Any federated simulation environment or synthetic training environment will be only as good as the network that interconnects their various components. A poor performing network will limit the ability for real-time or near real-time distributed simulation and will limit the ability to deliver an effective synthetic training experience. Low-capacity networks can unacceptably constrain the types and amounts of data that can be shared across the environment. Low performing networks can lead to unacceptable latency that will translate to a poor user experience. The network must be able to provide both the capacity and performance to enable these key simulation and training paradigms. This becomes even more true with the emergence of performance-sensitive technologies such as Augmented Reality / Virtual Reality (AR/VR).

This paper presents the High-Performance Synthetic Training Environment (STE) Optimization Network (HP-SON), a capability currently being developed by Sabre Systems, Inc. for the US Army Futures Command Synthetic Training Environment (STE) Cross Functional Team (CFT). Born from the desire to inter-connect a wide range of geographically distributed simulation and training elements across a variety of network types, HP-SON incorporates a suite of artificial intelligence / machine learning (AI/ML) algorithms to optimize the network to enable the Army’s full large-scale, distributed vision of the STE. HP-SON’s AI/ML employs state-of-the-art AI/ML algorithms to learn and predict network performance and user activity, resulting in a highly-reliable set of predictive network analytics. These predictive network analytics are then used to drive several predictive AI/ML-driven features, including data compression, traffic scheduling, protocol optimization, and forward data caching. Together, these capabilities dramatically improve perceived network performance and capacity and produces an improved user experience.

Keywords
DISTRIBUTED, MACHINE LEARNING, NETWORKS, SYNTHETIC ENVIRONMENT

Additional Keywords
Artificial Intelligence