Consensus rankings in prioritized converge-cast scheme for wireless sensor network

Data packets collected in wireless sensor networks may contain information of different importance, including regular updates and critical data triggered by target events. In a converge-cast tree, packets routed towards the data sink may be dropped due to network contention. In order to improve the reliability of critical packets, we propose a prioritized transmission scheme named Pritrans. By overhearing the data packets transmitted by the neighbors, each node will compute the consensus relation based on the packet contents and maintain a consistent ranking within the neighborhood. The parent node will send feedback messages to adjust the packet rate of child nodes proportional to their ranking, such that the channel can be prioritized for transmission of important packets. A preemptive queuing method is applied in the intermediate nodes, enabling an incoming packet with high priority to evict a low priority one when the queue is full. We provide a statistical analytical model to demonstrate and examine behavior of the proposed scheme. We have also simulated Pritrans using TOSSIM. Simulation results show that the proposed scheme can effectively increase the delivery rate of critical packets in a congested network scenario.