Machine Learning for Data Trust Evaluations in Blockchain-Enabled IoT Systems

Recently, there has been a surge of interest surrounding the integration of blockchain with the Internet of Things (IoT), aiming to address IoT’s inherent issues like single points of failure and concerns related to data integrity. However, although blockchain provides decentralization and transparency, it does not guarantee the accuracy and reliability of IoT-generated data. Therefore, additional measures are needed to assess and verify the reliability of IoT data stored on blockchains. In this demonstration, we present a novel approach that employs support vector machine (SVM) models in edge servers and multiple machine learning (ML) models executed by validators for data trust evaluations in blockchain-enabled IoT systems. Our approach introduces a composite trust metric that combines past device reputation on the blockchain with real-time data assessment enabled by SVM models. This composite measure provides a dynamic method for determining the trustworthiness of data at the point of submission. The multiple different ML models used by validators work as a distributed ensemble, leading to improved classification accuracy. This novel approach helps to calculate reputation scores more accurately, increasing the system’s reliability. We illustrate the feasibility of our approach through a description of our prototype implementation.