Abstract: In High Level Architecture (HLA) based distributed simulation, a federation is a set of federates, where each federate is normally executed sequentially. Our recent work on HLA-based distributed supply chain simulation shows that a federate with well above-par workload will become the performance bottleneck of the entire federation, since faster federates have to wait for the slower one to advance its simulation time. In this paper, we propose a parallel federate architecture which exploits the internal parallelism of a single federate. The parallel federate architecture is an integration of a parallel simulation protocol and HLA-based distributed simulation. A parallel federate can be executed on a SMP system to improve the performance of the entire federation. Our approach gains the advantages of both the interoperability, reusability and scalability of the HLA and faster execution of the parallel simulation protocol.