A tactical network optimization engine using simulation

This conference note describes a real-time tactical mobile network optimization agent (TNOA) utilizing real-time simulation. By modeling the wired and wireless links in near real time using the massively parallel, high performance network simulator QUALNET we optimize system performance via configuration control. Our heterogeneous tactical network consists of Link 16, Link 11, S-Tadil-J, and JREAP links which have line-of-sight wireless, UHF SATCOM and generic IP transports. Under a Navy contract we have developed a centralized network monitoring system, Decoupled TDL, which creates a global network status view which supports our optimization. Since some elements of our network do not provide communication telemetry our optimization engine does not have direct information about these network links. For these unobservable nodes we employ some statistical methods to infer network status and node position and gain a complete (distorted) picture of our heterogeneous wireless network. Unfortunately unobservable nodes are typically the most dynamic (e.g. planes). Also we use a legacy wireless system instead of a network where the researchers are free to modify the system. We use only the configuration parameters available in the fielded legacy system. Our evaluation procedure is also unique since we use two independent simulations, our tactical network under optimization (TNUO) which consists of wireless mobile nodes under random waypoint models with heavy bandwidth requirements and our optimization simulation which models the TNUO. We expect significant network performance improvement over a network configured by operators using the processes currently in place.