Wireless Sensor Network based Anomaly Detection using SVM-RFE-MRMR

There has been a meteoric rise in the use of wireless sensor networks (WSNs) in recent years. WSN’s compact size and low cost are enticing many different industries to adopt them for use in a wide range of applications. There are several applications of IoT technology, including environmental monitoring, building security, and precision agriculture. Given that most WSNs are deployed in an unsupervised and potentially hostile environment, security risks are significant. To ensure the privacy of data during transmission from sensors to the base station, a number of methods have been proposed. The work in question is concerned with a crucial part of data and network security: detecting attacks. With the goal of preventing malicious attacks and maintaining the security of wireless sensor networks, anomaly detection has emerged as a significant obstacle. After the input is given, It consist of three phases such as preprocessing, Feature selection and training the model using SVM-RFE-MRMR. The SVMRFE-MRMR methodology is used for classification in the suggested method. While SVM-RFE takes feature-decision correlation into consideration, it does so by neglecting feature-feature relationships, MRMR is able to provide features with little redundancy and maximum relevance. Because of this SVM-RFE-MRMR is used to generate a set of features that is both minimal in redundancy and maximal in relevance. The proposed model produces an accuracy of about 98.6% when compared other models such as SVM and CNN.