Intellig_block: enhancing IoT security with blockchain-based adversarial machine learning protection

Abstract

Internet of things (IoT) systems were becoming increasingly complex due to advancements in open innovation, especially in the realms of intelligent automation and artificial intelligence (AI). However, their effective deployment was impeded by security concerns and the need for enhanced threat detection capabilities. To address these challenges and bolster the security of IoT devices, an architecture called "intellig_block" was developed. This architecture seamlessly integrated blockchain (BC) and AI technology to mitigate vulnerabilities and enhance system efficiency. The goal was to harness the advantages of BC and AI to offer effective solutions for the security challenges confronting IoT systems. The primary focus centered on thwarting contamination and evasion attacks on intrusion detection systems (IDS) powered by machine learning (ML). At that time, many existing solutions relied on traditional statistical frameworks or ML techniques, resulting in increased deployment and runtime costs. In contrast, the "intellig_block" architecture hashed the template file and embedded it as a smart contract to implement the categorization algorithm. The results of the experiments conducted at that time were quite promising: the execution time was short with minimal gas overhead. A potential method was proposed at that time for effectively identifying cyber threats in ML models using the "intellig_block" architecture, which could significantly fortify IDS. Smart contracts (SC) have been introduced as a solution to safeguard IDS results against adversarial machine learning (AML) attacks within BC. In that context, IoT devices leveraged these SC to promptly detect AMLs in real-time data streams. Comprehensive performance analysis and experimental findings at that time substantiated the efficacy of the model in shielding IoT devices against unreliable services, all while maintaining cost-effectiveness within a reasonable time frame and at an affordable cost.