EWMS - Extensible Workflow Management for Simulations

Kurzfassung

The DLR Institute of Flight Guidance is one of the leading establishments in the field of air traffic management research worldwide. Research is conducted using several simulation environments. Simulators like the Tower Simulator, the Approach and Radar Simulator, and different fast-time simulation tools focus on different parts of air traffic and help to examine various aspects of the air traffic environment (e.g. controller view, pilot view, general traffic situation, etc.). The institute's research activities include validation and verification of new systems and procedures, such as airport infrastructure, routes, or assistance systems for air traffic controllers and pilots. Most software systems to operate the simulators are commercial products that concentrate on simulation, and do not emphasise on analysis of simulation outcome. The software environments for the Tower and Radar Simulator were originally designed in order to train air traffic controllers, with little provisions for in-depth data analysis. Other tools do provide detailed output for subsequent analysis, but output content and format varies widely between tools, requiring specific post-processing for each simulation environment. In the past, DLR developers have implemented numerous tools to record and analyse data from individual simulations. These tools were developed independently for different simulators, different simulation campaigns, using several different programming languages. For use in specific campaigns the tools often had to be adapted or re-implemented from scratch. Finally, the underlying algorithms were often not sufficiently tested and documented, and tool validation itself was often a by-product of simulation evaluation. The Extensible Workflow Management for Simulations (EWMS) was developed to support, simplify and enhance the process of simulation analysis and evaluation. The main objectives of the EWMS concept are: Data backup strategy – Considering the effort needed to gain simulation data (up to six months of preparation and several weeks of simulation execution) the recorded data presents a substantial value. Thus, the EWMS employs a client-server approach to offer a consistent backup strategy, using an off-the-shelf data repository system for backup and versioning of project data, to prevent data loss and to ensure adherence to quality management standards for data handling. Transparent post processing – The EWMS was designed to support the scientific evaluation of simulation data by those scientists that run the actual simulation campaigns. These scientists need not be familiar with the technical details of the simulation environment, or with the mathematical and computer science aspects of the evaluation algorithms. The EWMS provides an environment that automates several aspects of the overall workflow and makes them transparent to the user. This includes backup and version control, as well as the transformation of simulation specific output into a generic format for further processing. Standardised reporting – Until now, the different simulation environments and even different campaigns used their own algorithms and methods for data analysis. In some cases, these differed even between users of the same environment. Consequently, comparison of results was often difficult, if not impossible because of variations in the underlying assumptions. The EWMS offers standardised, well documented and peer-reviewed algorithms and methods for analysis and reporting, enabling comparison between different simulation environments, ensuring and enhancing the overall quality of the results. The EWMS environment provides an abstraction layer that separates the data format of a specific environment from the generic reporting capabilities. This abstraction guarantees that all users can perform the same analyses using common algorithms and methods, regardless of the specific simulation environment used. Scientists can focus on data interpretation while keeping the overhead for data manipulation to a minimum. The results can be displayed numerically or as standardised diagrams, and can be exported for easy inclusion in documents. The EWMS is built around a generic core supported by numerous modules that add support for simulation environments or provide new algorithms for analysis and reporting. The core provides generic functionality including the overall user interface, the repository server connection and a built-in SQL database. Those capabilities are made available to the simulation-specific environments and the reporting modules. The simulation-specific code encapsulates the handling of simulation input and output, as well as data transformation into the generic format required by the reporting modules. Reports publish their input requirements and are shown only when the active simulation environment is capable of delivering the required data. This modular approach simplifies distributed software development, integrating developers from different departments that work on their specific areas of expertise. Looking forward, the EWMS development continues with a focus on two major issues. Based on user requests, additional reporting capabilities will be added that will be of benefit for all users of the supported simulators. Also, additional simulation environments will be integrated to widen the user base.