A Distributed Blackboard Approach Based Upon a Boundary Node Concept

Abstract

An architecture for distributed decision making and command / control, which is based on an extrapolation of the Blackboard architectural style, is presented. The proposed architecture utilizes boundary rules as connection / coordination points between multiple blackboards, allowing a distributed knowledge base to span multiple physical systems, while minimizing data duplication and data replication communications. When a boundary node is asserted, this is replicated to all other blackboards containing this node. Network state update information is, thus, limited to updating boundary node status and updates regarding system status (i.e., a system being detected by other systems as offline or back online, etc.). This paper provides an overview of the proposed architecture and its implementation. It describes several possible implementation scenarios for the proposed concept and compares its performance, under a theoretical scenario, to other distributed command / control approaches. Through this work, the efficacy of the proposed technique is demonstrated, from a communications requirement reduction perspective.