Anomaly detection in computer networks: A state-of-the-art review

Abstract

The ever-lasting challenge of detecting and mitigating failures in computer networks has become more essential than ever; especially with the enormous number of smart devices that get connected to all sorts of network everyday. Whether the root cause of a given anomaly is a security breach, a component failure, an environmental factor, or even any combination of these reasons, anomalies need to be detected and mitigated timely and properly. In this paper, we review and evaluate the state-of-the-art studies on the problem of anomaly detection in computer networks. We provide an elaborate description of the anomaly detection problem, and depict the different categorizations of its solutions. We also illustrate some recent state-of-the-art solutions on the network level, and depict current trends in handling malware-induced anomalies in smartphone networks. Additionally, we evaluate the presented solutions and highlight their shortcomings.