Deep Learning Model for IOT-Integrated WSN Network for Security Threat Classification

Abstract

Worldwide, industry has seen a radical transformation because of technological advances, i.e., real-time automation systems that employ sensors to gather physical parameter data and transmit that data to remote computers via wireless sensor networks. Even while such technology offers many advantages like higher operational effectiveness, real-time data visualization for better decision-making, cost reduction, etc., it's also subjected to significant data security threats. For this, conventional security methods like cryptography and authentication & access control are also challenging to implement due to the variety of connections and the constraints on compute, energy, and storage capability of industrial equipment. Therefore, in order to safeguard privacy, accessibility, and data integrity when collecting data, transferring, storing, etc., strong security measures must be implemented. Derived from this, the author has assessed the efficacy of different deep  learning (DL) model to verify their validity in terms of how well it work when simulated using a digital threat dataset so that they can be proposed for real-time IoT-integrated WSN data security systems. It has been found that LSTM model for Wireless Sensor Network (WSN) attack categorization performs remarkably well, attaining high recall (98.68%), accuracy (98.68%), and precision (98.61%).