OCCOA for clustering-based intrusion detection system with MLP-RNN architecture

WSN play a critical role in various applications such as environmental monitoring, surveillance, and healthcare. However, their deployment in open and dynamic environments makes them vulnerable to security threats. IDS are essential for safeguarding WSNs against malicious attacks and ensuring the integrity of data transmission. This research introduces a novel Clustering-based Intrusion Detection System (CIDS) tailored for WSNs. The CIDS framework encompasses two primary components: optimal CH selection and intrusion detection within the WSN. Clustering is executed based on specific constraints such as energy, distance, RSSI, delay, and PDR. The selection of optimal CHs is facilitated by the hybrid optimization approach called OCCOA, which combines Osprey optimization algorithm (OOA) and Coot Optimization Algorithm (COA). Following CH selection, intrusion detection is performed within the WSN using DL strategy. The CIDS employs a Multilayer Perceptron-Recurrent Neural Network (MLP-RNN) model to detect intrusions by assessing trust and security. Additionally, the weights of the MLP-RNN model are finely tuned through the OCCOA algorithm for optimal performance. Moreover, the proposed method achieved higher performance by considering the measures, namely accuracy, sensitivity, precision and specificity with the values of 98.25%, 99.96%, 99.37% and 97.52%, respectively.