Feature engineering based performance analysis of ML and DL algorithms for Botnet attack detection in IoMT

The internet of medical things is one of the popular application of internet of things (IoT) where various small medical devices are interconnected with each other to share sensitive data. But IoT device and network are vulnerable to several security attacks such as sniffing, spamming, flooding etc. initiated by Botnets which is a major problem for time critical application such as healthcare. Any kind of security attacks may lead to data breach, data alteration and non-availability of data may endanger the life of a patient in critical situation. To prevent these kinds of attacks machine learning and deep learning techniques can be applied to create an effective intelligent botnet attack detection engine (IBADE). This paper presents the performance analysis of 8 popular Machine Learning Models such as naïve bayes, decision tree, random forest, support vector machine, logistic regression, single-layer perceptron, convolution neural network and multi-layer perceptron to choose the most effective one for intelligent botnet attack detection engine (IBADE) modeling. N-BaIoT Dataset is used to train and test the above mentioned algorithms while principal component analysis (PCA) and linear discriminant analysis are used for feature reduction to get higher performance. Experimental result shows that the performance of Random Forest based ML model is best when used with PCA in terms of Accuracy = 0.99 and Precision = Recall = F1 score = 1. Hence Random Forest based model is the most effective to detect various Botnet attacks in an IoT network.