Design of a knowledge-based system for a distributed very large area network (VLAN

Distributed problem solving is a subfield of artificial intelligence that deals with the interactions of groups of intelligent agents attempting to cooperate to solve problems. A coarse-grained problem decomposition with multipurpose agents characterizes distributed problem solving in the general subfield of distributed artificial intelligence. A central issue in distributed problem solving is how to distribute control and still maintain coherent behavior. Interaction is structured around coordination as an organizing principle. Coordination is implemented by focusing on what agents share with each other.;The central result of this dissertation is distributed problem solving using a hierarchical model, a Very Large Area Network (VLAN) or satellite network, which is a network including at least one satellite. The focus is on knowledge management--organization of knowledge within each node and with relation to the entire system, distribution of knowledge amongst the nodes, and communication of knowledge between nodes.;The question addressed is: How can we manage the data and control of a knowledge-based system in a VLAN to effectively solve distributed problems? A problem-solving framework in a distributed networking environment is presented. The framework has three major components including a control component that specifies how the interaction occurs between nodes, a communications component that provides the high-level protocols for node interaction, and a knowledge organization component that specifies how knowledge is organized in individual nodes and distributed in the network. The agents are decentralized in that both control and data are logically and geographically distributed. The nodes are grouped physically as well as functionally.;This model has many potential applications including distributed world-wide information systems and space traffic control problems. The VLAN model design lends itself well to handling physically (spatially) large distributed problems. One foremost problem, the missile detection-destruction problem, is presented within the framework. The missile detection-destruction problem is simulated and a critical analysis of the results is presented. Effects of the assumptions, restrictions and simplifications are discussed.