Reputation Management in Multi-Agent Systems Using Permissioned Blockchain Technology

Abstract

The multi-agent framework is a well-known approach to realize distributed intelligent systems. Multi-agent systems (MAS) are increasingly employed in safety-and information-critical domains (e.g., eHealth, cyber-physical systems, financial services, and energy market). Therefore, these systems need to be equipped with mechanisms to ensure transparency and the trustworthiness of the behaviors of their components. Trust can be achieved by employing reputation-based mechanisms. Nevertheless, the existing methods are still unable to fully guarantee the desired accountability and transparency. Aligned with the recent trends, advocating the distribution of trust to avoid the risks of having a single point of failure of the system, this work extends existing efforts on combining blockchain technologies (BCT) and MAS. To attain a trusted environment, we provide the architecture and implementation of a system that allows the agents to interact with each other and enables tracking how their reputation changes after every interaction. Agents reputations are computed transparently using smart contracts. Immutable distributed ledger stores reputation values, as well as services and their evaluations to ensure trustworthy interactions between the agents. We also developed a graphical interface to test different scenarios of interactions between the agents. Finally, we summarize and discuss the experience gained and explain the strategic choices when binding MAS and BCT.