Big Data analysis based intrusion detection in WSN with reduced features

Energy and Security remain two of the biggest obstacles that Wireless Sensor Networks (WSNs) must overcome. Protecting WSNs from Denial of Service (DoS) attacks are some of the security challenges associated with WSNs. The Intrusion Detection System (IDS) should guarantee the security of the WSN services. This IDS must be able to recognize as many security risks as it can and be compatible with WSN features. In this paper, intruder node detection is accomplished using various machine learning approaches. Our work focuses on Big Data analysis based attack detection in WSN with the reduced dataset. In this work, we utilized the WSN - DS dataset. To increase classification accuracy and reduce processing complexity, feature selection is done on the dataset and a reduced dataset is created. Flooding, Blackhole, Grayhole, and TDMA attacks are the four forms of DoS attacks that are taken into consideration in this work. The parameters used to assess the attack detection are the training time to build the machine learning model, and the number of Instances that are Correctly- Classified and Incorrectly- Classified. The outcomes demonstrate that Random forest outperforms other classifiers with a high accuracy rate of 98.17% for the reduced dataset. The Bagging classifier takes less time to train the model than Random forest as well as gives an accuracy of 98% for the reduced dataset.