Towards Secure Coordination in Multiagent Systems: Privacy-Preserving Techniques

Multi-agent systems (MAS) are positioned among the major technical advances of distributed artificial intelligence, whereby independent agents can cooperate and coordinate to meet complex and evolving tasks. But the privacy problem is also of considerable importance, especially in situations where sensitive information is to be subjected to sharing because of the given decentralization and heterogeneity of the MAS in question. The paper is a critical study of the literature on the privacy-preserving techniques (PPTs) to ensure safe and secure coordination of the designs of the MAS. The most significant of these, including the technique of differential privacy, the secure multiparty computation, homomorphic encryption, federated learning and blockchain integration, are also dealt with critically from the point of view of their suitability in the different fields treated respectively by the way of example. These include health care, smart grids, autonomous vehicles and swarm robots. The review sense of depth will be found from the elaboration of the many definitions of privacy for the purposes of MAS design and implementation. The paper will also attempt to identify gaps in the literature and suggest future lines of research in terms of how it relates to the MAs designed.