Dynamic Synthetic Environments for Defence A Service Based Approach - - PDF document

IT2EC 2020 IT2EC Extended Abstract Template Presentation/Panel

Dynamic Synthetic Environments for Defence – A Service Based Approach

Andrew Churchward

Principal Engineer – XPI Simulation Ltd, UK

Abstract — The provision of a dynamic and deformable terrain which can update to reflect events occurring within a Synthetic Environment is often lacking in modelling and simulation systems and limited to standalone simulations

• r simplistic changes in terrain state. There is a need for a dynamic terrain representation in distributed simulation

systems, to provide a coherent view of the synthetic natural environment. The author presents work that has been carried out under UK Ministry of Defence research projects investigating an approach to representing a coherent and consistent dynamic and deformable terrain within Synthetic Environments. This approach employs a common data repository (hosted by a standards based open-source geospatial data server) with supporting deformation micro-services to enable a coherent dynamic and deformable terrain representation across multiple simulation clients. The UK is contributing to a NATO Modelling and Simulation Group (MSG) Dynamic Synthetic Environments for Distributed Simulation initiative in order to align its approach with the international community.

1 Introduction and Research Context

A long-term aim of Synthetic Environment (SE) research has been to move towards higher fidelity correlated representations of natural environments that can be delivered to disparate simulation systems. In 2015, the UK Defence Science and Technology Laboratory (Dstl) commenced a 4 year programme of work to investigate the improved representation of

• perational environments in simulation systems. This

programme, the Simulation Composition and Representation of Natural and Physical Environments (SCORE) project, undertook a number of investigations with the aim of: ‘To inform the development of coherent and consistent synthetic representations of the operating environment to provide more effective defence capabilities in support of training, concept development and experimentation, through-life acquisition and evaluation.’ The SCORE programme included a number of work packages that investigated methods and architectures for provisioning dynamic simulation data repositories. This included service-based approaches to storing, updating, and serving datasets to simulation components. This paper describes research conducted by XPI Simulation to investigate a dynamic and deformable terrain which can update to reflect events occurring within a Synthetic Environment. This included prototyping activities that developed a concept demonstrator based on an open-source geospatial data server combined with a number of custom data processing services

2 A Common Knowledge Repository

Simulation systems are typically constructed using a Systems of Subsystems Approach (SOSA) supported by

• ne or more standards describing datasets containing a

vast array of information. Typically each individual simulation system requires several datasets (frequently in proprietary formats) and

• ften there is a great deal of redundancy and overlap in

the information contained within them. This redundancy

• f information and need for proprietary formats leads to

the need for rework across multiple systems which results in increased costs and correlation issues between the various datasets in use. An approach to addressing the need for multiple datasets in simulation systems is to employ a common knowledge repository that hosts a single simulation supporting dataset. The repository is supported by a number of refinement services that are able to, on demand, consume and refine data into a format required by a simulation component. It is intended that, by developing such a repository, the need for multiple redundant datasets in a variety of formats to represent the environment to be modelled with an SE may be reduced. The concept is illustrated in Figure 1 - Common Knowledge Repository

Figure 1 - Common Knowledge Repository

Under the SCORE project, XPI Simulation developed the common knowledge repository concept into a prototype demonstrator based an open-source geospatial data server. The demonstrator included a number of candidate refinement services to refine and serve a tiled dataset including elevation and imagery data.

IT2EC 2020 IT2EC Extended Abstract Template Presentation/Panel From the development of the concept demonstrator, a number of key points were identified that both impose requirements on, and govern the operation of, the system:

• 1. A knowledge repository will start at runtime

with one set of source data, in a common format, and at the highest resolution available;

• 2. Refinement services generate data products from

the source data on demand, and for geographic areas required by consuming federates;

• 3. Refinement services must publish information

describing the data products that they are able to generate to enabling consuming federates to request data products; and

• 4. It is likely that multiple simulation components

will be operating within similar areas of an SE, and as such, there are efficiencies in caching the refined data products By delivering a single dataset, the common knowledge repository approach to simulation data delivery has a significant benefit in improving correlation across components within a simulation system.

