Congestion performance improvement in wireless sensor networks

TL;DR: This work is the first to use LIFO to improve delay and fairness in congested WSNs, and divides the single queue in each node into multiple weighted sub-queues logically, and forward packets in each sub-queue based on its weight to enhance fairness.

Abstract: Wireless sensor networks (WSNs) have expanded their monitoring and tracking applications into wide areas, such as civil, military, and aerospace fields. Despite their important role, their performance under harsh conditions still remains to be improved. Specifically, when a WSN suffers congestion, the base station (BS) can hardly receive any data from the faraway sensor nodes while it still gets a moderate amount of data from the near-by nodes. Since the goal of WSN applications is to monitor the whole designated area, such unfairness is not acceptable. In addition, the average latency during congestion is intolerably long, failing the data freshness requirement of WSN applications. Although the fairness and latency performance of congested WSNs is very crucial for WSN applications, their degradation during congestion is usually ignored by most current congestion control methods which focus more on avoiding or recovering from congestion rather than on the congestion process itself. To improve the performance during congestion, we propose a multi-queue-LIFO (Last-In, First-Out) approach. Instead of the intuitively and frequently employed FIFO (First-In, First-Out), we are the first to use LIFO to improve delay and fairness in congested WSNs. To further enhance the fairness performance, we divide the single queue in each node into multiple weighted sub-queues logically, and forward packets in each sub-queue based on its weight. This method balances the data reception from other nodes at the BS. Both theoretical analysis and extensive experiments verify the performance improvement of our approach.