Cooperative Intelligent Systems for Communication Network Management

摘要

This report documents new techniques for building intelligent, cooperative agents for assisting in the management of communication networks. First, we undertook a redesign and upgrade of an existing testbed for a distributed simulation environment. Second, we developed a generalization of previous work in distributed automated reasoning to provide for distributed expert systems built using an extension of conventional production rule technology. Third, we enhanced a distributed reasoning system by utilizing a stronger inference rule known as hyper-resolution. Fourth, we developed a distributed constraint based planner. Our results demonstrate a technique for solving the coordination problem among multiple problem solving agents in a distributed extension of the classical artificial intelligence blocks world domain. We have analyzed real world communication equipment, presented a typical problem scenario, and identified the expert knowledge required to solve the problem. We have experimental data that demonstrate significant improvements in the performance of distributed reasoning tasks. We also have experimental data that illustrate the effectiveness of our distributed constraint based planner. Finally, the problem of link activation scheduling is discussed. We show that our planner is able to solve this problem.