Anomaly detection in Smart Traffic Light system using blockchain: Securing through proof of stake and machine learning

Abstract

<p>The Smart Traffic Light system plays a crucial role in smart cities, contributing significantly to enhancing the overall urban living standards, supporting sustainable practices, assuring public safety, and optimizing operational efficiency. Smart lighting systems leverage advanced technology such as the Internet of Things (IoT), data analytics, and networking to construct a complex and flexible lighting infrastructure. Anomalies within the context of a Smart Light system’s data pertain to patterns, events, or behaviors that are unexpected or uncommon, displaying a considerable deviation from the system’s typical operational state. The distributed and tamper-resistant ledger of blockchain technology renders it well-suited for maintaining an open and immutable record of data and events pertaining to smart lighting systems. The integration of blockchain technology and anomaly detection techniques enables the establishment of a resilient and reliable smart lighting system. The primary objective of this study is to investigate the application of the Isolation Forest algorithm for anomaly detection and its enhancement through the implementation of Proof of Stake for enhanced security measures. The Isolation Forest algorithm is utilized for anomaly prediction and evaluation of accuracy and precision. This is achieved by employing synthetic ground truth labels. Subsequently, the non-anomalous data is incorporated into the system as blocks using blockchain technology. The experimentation involves the use of synthetic numerical data derived from a dataset accessible on Kaggle. This process is conducted in two distinct phases. The first phase focuses on the analysis of synthetic numerical data, specifically targeting the identification of anomalies, prior to the implementation of blockchain technology. In the second phase, the same technique is applied to the synthetic numerical data following the integration of blockchain technology. Various machine learning (ML) models were utilized to analyze the data, resulting in improved accuracy from 48% to 94%, as evidenced by the validation of the obtained results. The empirical results indicate that the utilization of blockchain technology in data processing leads to an improvement in accuracy. Furthermore, the random forest algorithm exhibits robust performance when integrated with blockchain technology.</p>