SMSR: A Scalable Multipath Source Routing Protocol for Wireless Sensor Networks

In wireless sensor networks (WSNs), providing resilience (fault tolerance) is a challenging issue. A lot of multipath routing protocols, therefore, have been proposed to achieve the goal; however, they usually suffer from control message overhead or a lack of scalability. Although some protocols utilize partially disjoint paths or longer alternate paths in order to reduce such overhead, they cannot guarantee resilience, because a single failure on a shared node breaks all the paths. In this paper, we therefore propose a scalable multipath source routing (SMSR) protocol. In SMSR, a sink node collects each sensor node's one-hop upstream neighbor information during an initialization phase which the sink node then uses in order to construct several shortest node-disjoint downstream (sink-to-sensor) paths. When transmitting downstream packets, the source routing technique is exploited. On the other hand, each sensor node forwards upstream (sensor-to-sink) packets to one of its upstream neighbors, through the concept of gradient-based routing. The initialization phase depends on only one-time flooding and n (network size) times unicasting, and each sensor node manages only one-hop upstream neighbor information. In this sense, SMSR is scalable in terms of the overhead and the size of routing tables in sensor nodes. Particularly, since SMSR provides several node-disjoint paths with low overhead, it can guarantee resilience efficiently. Through experiments using both ns-2 simulation and our real world test-bed, we verify that SMSR achieves the goal better than other existing routing protocols.