A communication optimization for conservative interactive simulators

We are using an augmented stop-and-wait protocol to support interactive simulation in FreeMMG, our partially decentralized support model for state-persistent and massively distributed real-time strategy games. In this letter we briefly present FreeMMG and then evaluate an optimization to stop-and-wait that gradually reduces update frequencies of some participant simulators when input inactivity is detected. We performed simulations of a FreeMMG network in ns-2 with 500 clients, each client running up to 9 simultaneous simulators, where a single simulator is one subscribed area of the game world. Experiments show a 29,5% reduction in communication between clients when the optimization is enabled