Engineering large-scale agent-based systems with consensus

摘要

The paper presents the consensus method for the development of large-scale agent-based systems. Systems can be developed as networks of knowledge based agents (KBA) which engage in a collaborative problem solving effort. The method provides a comprehensive and integrated approach to the development of this type of system. This includes a systematic analysis of user requirements as well as a structured approach to generating a system design which exhibits the desired functionality. There is a direct correspondence between system requirements and design components. The benefits of this approach are that requirements are traceable into design components and code thus facilitating verification. The use of the consensus method with two major test applications showed it to be successful and also provided valuable insight into problems typically associated with the development of large systems.