Adaptive and real-time planning architecture in IDSP system

We propose an adaptive real-time planning architecture, based on the multi-agent concept, for a service provisioning system. A planning agency, which is composed of heterogeneous agents, is allocated to each terminal and each service provider. It gets an plan and breaks it down hierarchically. We evaluated this architecture in the service provisioning system called IDSP (Intelligent Dynamic Service Provisioning) over an ATM experimental network. We found that this architecture had a stable response time and was effective. In addition, we propose an agent interaction mechanism that solves remaining problems by using a residue-driven architecture.