SMARPchain: A Smart Marker Based Reputational Probabilistic Blockchain for Multi-agent Systems

Designed as a distributed data storage system, blockchain employs a consensus mechanism to validate the addition of new data blocks while saving the need for mediation by a centralized server. However, when the blockchain technology is applied in decision-making applications, the consensus mechanism requires the decision to be deterministic, which will ignore different recommendations made by multiple agents and may put the computing system at risk. In this paper, we propose and develop SMARPchain, a paradigm of reputational probabilistic blockchain which is based on smart markers. The distinguishing characteristic of our SMARPchain approach is that it enables multiway branching which allows the result produced by each agent to be kept on the blockchain as a branchchain. Moreover, SMARPchain computes a probabilistic score on each branchchain based on the associated agent’s reputation to help the decision maker reach an informed decision. SMARPchain satisfies three desirable requirements of recognizability, compatibility, and authenticability, and evaluation results show that SMARPchain will enhance the security and robustness of multi-agent decision making systems with acceptable overhead.