Performance Evaluation of Intrusion Detection System for Smart Devices using Neural Network Classifiers

The multiple security threats arising in smart devices cause more computing power and reduce storage capabilities. Authentication and Encryption methods are not adequate to provide enough security for a smart device. An intrusion detection system improves protection and uses few resources on the smart device. An Intrusion Detection processwas proposed to detect intrusions in smart devices in a cloud environment using Data Mining techniques. A Network - based approach is used to eliminate the overhead dispensation from smartphones. A neural network classifieris implemented for all user call logs. A technique that runs on a smartphone of the user collects information from the user and transfers it to the remote server. These call logs are then moved to the already qualified classifier which analyzes. The ogs and sends back the feedback to the smartphones whenever abnormalities are found. The different neural classifiersare used to identify the classifier with better performance using the Weka data mining tool.