RTIFQLD: An Integrated Framework for RealTime IoT Forensic Analysis via Incremental QLearning Modelled with Resource Aware DQN Operations
Gangavarapu Rajesh Babu, Virendra Kumar Sharma · 2023
As the proliferation of Internet of Things (IoT) devices permeates every facet of human life, the need for effective and efficient forensic analysis within this intricate ecosystem has become more critical than ever. The imperative to swiftly and accurately identify anomalous behaviors or security incidents carries substantial ramifications not only for cybersecurity but also for public safety and critical infrastructure protection operations. Current forensic analysis methods for IoT often suffer from issues related to scalability, adaptability, and real-time analysis capabilities. Traditional machine learning techniques frequently require batch processing, making them unsuitable for real-time deployment. Moreover, these techniques often lack the ability to adapt to the dynamically evolving nature of IoT devices and the myriad threats they face, resulting in suboptimal forensic accuracy and responsiveness. This paper presents an integrated framework for real-time IoT forensic analysis that successfully addresses these limitations. We introduce a novel approach combining Incremental Q-Learning operations, Resource-Aware Deep Q-Networks (DQN) enhanced with Bacterial Foraging Optimization, and a robust ensemble of classifiers including Naive Bayes, Support Vector Machines, Logistic Regression, and Multi-Layer Perceptron for forensic analysis. Our contributions yield several advantages, including superior scalability, real-time analysis, and adaptability to emerging threats. Empirical results demonstrate a significant improvement in the precision of forensic event classification by 8.3%, accuracy by 4.5%, recall by 3.5%, and specificity by 4.9% across different scenarios. By bridging the gap between real-time requirements, resource awareness, and advanced machine learning techniques, this framework sets a new standard for IoT forensic analysis, enhancing both its effectiveness and efficiency levels. This work represents a significant stride in the march toward a safer and more secure IoT ecosystem, with tangible improvements in key performance metrics for forensic analysis.