An Enhanced Grey Wolf Optimisation–Deterministic Convolutional Neural Network (GWO–DCNN) Model-Based IDS in MANET

Abstract

At present, ensuring the security of MANET is a highly challenging chore due to the dynamic topology of the network. Hence, most of the existing Frameworks for Intrusion Detection Systems (IDS) seek to predict the attacks by utilising the clustering and classification mechanisms. Still, they face the major problems of reduced convergence speed, high error rate and increased complexity in the algorithm design. Therefore, this paper intends to utilise integrated optimisation and classification methods for accurately predicting the classified label. This framework comprises the working modules of preprocessing, feature extraction, optimisation and classification. Initially, the input datasets are preprocessed by filling the missing values, and normalising the redundant contents. After that, the Principal Component Analysis (PCA) technique is employed for selecting the set of features used for improving the classification performance. Consequently, the Grey Wolf Optimisation (GWO) technique is utilised for selecting the most optimal features based on the best fitness value, which reduces the overall complexity of IDS. Finally, the Deterministic Convolutional Neural Network (DCNN) technique is utilised for predicting whether the classified outcomes are normal or attacks. For validating the results, various performance metrics have been assessed during the analysis, and the obtained results are compared with the recent state-of-the-art models.