Abstract
High Level Architecture (HLA) is a distributed simulation architecture used in many fields of computer science. In most simulation environments, the simulation entities (called federates in HLA) are connected by several interconnected local area networks (LANs). Those federates in the same LAN have better locality properties, such as latency, bandwidth and other proximity metrics, than those federates in different LANs. Under this premise, this paper leverages the locality properties to design a publish/subscribe scheme based on the structured peer-to-peer overlay as the messaging foundation of HLA. The contribution of this paper is to increase the affordable federate size or workload size that is hard to be achieved in traditional centralized approaches under the same hardware support. Moreover, we implemented the proposed publish/subscribe scheme in accordance with HLA for performance benchmark and the results show this scheme is feasible and scalable.