Congestion Control of Wireless Sensor Networks based on L1/2 Regularization

Abstract

The characteristics of the throughput, packet loss rate and delay of the network are the kernel problems that must be addressed to alleviate the congestion of the wireless sensor network in the process of transmitting data. In this study, a congestion control of wireless sensor networks based on L 1/2 regularization algorithm was proposed to solve the problem of congestion near the central node. First, the algorithm compressed and observed the collected data at the transmission layer, the fuzzy neural network was used to adjust the dimension of the compressed sensing observation matrix automatically. Then, a fuzzy PID queue management algorithm was created to maintain the length of node queues around the expected value. Finally, the compressed data was retracted by the L 1/2 regularization method at the link level. The result show that the algorithm can alleviate the congestion problem in wireless sensor networks effectively. The throughput of the network increases by 35%-50%, the packet loss rate decreases by 20%-45%, and the delay reduces by nearly 5s. The experiment results demonstrate that can achieve the global control of network congestion with high reconstruction precision and low data loss under different congestion conditions.