longer but faster to pass by avoid waiting for passage at congested routers. This enables high priority traffic to get through quickly via short paths as their priority (often representing their social value) enables them to pay high prices with little wait. Thus, our approach segregates traffic flows of different applications and lowers congestion and delays for all applications. We further show that our dynamic path allocation technique performs well both in normal and emergency situations in which network is partially damaged. Our dynamic pricing mechanism quickly adapts routing to the damaged network, increases utilization of the partial network and minimizes delays.