An Intelligent Clone Detection Framework in Wireless Sensor Networks Using Hybrid Multi-Objective Optimization with Random Vector Functional Neural Network Architecture

Abstract

The progress of wireless sensor networks technology normally called WSNs, which was obtaining a considerably enlarged volume of attention from scholars for the past few years. Its vast amount of sensor nodes (SN) is the main feature that creates it advantageous to the technology. The SNs are normally employed for varied applications. In spite of this, the disadvantages of WSNs are frequently used in hostile and critical situations but they don’t contain physical access. This realism makes them very susceptible to node replication or clone node threats. In recent times, deep learning (DL) has become a great tool for detecting clone attacks in WSNs, which are exposed to security attacks.  This study presents a Randomization-Based Neural Network Architecture for Intelligent Clone Detection Using Hybrid Multi-objective Optimization (RNNICD-HMO) approach. The proposed RNNICD-HMO approach is to accurately detect cloned SNs and prevent security breaches in WSNs. Initially, data preprocessing is performed data cleaning and normalization, to ensure high-quality and consistent input for analysis.  Next, the salp swarm algorithm (SSA) has been applied for feature selection to identify the most informative attributes. Subsequently, a random vector functional link (RVFL) network was utilized for efficient and reliable classification of cloned and legitimate nodes. To further enhance detection accuracy, a hybrid multi-objective ant colony system and simulated annealing approach (MACSSA) is applied for optimal parameter tuning of the classification model. The experimental study of the RNNICD-HMO system takes place utilizing dissimilar benchmark databases. The simulation outcomes stated an enhanced outcome of the RNNICD-HMO algorithm interms of various measures.