A hybrid heuristic-assisted deep learning for secured routing and malicious node detection in wireless sensor networks

Abstract

Diverse routing and security protocols are implemented to enhance the efficacy when performing the packet transmission, however, finding the optimal path is highly challenging since it reduces the transmission consistency over the sensor network. Here, the security is enhanced in Wireless Sensor Network (WSN) routing where a new meta-heuristic algorithm and deep learning framework are suggested. The designed WSN model consists of various models like trust model, routing model, clustering model, and energy efficient model. Moreover, in the trust model, malicious node or attack detection is done by “Bidirectional Long-Short Term Memory (Bi-LSTM) with Recurrent Neural Network (RNN)”. By selecting the optimal path, an energy-efficient routing is implemented for secure data transmission. Here, the secure routing is implemented through the hybrid optimization algorithm named Exploration-based Pelican Black Hole Optimization (E-PBHO). The overall performance is enhanced by evaluating the standard performance measures like alive and dead nodes, network lifetime, throughput, and energy consumption. Here, the developed model provides 92% and 93% in terms of accuracy and sensitivity. Thus, the empirical outcome of the suggested model offers superior performance over than the existing approaches.