A Hadoop Based Framework Integrating Machine Learning Classifiers for Anomaly Detection in the Internet of Things

In recent years, different variants of the botnet are targeting government, private organizations and there is a crucial need to develop robust framework for securing IoT (Internet Things) network. this paper, Hadoop based proposed identify malicious traffic using modified Tomek-link under-sampling integrated with automated Hyper-parameter tuning machine learning classifiers. The novelty paper utilize big data platform benchmark datasets minimize computational time. loaded in Distributed File System (HDFS) environment. Three approaches namely naive Bayes (NB), K-nearest neighbor (KNN), support vector (SVM) used categorizing traffic. Artificial immune network optimization deployed during cross-validation obtain best classifier parameters. Experimental analysis performed on platform. average accuracy 99% 90% obtained BoT_IoT ToN_IoT datasets. difference ToN-IoT dataset due huge number samples captured at edge layer fog layer. However, BoT-IoT only 5% training test from complete considered experimental as released by developers. overall improved 19% comparison state-of-the-art techniques. times reduced 3–4 hours through Map Reduce HDFS.