Article Preview
Top1. Introduction
In recent years, the Cyber-Physical Systems (CPS) becomes very popular in network control systems, instance power systems reduction systems, and multiple linear induction traction systems. The above-mentioned systems are utilized to transfers sensor measurement data of the physical processes using a wireless network to the remote end. Because of the unreliability of the wireless network, the attacker implements cyber-attacks for reducing the performance of systems. Hence, the security issue becomes a hot topic, nowadays (Xu, et al., 2015). Potential physical and cyber-attacks by the adversaries lead to severe consequences in societies, like extensive damages to the economy, customer information leakage, endangering human lives, and destruction of infrastructures (Vempaty, et al., 2013). However, several security controls, which includes access control, authentication, intrusion detection, firewall, and the encryption tools are introduced for protecting privacy, CPSs, but still, the security problems are challenging due to the complexity of monitoring network and the physical elements (Moustafa, et al., 2017 ; Mano, 2017; Bisandu, 2018). In addition, the Intrusion Detection Systems (IDS) are established for enhancing accountability and audit. Moreover, the implementation and design of IDS in the CPS is changed from the Information and Communication Technology (ICT) systems. The major cause is that the security dimensions have various priorities in ICT and CPS. Here, the CPS has the highest priority, and then integrity, and the last priority is confidentiality (Cssp, 2009).
Cyber security systems containing various components, like protocols, network devices, and sensors, which is utilized to handle and monitor the collected information. The security and privacy of the CPSs have increased progressively for maintaining data integrity and confidentiality. Privacy-preserving techniques are introduced for avoiding sensitive information (Pawar and Anuradha, 2020 ; Jadhav et al., 2016). Asymmetric and symmetric encryption are the traditional tools to protect privacy, and is broadly utilized for preventing the unauthorized users accessing sensitive information transmitted through the networks (Baby and Chandra, 2016). Identifying anomalous behavior of the physical process with respect to fault identification where both cyber-attacks and faults produce anomaly in the physical process. Moreover, the patterns of cyber-attacks as well as faults are different and hence, the attacks are not treated as faults in the CPS (Basile, et al., 2006). The IDS is classified to anomaly-based and signature-based approaches (Veeraiah and Krishna, 2018). Here, the anomaly and signature-enabled Industrial IDS (IIDS) are complementary. However, the Signature-enabled IIDSs remove and detect the illegal packets in the CPS in which the anomaly-based IIDSs is utilized to extract normal behavior and to detect the anomaly (Garcia-Teodoro, et al., 2009 ; Liu, 2020). Nowadays, the network intrusion detection approaches are progressed to the highly sophisticated levels, which involves advanced signal processing methods, but the time series analysis, wavelets, and the principal component analysis, when compared to other techniques, are not limited. Hence, the broadly utilized detectors are non-signature and signature for detecting network anomalies (Sadreazami, et al., 2017).