The Trend Toward Common Architectures Peter Swan Director - PDF document

IT 2 EC 2020 The Trend Toward Common Architectures Architecture & Interoperability The Trend Toward Common Architectures Peter Swan Director International Sales, Cambridge MA, USA Abstract — Defense forces around the world are starting to realize the benefits of a common distributed simulation architecture for collective training, whether in the land or air domain. Leaders are frustrated by having to pay for terrains and models multiple times, spending time and effort attempting to federate incompatible training systems, and being unable to guarantee a fair fight between these systems. This paper will describe the different architectural approaches taken to resolve these issues and to deliver collective training on two current programs that MAK is involved in. 1 Introduction 2.2 STE CSE System Architecture The benefits of distributed simulation have been well known, even since the early days, epitomized by the The STE vision is to develop and deliver a Common Defense Advanced Research Projects Agency (DARPA), Synthetic Environment and common One World Terrain SIMNET (SIMulation NETworking) Program [1]. Those underpinning the virtual simulators, semi-automated early systems struggled with high latencies and a lack of forces and rendering engines across all future US Army bandwidth. It was not feasible to transmit large quantities collective trainers. This next generation collective training of data around – just the bare minimum needed for software has the following attributes: distributed simulations to interoperate. However, the  advent of fiber networks providing 200+ Mbs into our A Computer Generated Forces (CGF) homes, and the availability of pervasive cloud application and a multi-role virtual simulator environments, is making simulation-based training application, built on a common unified available at the point of need, as well as enabling simulation engine running in the cloud and on distributed, collective training. edge computing systems  A multi-role virtual simulator, Image Generator This presentation will describe two specific programs and (IG), scenario authoring tool, Instructor Operator how they are leveraging modern networking and cloud Station (IOS), and After-Action Review (AAR) environments to deliver training. suite, built on a common unified visualization engine that renders 3D scenes in realistic detail  2 The US Army Synthetic Training A common unified terrain engine embedded in all of the components above, allowing them Environment to natively operate on whole-earth WGS-84 terrains  2.1 STE Overview A whole-earth terrain server that can host global terrains at the required fidelity, and serve terrain information through open industry The US Army initiated the Synthetic Training standards to the various STE components Environment effort in 2016 with the mission of providing  a cognitive, collective, multi-echelon training and mission An open systems architecture with a suite of documented APIs, SDKs, and editors to allow rehearsal capability for the Operational, Institutional and Self-Development training domains; of converging the users to customize, extend, and adapt any elements of these systems, and to directly virtual, constructive and gaming training environments into a single Synthetic Training Environment (STE) for integrate new simulation models into both the CGF and the virtual simulators Active and Reserve Components as well as civilians; and  to provide training services to ground, dismounted and A truly modular architecture that supports aerial platforms and command post (CP) points-of-need scalability by combining multiple instances of (PoN) [2]. After evaluating initial prototypes, VT MAK the simulation engine to share the simulation was selected to deliver the Training Simulation Software load and communicate with each other through a (TSS) and Training Management Tool (TMT) components common, network-protocol-independent of the STE Common Synthetic Environment (CSE). interface  A hardware-independent interface that supports a variety of input and output devices, including virtual reality (VR) and augmented

IT 2 EC 2020 The Trend Toward Common Architectures Architecture & Interoperability reality (AR) head-mounted displays (HMDs), constrained largely by the fact that HLA transmits data multi-projector domes, motion platforms, and separately for each entity. immersive cabins and cockpits. For larger entity counts, MAK is developing a scalability architecture we are calling Legion. A centralized Entity 2.3 Hybrid Cloud Model Server maintains a mirror of each Sim Engine’s Data Store. Data stays in the same contiguous layout all the The system architecture is based on a hybrid cloud way from the Sim Engine’s Data Store, to the network, to model, with the flexibility to enable training wherever it the Entity Server’s Data Store . This eliminates expensive is needed, whether at the point of need, at homestation, in marshalling and copying. Network API abstraction a training center, or distributed across many sites. The allows various network protocol choices. The default architecture must support the processing where it is TCP implementation ensures reliability (even over needed for optimal performance. For example: WANs), and allows the efficiency of sending large packets that include many entities’ state .  rendering is done locally on edge computers for Mixed Reality applications using headsets, to When there are too many Sim Engine instances for one avoid delays and jitter caused by the network, Entity Server to handle, the load is shared across multiple but can be done remotely in a cloud for desktop Entity Servers. The Entity Server filters the data against training applications.  interest criteria, builds a large message of just the data Simulation is conducted locally for small local required by a particular client, and sends the message exercises when connectivity is missing, but can directly to the network. run entirely in the cloud for large constructive simulations. A holistic Training Management Tool (TMT) will handle 3 UK DOTC(AIR) the planning, preparation, execution, and assessment of these exercises whether local or distributed, on a LAN or In May 2019, the UK RAF awarded Boeing Defence UK in the cloud. It will use an Intelligent Tutoring/AI system the contract to deliver the Defence Operational Capability to take the training developer through the process of (Air) Core System and Services (DCS&S). The DCS&S developing the complete required Training Support contract will create a capability, known as Gladiator, that Packages; scheduling and allocating resources; will support multiple complex training scenarios, initializing and monitoring the system; executing the simultaneously and independently of each other. The exercise; and performing assessment of individuals, system will allow personnel at different sites to train teams and the training itself. together in their own high-fidelity simulators, linked by a secure network to a new hub at RAF Waddington. The 2.1 Scalability system will securely manage training events across locations and classification levels, allowing RAF crews to At Initial Operating Capability (IOC) the architecture will experience the same battle environment and threats, support up to 80,000 entities for a Brigade level exercise, including in joint training with their counterparts in the with the capacity to support millions of entities United States. representing a city’s population . This will be achieved by utilizing multiple simulation engines running in parallel and taking advantage of: So, rather than a clean sheet, new architecture as envisioned for STE, DCS&S will link together existing  Spatial Organization – responsibility for and new training centers developed by different suppliers simulating entities is shared across multiple such as BAE Systems, Boeing, CAE and Thales. It will need to resolve the correlation and ‘fair fight’ issues simulation engines, each covering a specific geographic area. created when different simulations are linked together.  Ownership transfer - ownership is transferred from one simulation engine to another as the The DOTC(Air) network will be implemented using the entity moves from one region to another. High Level Architecture (HLA) standards specified in the  Load balancing - the size of the region covered NATO Education and Training Network (NETN) by a simulation engine is automatically adjusted federation agreement document. It will provide centralized based on the density of entities in the region. services for instructor control, threat modelling, pattern of  Interest Management – the simulation clients life, terrain, weather, communications and visualization. register interest in entities based on certain criteria, typically only those that they might DCS&S will provide the common Computer Generated interact with. Forces application, Image Generator, and Role Player Stations for UK RAF distributed training exercises based Within a High Level Architecture (HLA) federation, the on MAK’s COTS products, extended to support Tactical simulation engine will support on the order of 50,000 complete and intelligent semi-automated entities,