A sentinel based exception diagnosis in market based multi-agent systems

Market based Multi-agent systems (MAS) operate in an open and dynamic environment where agents enter and leave the system unpredictably. The dynamism and unpredictable nature of an open environment gives rise to unpredictable exceptions. It becomes essential to have some exception diagnosis mechanisms in these systems to effectively manage their exception diagnose processes. The defining characteristic of mechanisms is their ability to provide a uniform exception diagnosis capability to independently developed agents. In this paper we present a brief discussion of our proposed sentinel based approach to exception diagnosis in an open MAS and discuss the accuracy and performance overhead of our approach when applying in a market based MAS.