GOF-SLFN- An Intelligent Attack Detection System against Denial of Service (DoS) attacks based on Glow Worm Swarm optimized Single Layer Feed Forward Networks for vehicular Cyber Physical Systems (VCPS)

Abstract

Vehicular cyber-physical network use either wireless communications or internet to communicate with each other for an efficient data transfer. Data includes transportation details, safety, mobility and sustainability. However, these communication networks for vehicular transportation are susceptible to different cyber-attacks, which may lead to severe damages or huge casualties in vehicular infrastructures. Hence the intelligent attack detection system needs to be obtaining the lime light of research based on recent machine and deep learning algorithms. In this paper, the GOF-SLFN (Glowworm optimized features for Single-layer feed-forward network) is proposed to detect the attacks among the vehicular cyber physical systems. Since DoS attacks can happen at any layer of OSI Model such as application, network and transport the foremost purpose of the paper is to detect the DoS attacks using the proposed intelligent algorithm. The glowworm algorithm extracted the optimum features based on the probabilistic mechanism and SLFN has been proposed using glowworm to detect the DoS attacks. The proposed system outperformed in detection when compared with artificial neural network, SLFN, Multi-layer Perceptron and random forest. The proposed GOF-SLFN has been again compared with firefly, Bat and Latent Dirichlet Allocation (LDA) and found the proposed intelligent system provided promising results for detection of DoS attacks in VCPS. The proposed system has been tested on the real time scenarios created with SUMO and OMNET++ environment integrated with CIC IDS 2018 datasets. The performance metrics considered were based on accuracy, sensitivity and specificity.