Evolutionary Optimization with Variational Auto encoder based Denial of Service Attack Detection and Classification in Wireless Sensor Networks

Wireless sensor networks contain a set of independent sensor nodes (SNs), typically utilized for data collecting and tracking applications. Intrusion detection system (IDS) is an indispensable security tool to protect infrastructures and services of wireless sensor networks (WSNs) from unpredictable and invisible assaults. Certain works of machine learning (ML) were presented for ID in WSNs and have reached reasonable outcomes. But such works still require higher accuracy and efficiency towards imbalanced data complexities in network traffic. This article introduces an Archimedes Optimization with Variational Auto encoder based Attack Detection and Classification (AOVAE-ADC) for WSN. The purpose of the AOVAE-ADC technique lies in the identification and classification of attacks exist in the WSN. To attain this, the presented AOVAE-ADC technique normalizes the WSN data using min-max data standardization technique. Next, the presented AOVAE-ADC technique utilizes VAE classification process to detect attacks or intrusions. At last, the VAE parameters are effectually chosen by the use of AO algorithm. The experimental validation of the presented AOVAE-ADC algorithm on benchmark dataset stated the improved outcome of the AOVAE-ADC model over other recent approaches.