Remediation of chaos in cognitive Internet of Things sensor network

Abstract

The cognitive Internet of Things (CIoT) is an emerging field that integrates cognitive capabilities into IoT systems, enabling devices to learn, reason, and make autonomous decisions. This advancement enhances the intelligence and adaptability of IoT applications. However, the vast, unpredictable, diversified, and time-dependent nature of data generated by these applications presents significant challenges in data management. Without a cognitively inspired framework, effectively managing this complex data becomes increasingly difficult. The behaviour of the data stream should be cognitively assessed in a number of scenarios to ensure that CIoT applications continue to operate smoothly and exhibit non-chaotic behaviour. In order to address it, this research proposes a novel design to detect the chaotic behaviour of the multisensory data stream and further tries to step it up to correct and return from a chaotic state to a non-chaotic state. In the proposed design, the Lyapunov exponent is computed for the detection of chaotic behaviour in the massive heterogeneous data stream, and if the system is found chaotic, then it designs the three novel algorithms, i.e., total variation (TV) regularization, maximum a posteriori (MAP) estimation, and informative value replacement for chaotic sensor data of the system from returning to non-chaotic state. This is done in a computationally efficient manner so that there is no extra burden posed on the fusion center. The suggested algorithm outperforms competing algorithms (accuracy > 99%) in an experimental evaluation, which is carried out utilizing environmental data spanning 21.25 years and uncovered by cross-validation using several measures.