Exact greedy algorithm based split finding approach for intrusion detection in fog-enabled IoT environment. (August 2021)

• Record Type: Journal Article
• Authors: Reddy, Dukka Karun Kumar; Behera, H.S.; Nayak, Janmenjoy; Naik, Bighnaraj; Ghosh, Uttam; Sharma, Pradip Kumar
• Is Part Of: Journal of information security and applications. Volume 60(2021)
• Publication Date: 2021-08
• Subjects: Intrusion detection system (IDS) -- Internet of Things (IoT) -- Fog computing -- Critical infrastructure (CI) -- Exact Greedy Boosting (XGBoost) -- Ensemble learning -- Anomaly detection; Computer security -- Periodicals; Information technology -- Security measures -- Periodicals; 005.805
• DOI: 10.1016/j.jisa.2021.102866
• Languages: English
• ISSNs: 2214-2126

Abstract

Abstract: The essential role of intrusion detection is to manage the critical infrastructure to detect malicious activity competently concerning the Internet of Things (IoT). The IoT network is used to communicate and control information among various components composing a critical system. The essential inclination of infrastructure swerves to confront security issues and challenges as network systems are examined to expose cyber-related threats. Additionally, accessing real-time information from the cloud leads to a massive problem of latency. Fog computing is a novel archetype to encompass the cloud to the network edge with practical computation and critical infrastructure. The fog layer makes the device vulnerable to numerous attacks because of rapid access to resources. The practical way of addressing these issues is to detect intrusions effectively and trace the path leading to the attack source. The intent of this paper is to present a security mechanism and assure truthful operation of IoT networks with the intrusion detection system. A network intrusion detection system is proposed based on the conception of the Exact Greedy Boosting ensemble method for device implementation in the fog node because of protecting critical infrastructure from timely and accurate detection of malicious activities. The proposed model explores the traffic flow monitoring in novel IoT Intrusion Dataset 2020(IoTID20) network traffic by identifying and classifying the type of attack based on anomalies from normal behavior. Further, the paper estimates the complete experimentation performance and evaluations with competitive machine learning algorithms. The experimental observation of the simulation work is evident in the proposed model's efficiency and robustness in categorizing the attacks.