Modeling cooperative agents for customer network control using planning and agent-oriented programming

A testbed, called TEAM-CPS, that explores distributed cooperative architectures for customer network control is described. TEAM-CPS is used to model customer and public network operations support systems as intelligent knowledge-based agents that cooperate as do teams of human experts working together on common problems. An earlier proof-of-concept approach is generalized by the addition of a new and more practical framework for building autonomous agents. This framework is defined by the layered integration of two artificial intelligence (AI) paradigms: agent-oriented programming and classical AI planning. The application domain and this new framework are described, along with general AI techniques developed for TEAM-CPS. A simplified customer network management problem is used throughout, and an annotated trace of a three-agent dialogue, produced in TEAM-CPS, exemplifies the type of cooperative behavior and problem-solving performance that can be expected with current prototypes of TEAM-CPS