A DEEP LEARNING APPROACH FOR SECURING IOT SYSTEMS WITH CNN-BASED PREDICTION OF WORST-CASE RESPONSE TIME

Abstract

The global industry and AI have promoted the development of the diversified intelligent applications.  Because of large number of devices getting associated and controlled with remote framework, there is a lot massive safety concerns are rising. In order to maximize real-time intrusion detection in the context of the Internet of Things, this article suggests an advanced CNN secure framework. This paper concentrated on the security of web of things, security in term of our IOT gadgets framework usage, for mapping our framework in this study we used a Convolutional Neural Network which help us to characterize our framework. In order to enhance the security and functionality of Internet of Things (IoT) systems, this study proposes a unique framework that blends real-time task modeling with a Convolutional Neural Network (CNN). In a distributed networked control environment, we evaluate system behavior by looking at CPU usage, job scheduling, and execution time. For reliable data transfer, tasks are evaluated in terms of their worst-case execution time (WCET), deadline restrictions, and release jitter. In order to detect any deadline breaches early on, a CNN is used to estimate the worst-case response time (WCRT) of tasks before they are executed. Experimental findings show that the CNN achieved an average prediction error of less than 3% and that all jobs were completed within the deadline. The suggested architecture provides a dependable, clever method to improve task stability and predictability in time-sensitive Internet of Things applications.