A detection of phishing probes in IOT devices using a hybrid feature fusion based deep belief network HFDBN

Abstract

Increase in the use of internet of things owned devices is one of the reasons for newlineincreased network traffic. Android devices are connected to IoT networks with fewer newlineauthentications. The flexibility of the internet connectivity with user devices created an newlineattractive bridge between the user and the internet. People started entering new newlinewebsites without any authentication and accessing their services. There is a high risk newlineof vulnerable attacks entering the Android mobiles in the form of data. While newlineconnecting the smart devices with publicly available networks many kinds of phishing newlineattacks can enter into the mobile devices and corrupt the existing system. The IoT newlinenodes can communicate with each other automated way and further enable the user newlineto connect with the clustered nodes easily. Due to intelligent virtual interfaces newlineoutperforming IoT networks, IoT networks are adaptive to heterogeneous integrations newlineof devices. Phishing is a slow and resilient attack stacking technique that probes the newlineusers. The proposed model is focused on detecting phishing attacks in the internet of newlinethings enabled devices through a robust algorithm called NWAT-Novel Watch and newlineTrap Algorithm through Predictive mapping, Predictive Validation and Predictive newlineanalysis mechanism is developed. For the test purpose CIC (Canadian Institute of newlinecyber security) dataset is used for creating a robust prediction model. This attack newlinegenerates a resilience corruption work that slowly gathers the credential information newlinefrom the mobiles. The proposed Predictive Analysis Model (PAM) enabled NWAT newlinealgorithm is used to predict the phishing probes in the form of suspicious processes newlinehappening in the IoT networks. The prediction system considers the peer-to-peer newlinecommunication window open for the established communication, the suspicious newlineprocess and its pattern are identified by the new approach. The proposed model is newlinevalidated by finding the prediction accuracy, Precision recalls F1score, error rate, newlineMathew s Correlation Coefficient