A Blockchain-Based Approach to Trust System Design in Multi-Agent Environments

Introduction. This paper presents a decentralized approach to building a trust system for distributed multi-agent environments using blockchain technology. The developed system, TrustLedger, is based on Ethereum smart contracts, ERC-20 tokens, and token-weighted voting mechanisms. This approach enables transparent, secure, and scalable interactions between agents, even in scenarios with partial distrust and without centralized authority. TrustLedger includes mechanisms for agent registration, role and task creation, proposal submission, voting, token staking, and the automatic distribution of rewards and penalties. Agent reputation is built through honest participation in the system, while the influence of dishonest participants decreases as their voting power is gradually reduced. A simulation of an energy distribution scenario demonstrated the system’s ability to effectively identify and isolate malicious agents while maintaining high levels of internal trust. In contrast to centralized trust models, which are vulnerable to failures and abuse, the proposed solution provides resilience to network dynamics, latency, communication disruptions, and adversarial behavior. The advantages of smart contracts – automation, transparency, and immutability – enable “trust by design,” reducing dependence on human intervention or centralized intermediaries. The described system is a promising tool for use in smart grids, IoT environments, and collaborative robotics, where autonomous agents must interact without prior trust. The purpose of the paper. To develop and evaluate the decentralized trust system TrustLedger for multi-agent distributed environments using blockchain technology. Results. A prototype of the TrustLedger system was implemented, demonstrating the ability to isolate dishonest agents and maintain consensus stability. Simulation results showed a consistent reduction in the influence of malicious agents over several rounds of interaction. Agents that regularly acted against collective interest lost their reputation and voting weight, confirming the effectiveness of the incentive and penalty mechanisms. Conclusions. The TrustLedger system has proven to be an effective decentralized solution for managing trust in multi-agent systems. By utilizing smart contracts and token-weighted voting, it enables transparency, security, and self-regulation without the need for centralized control. This approach provides a resilient environment for agent interaction, even in dynamic and potentially adversarial settings. Keywords: multi-agent systems, distributed systems, TrustLedger, blockchain, trust system, smart contracts, ERC-20, decentralized governance, agent reputation, voting.