A FD-EDL and Novel Clustering-Based Intrusion Detection System Using G-WEFRPO in MANET Environment

Abstract

Recently, Mobile Ad-hoc Networks (MANETs) have created great interest in wireless communication. Several vulnerabilities are present in these networks. Thus, the pre-existing techniques offered numerous solutions. However, improvement is still required for augmenting the Detection Rate (DR). In this research approach, a Frechet Distribution-based Ensemble Deep Learning FD-EDL with hybrid optimization for an Intrusion Detection System (IDS) in MANET is proposed for augmenting the DR. Primarily, the trust value is computed. After the trust evaluation, the cluster formation and the Cluster Head (CH) selection are done utilizing the Diagonal with Cosine Similarity based K-Means (DCS-KM) algorithm. Then, by utilizing the Ad-hoc On-demand Distance Vector (AODV) algorithm, the path is generated for data transmission. For avoiding packet loss, the split and share strategy is designed in the generated path. Next, by utilizing the Polynomial Structured with Nullified Coupled Markov Chain (PSNCMC) model, noise interference is estimated and mitigated. Subsequently, the data is aggregated. The features are extracted from the aggregated data, and by utilizing Gazelle with Weighted Entropy Functional Red Panda Optimization (G-WEFRPO), the significant features are chosen. Next, for detecting intrusion in the MANET environment, the chosen features are inputted to the classifier. Based on performance metrics, the proposed method's performance is analogized with the baseline techniques in experimental analysis. The proposed system obtains a higher DR than conventional models. Hence, it is highly beneficial for IDS in MANET.