Fuzzy computation on ensemble deep network for the performance of vampire attack detection model in WSN

Abstract Context Wireless sensor network (WSN) is susceptible to vampire attacks. It is a type of denial of service (DoS) attack that drains sensor nodes’ energy, leading to network failure. The study addresses the challenge by detecting these attacks to conserve energy and maintain network functionality. Objective The research aims to develop a fuzzy ranking-based ensemble deep network (EDN) for vampire attack detection in WSN, thereby extending the network’s lifespan and improving security. Methods The proposed model involves three phases: data collection, feature selection using the enhanced piranha foraging optimization algorithm (E-PFOA), and attack detection using EDN, which combines long short-term memory (LSTM), recurrent neural network (RNN), gated recurrent unit (GRU), and temporal convolution network (TCN). The final detection outcome is determined through fuzzy ranking. Results The sensitivity, specificity, and F1-score of the implemented vampire attack detection model were attained with 96.85%, 95.56%, and 96.10% for the k-fold value of 5, and these values are significantly higher than those of conventional approaches. The ROC curve and confusion matrix further validated the model’s effectiveness. Conclusion The study successfully implemented a fuzzy ranking-based vampire attack detection model that outperforms traditional methods, offering a promising solution for securing WSN against vampire attacks and ensuring the network’s lifetime.