An overview of flow-based anomaly detection. (2018)

• Record Type: Journal Article
• Authors: Sharma, Rohini; Guleria, Ajay; Singla, R.K.
• Is Part Of: International journal of communication networks and distributed systems. Volume 21:Number 2(2018)
• Page Start: 220
• Page End: 240
• Publication Date: 2018
• Subjects: network flows -- anomaly detection -- security -- privacy -- flow-based dataset -- statistical techniques -- machine learning -- clustering -- frequent pattern mining -- software agents; Computer networks -- Periodicals; Telecommunication systems -- Periodicals; Electronic data processing -- Distributed processing -- Periodicals; 004.6
• Languages: English
• ISSNs: 1754-3916

Abstract

Intrusions in computer networks are handled using misuse or anomaly-based solutions. Deep packet inspection is generally incorporated in solutions for better detection and mitigation but with the growth of networks at exponential speed, it has become an expensive solution and makes real-time detection difficult. In this paper, network flows-based anomaly detection techniques are reviewed. The review starts with motivation behind using network flows and justifies why flow-based anomaly detection is the need of the hour. Flow-based datasets are also investigated and reviewed. The main focus is on techniques and methodologies used by researchers for anomaly detection in computer networks. The techniques reviewed are categorised into five classes: statistical, machine learning, clustering, frequent pattern mining and agent-based. At the end the core research problems and open challenges are discussed.