An Evaluation of Machine Learning Algorithms To Detect Attacks in Scada Network

Today, with the digitization of manufacturing, industries are becoming more and more competitive and offer personalized products and services. With The Industry of the Future also known as Industry 4.0 or Fourth Industrial Revolution, architecture has fundamentally changed. It is based on intelligent automation and integrates a new technologies that allow permanent communication between machines and systems to ensure an increasing improvement in production. Despite the advantages of industry 4.0, there are also challenges to be addressed: Network security has become very important than before, therefore we have to guarantee better intrusion detection, based on machine learning algorithms. To this end, we examine four supervised learning methods, namely: Naïve Bayes, Support Vector Machines (SVM), Trees J48 and Random Forest, and compare the results using different performance measures (accuracy, recall, time to build the model etc). The data set used in this study comes from a laboratory scale gas pipeline, entitled “10% Random Sample Gas Pipeline The objective of this research is to select the best algorithm to detect and predict attacks in SCADA Network, in order to take preventive measures against the risk of intrusion.

[2]  Karen A. Scarfone,et al.  Guide to Industrial Control Systems (ICS) Security , 2015 .

[3]  Igor Nai Fovino,et al.  Critical State-Based Filtering System for Securing SCADA Network Protocols , 2012, IEEE Transactions on Industrial Electronics.

[4]  Paola Zuccolotto,et al.  Variable Selection Using Random Forests , 2006 .

[5]  Sara Matzner,et al.  An application of machine learning to network intrusion detection , 1999, Proceedings 15th Annual Computer Security Applications Conference (ACSAC'99).

[6]  I. Maqsood,et al.  Random Forests and Decision Trees , 2012 .

[7]  Tina R. Patil,et al.  Performance Analysis of Naive Bayes and J 48 Classification Algorithm for Data Classification , 2013 .

[8]  Xinghuo Yu,et al.  An unsupervised anomaly-based detection approach for integrity attacks on SCADA systems , 2014, Comput. Secur..

[9]  Thomas J. Watson,et al.  An empirical study of the naive Bayes classifier , 2001 .

[10]  Vikramaditya R. Jakkula,et al.  Tutorial on Support Vector Machine ( SVM ) , 2011 .

[11]  Neelam Sharma,et al.  INTRUSION DETECTION USING NAIVE BAYES CLASSIFIER WITH FEATURE REDUCTION , 2012 .

[12]  Ulf Lindqvist,et al.  Using Model-based Intrusion Detection for SCADA Networks , 2006 .

[13]  Jean-Michel Poggi,et al.  Variable selection using random forests , 2010, Pattern Recognit. Lett..

[14]  R. Kaboré,et al.  Revue des systèmes de détection d'anomalies dans les réseaux SCADA et attaques internes , 2017 .

[15]  Leo Lebanov,et al.  Random Forests machine learning applied to gas chromatography - Mass spectrometry derived average mass spectrum data sets for classification and characterisation of essential oils. , 2020, Talanta.