3 A Dynamic and Deformable Terrain

Following the common knowledge repository study, XPI Simulation conducted research into the provision of a dynamic and deformable terrain within SEs. This focused

• n the development of an approach to delivering a

dynamic dataset that could be updated during runtime to reflect events occurring within the SE. A primary focus of this research was investigating how the architecture developed for the common knowledge repository might enable a dynamic data system in which a dataset stored within the common knowledge repository could be updated in real-time. From the investigation, several design considerations were identified that were considered key to allowing such a system to react to SE events in a timely manner, whilst also preventing fair-fight issues. These were:

• 1. All updates to the terrain data set (to reflect

deformations in the terrain) should be made to a single reference dataset which is then served to all simulation components. This ensures all components operating within an SE share a common view of the terrain;

• 2. When deforming a terrain dataset, the smallest

possible area of the terrain dataset should be modified to ensure that data processing is kept to a minimum. This approach aids in providing the newly deformed terrain to simulation components in a timely manner;

• 3. In the event of a deformation, simulation

components should request the newly deformed data in a ‘pull’ manner as the component may be some distance away from a deformation rendering the new data unimportant; and

• 4. In the event of a data update, a notification

should be published to all components within the SE, enabling them to request the newly deformed areas of terrain. This notification should contain the type of data modified (i.e. elevation or imagery etc.) and the geographic bounds of the deformation; It was expected that the common knowledge repository approach (a single tiled high-resolution dataset with supporting refinement micro-services) would facilitate a common dynamic and deformable terrain by addressing some of the key considerations listed above. For example, storing a tiled terrain dataset in a common knowledge repository would allow modifications to the data to be performed centrally. The modified (deformed) data would then be refined on demand and served to simulation components, providing a correlated view of the terrain. In employing a tiled dataset, the common knowledge repository supports updating the least amount of data possible, keeping processing to a minimum. A prototype dynamic and deformable terrain system was developed by XPI based on the common knowledge

• repository. This expanded the previously developed

prototype system through the development of a number

• f deformation services, each of which updated the

repository data to reflect events occurring within the SE. Each deformation service specialised in

• ne

dimension of the terrain dataset (i.e. elevation or imagery etc.) and monitored the SE for events (such as a detonation) that would affect that data dimension. On such an event, the relevant deformation micro-service retrieved the tiles of terrain data affected by the event from the common knowledge repository. These tiles were then modified (according to the event within the SE), and stored back in the common knowledge repository as deformed data. On the event of a deformation, newly deformed data was uploaded to the repository, and a notification of a terrain data update was issued to simulation clients. This notification contained the type of data modified along with the geographic bounds of the deformation. The geospatial data server employed by the common knowledge repository architecture supported storage of a dataset in a layered structure. Within this layered structure, new deformed data tiles could be ‘overlaid’ on top of the original data tiles within the database. A request for data to server would result in the layers being temporarily ‘flattened’ resulting in the retrieved data containing a mixture of original and deformed data. This is illustrated in Figure 2 - Deformable Terrain Layered Structure. The layered approach allowed the data refinement services to retrieve the original source and newly deformed data tiles as a single data layer. Thus the resulting refined data provided to simulation components reflected the deformed areas.

IT2EC 2020 IT2EC Extended Abstract Template Presentation/Panel

Figure 2 - Deformable Terrain Layered Structure

By storing each of the individual deformed data areas as a layer within server, the result of each of the deformation events can be viewed individually. Additionally, the terrain database can be ‘stepped- through’ in an incremental manner showing each deformation in turn. This capability is ideal for supporting activities such as after-action review, and allows the data set to be ‘rolled back’ to a non-deformed state by deleting all but the original layer. A demonstration of the prototype system was given during which detonations events were published into the SE resulting in deformation of terrain elevation and imagery data. The result is shown in Figure 3, where the

• riginal terrain, along with the results of both a first and

second detonation can be seen.

Figure 3 - Dynamic Terrain Munitions Effects

4 Conclusions and Lesson Identified

The following conclusions and lessons learned have been identified by conducting this study:

• 1. A common knowledge repository storing a tiled

high-resolution dataset, with accompanying refinement services can improve correlation across components within a simulation system;

• 2. Refinement services operating on-demand must

publish information describing the data products that they are able to generate in order to allow consuming federates to request data products;

• 3. It is likely that multiple simulation components

will operate within similar SE areas, and there are efficiencies in caching refined data products

• 4. A tiled dataset is beneficial to a deformed terrain

system in that it reduces data processing in allowing the smallest possible area of the dataset to be processed; and

• 5. A layered database could support activities such

as after-action review, and allows the data set to be reverted easily to a non-deformed state.

5 ACKNOWLEDGEMENTS

The SCORE project was funded by the UK MOD Chief Scientific Advisors research programme which is delivered by the Defence Science and Technology Laboratory (Dstl).

AUTHOR/SPEAKER BIOGRAPHIES

Andrew Churchward is a Principal Engineer at XPI Simulation and has over 15 years’ experience of delivering Modelling and Simulation solutions for a range of applications from unmanned air vehicle trials using live and synthetic flying platforms, to synthetic mission training systems, to researching improved real- world effects within synthetic environments.