Reliable Communication Through Energy-Efficient Congestion Control Mechanism Utilizing Deep Learning and Meta-Heuristic Optimization in Wireless Multimedia Sensor Networks

Abstract

Wireless Multimedia Sensor Networks (WMSNs) are plagued with issues of battery life limitation and congestion, which lead to packet loss, energy consumption, and delay. In this paper, an energy-aware congestion control architecture named OCNN-TDO-WMSN, combining Orthogonal Convolutional Neural Networks (OCNN) and Tasmanian Devil Optimization (TDO), is proposed. The system uses a rate-based congestion control with cluster routing to reduce energy and delay. Double Fuzzy Clustering-driven Context Neural Networks (DFCCNN) are employed for clustering, OCNN for cluster head election, and TDO for adaptive packet rate adaptation. Throughput is additionally enhanced through the use of a Semi-Decentralized Energy Routing Algorithm (SDERA). The proposed method is simulated in NS-3 and is tested with parameters of energy consumption, lifetime, delay, throughput, packet delivery ratio (PDR), and reliability. Results indicate that OCNN-TDO-WMSN reaches a throughput of 50 Kbps and a PDR of 95%, better than the current methods.