DDoS attack detection in smart home applications

Abstract

AbstractIn smart applications, edge nodes are deployed to perform faster computations. Due to the limited computational capability of edge nodes, collaborative computing is used in which multiple edge nodes collaborate for request processing. For faster processing, these edge nodes are used in many applications namely, smart homes, smart farming, healthcare and so forth. In this paper, we have discussed the use of edge nodes in smart home applications. The smart home application contains different types of sensors and these sensors generate various types of data. Edge nodes are used in these applications for the immediate processing of data. A data classifier is used to classify the data and to reduce delay in data processing. However, the data classifier is more susceptible to DDoS attacks. Hence, an efficient attack detection mechanism is required to detect DDoS attacks. We have used a Feature Selection SVM (FSSVM) algorithm to select optimal parameters for attack recognition. In this algorithm, the information gain ratio is used for optimal parameter selection, and SVM is used for classification. The FSSVM algorithm has been compared with KPCA‐SVM, SVM, and Naive Bayes. Simulation results show that the FSSVM algorithm provides better accuracy compared to KPCA‐SVM, SVM, and Naive Bayes algorithms.