Secure Multi-Agent Edge Computing Architecture for Autonomous IoT Systems with Blockchain-Based Trust Management

Abstract

Autonomous Internet of Things (IoT) systems, including unmanned aerial vehicles (UAVs), connected autonomous vehicles (CAVs), and intelligent industrial robotics, demand ultra-low latency, decentralized execution, and continuous zero-trust security. Traditional centralized cloud infrastructure introduces unacceptable latency delays, network congestion, and single-point of-failure risks. While Edge Computing mitigates latency by offloading computations closer to IoT data sources, it exposes systems to severe security threats, such as rogue node insertion, distributed denial-of-service (DDoS) attacks, data tampering, and malicious task manipulation. Multi-Agent Systems (MAS) enable localized, distributed decision-making across autonomous devices; however, maintaining reliable dynamic cooperation without a centralized authority remains a fundamental challenge. This paper presents a comprehensive review and structural design of a Secure Multi Agent Edge Computing (SMAEC) architecture integrated with Blockchain Based Trust Management. We systematically investigate the convergence of multi-agent reinforcement learning (MARL), lightweight smart contracts, permissioned distributed ledgers, and zero-trust identity verification. Furthermore, we analyze consensus mechanisms optimized for resource constrained edge gateways, evaluate cryptographic overheads, and present experimental benchmarks illustrating trade-offs between processing throughput, network latency, and cryptographic security. Our analysis demonstrates that lightweight Practical Byzantine Fault Tolerance (PBFT) integrated with smart-contract-based reputation scoring achieves sub-50 ms task execution latency while successfully mitigating 99.4% of malicious collusion attacks.

KEYWORDS: Multi-Agent Systems, Edge Computing, Blockchain, Trust Management, Autonomous IoT, Smart Contracts, Distributed Consensus, Zero-Trust Security.