Home // SECURWARE 2017, The Eleventh International Conference on Emerging Security Information, Systems and Technologies // View article

Clustering based Evolving Neural Network Intrusion Detection for MCPS Traffic Security

Authors:
Nishat I Mowla
Inshil Doh
Kijoon Chae

Keywords: Intrusion Detection; Machine Intelligence; Clustering; Neural Networks; Medical Cyber Physical Systems.

Abstract:
In the era of Internet, exploits and vulnerabilities of our systems can be used by attackers to violate confidentiality, integrity, and availability. These attacks pose even more serious consequences when we consider medical networks such as Medical Cyber Physical Systems (MCPS). Therefore, the design of an efficient intrusion detection system is vital. However, the success of most of these systems is linked to custom statistical signature based solutions. It becomes a limiting constraint when there are myriad possible attacks emerging every day. To solve the above issues, several machine learning techniques have been developed to form robust detection systems. Nevertheless, these systems are not efficient with low-frequency attacks and are often considered as outliers though the consequences of missing upon such attacks can be dangerous. Therefore, this paper proposes an evolving machine learning technique, based on clustering and neural network classification to improve the detection accuracy of all forms of network intrusion traffic. Our experimental results on the standardized Knowledge Discovery and Data Mining (KDD) Cup 99 public dataset, show that the proposed mechanism can outperform the well-established boosted decision tree algorithm under different feature selected environments.

Pages: 20 to 25

Copyright: Copyright (c) IARIA, 2017

Publication date: September 10, 2017

Published in: conference

ISSN: 2162-2116

ISBN: 978-1-61208-582-1

Location: Rome, Italy

Dates: from September 10, 2017 to September 14, 2017