SMART INTRUSION DETECTION IN INDUSTRIAL DEVICES USING DEEP BELIEF NETWORK

SMART INTRUSION DETECTION IN INDUSTRIAL DEVICES USING DEEP BELIEF NETWORK

Prof. T. P. Deshmukh¹, Sandesh Rayate², Manoj Shinde³, Nachiket Thombre⁴

Assistant Professor¹ BE Students^2,3,4,

Department of Information Technology,

Sandip Institute of Technology and Research Centre, College of Engineering, Nashik - 422213

---------------------------------------------------------------------***---------------------------------------------------------------------

Abstract - Utilization of smart systems everywhere through mobile devices, laptops and home pc are now become flexible. The increase in web usage also increases the web application cyber threats to be happening in most of the third-party connectivity websites. A robust approach on detecting the threats present in the IoT applications are discussed here. In the proposed architecture the collection of number of possible attacks is collected from KAGGLE NIDS dataset. The system detects the similar occurrence of intrusion creating task and triggers the model to prevent through immediate notification. In the existing system IDP-IOT is based on agent technology to support mobility, rigidness, and self-started attributes. Due to IoT limitations, the proposed solution is implemented in the middle, between IoT devices and the router that can be installed in a gateway. In the proposed research work cloud based advanced intrusion detection model is developed. The robust architecture provides the collection of number of possible attacks in the massive internet of things network. The collection of intrusion models we call bags of attacks. The proposed machine learning algorithm creates a robust prediction system for detection of feasible intrusions in the IoT network, the vulnerability of the IoT attacks act as a key for detecting the intrusion present in the network. The proposed design focuses on creating a Novel architecture though Adaptive convolution neural network for improving accuracy and increased security.

Keywords: CNN, Machine learning, Deep learning, Embedded System, Kaggle, IDP-IOT.