Multi-Method and Multi-Agent Planning

omputers and communications networks transport data from multiple locations to multiple locations. Network configurations address Quality-of-Service issues (e.g., component reliability, equipment costs, data buffer sizes, bandwidth limitations, and routing protocols). Planning for and implementing network equipment, software, and security mechanisms costs hundreds of millions of dollars per year. The rapid growth in number of users and equipment choices multiplies the network configuration complexity. This article describes our Air Force-sponsored system to accelerate network configuration by an order-of-magnitude using multi-agent planning techniques and thereby dramatically reducing network configuration costs. We examine how the system cohesively integrates planning, simulation and data/knowledge entry.