Testbed 14: The Most Complex OGC Testbed Ever?
On 10-12 April, 2018, the OGC kicked-off its most complex testbed ever (Testbed-14, or T14 for short), at least that’s what I think - although I am still in debate about this fact with our CTO, George Percivall. In any case, I think we have an exciting eight months ahead of us.
These annual testbeds are OGC’s main research activity, conducted under the umbrella of the OGC Innovation Program. Every year, OGC works closely with sponsors to first define a number of requirements, embed them in an architecture, and begin - together with many other representatives - exploring new digital territory. According to last count, this year has reps from 88 organizations.
I'm a technologist, but before jumping to the overview of work items, let me emphasize one aspect of these testbeds. The testbeds bring together a large crowd of experts working together - and I really mean together - on exploring new technologies, developing solutions to known issues, and exchanging ideas about future developments. The level of collaboration, the willingness to do research together, and the level of engagement of everyone is staggering, even though we can only compensate a fraction of the costs for each participant (and I can compare this to over 15 R&D projects funded by the European Commission I participated in over the last two decades).
If you have never participated in an OGC Testbed, I invite you to join us for next year’s Testbed 15. If you’re already part of an OGC member organization, and you want to participate immediately, you can do so as an Observer. If not, contact sserich [at] opengeospatial.org (Scott Serich), Director Innovation Programs.
Participants at the Testbed 14 Kickoff, April 2018.
But enough about my very personal view: let’s get down to technology. Testbed-14 runs four threads: EOC, which deals with tons of satellite data that shall be accessed and processed, sort of ‘Big Data in Clouds’ at its best; NextGen, which explores the new RESTful approaches to features and other data, all within workflows, all properly secured; MoPoQ, which addresses machine learning, modeling, portrayal, and quality of service; and CITE, a compliance thread that among others, develops compliance tests for secured clients.
The Earth Observation Cloud (EOC) thread in Testbed-13 laid out the foundations that T14 will build upon. We will continue the development of the application package, which allows standardized ad hoc deployment and execution of Big Data processing applications in cloud environments. The applications in T14 include multiple processing steps executed on data from multiple satellite missions with data stored in several clouds. Quoting and billing features will pave the way for future commercialization of this standards-based cloud processing. Further on, the EOC thread looks into swathe data and the provisioning of the low level data as coverages at WCS interfaces.
All components need to be secured in the NextGen thread. Security will be built on OAuth2 and OpenID Connect. Knowing that there are tons of services out there featuring older technologies such as SAML or x509, we will take a closer look at multi-security environment integration. The secure, on-demand distributed data sharing and collaboration environment aspect will be taken up in the Federated Cloud task: Imagine a user sits at one screen and can seamlessly access data from multiple clouds based on data discovery and access policies that are jointly defined by members within a virtual organization. The NextGen WFS3.0 task will test and verify the latest draft WFS 3.0 core standard available on Github. It will be tested with native WFS 3.0 implementations as well as a facade on top of existing WFS 2.0 services. Overall, the new RESTful approach simplifies access to feature data and makes it much easier to implement a WFS, which now plays nicely with popular developer tools. Next to the actual tests, we will look into complex feature handling with WFS 3.0, that is, 3D data or complex queries. Current WFS 3.0 only implements the most common queries, which is bounding box, time, and simple property values.
There is good news for those of you working with WPS: T14 will develop the transactional extension that has been continuously requested for some time now. Testbed 14 will develop this WPS-T as an extension to WPS 2.0. WPS instances will be used in workflows that are driven by BPMN. For that reason, T14 will work out the OGC Best Practices for using BPMN to orchestrate OGC services. The goal is to give a consistent user experience across workflow engines with security enabled.
Further on, NextGen will explore feasibility and efficiency of streaming and visualizing CityGML data as part of an Augmented Reality solution.
The MoPoQ thread will look into a variety of tasks. It starts with Machine Learning, where we will explore how machine learning tools can be integrated into OGC service architectures to allow for standards-based access to results, learning cycles, and other feedback loops. It continues with improvements of service search and discovery in the aviation domain. Instead of exploring services that may satisfy some search criteria but eventually don't serve the user’s needs, semantic annotation mechanisms will be explored to help discover the right services directly.
Quality of Service and Experience (QoSE) will be further analysed using a number of WMS services. Here, focus is less on discovery and more on QoSE in general, including its definition, communication, and related best practices. A QoSE test framework will be developed to analyse existing services. Visualization of geodata as maps will be further addressed in the MapML and portrayal tasks. With the long-term goal to have native browser support for MapML, T14 looks into new approaches for vector geometries, properties, styles, CRS negotiations, and links between features, as well as temporal aspects of map data. In the portrayal tasks, we further investigate the portrayal ontology, review how to modernize the styled layer descriptor (SLD) and symbology encoding (SE) specifications, and respond to major findings of the recently released Portrayal Concept Development Study.
Application schema modeling and conversion is an ongoing topic present in the previous 10 testbeds at least.This year’s focus is on enhancing the conversion of UML-based application schemas to ontologies, including application-specific subsetting at conversion time. This includes an analysis of serialization handling, including JSON, JSON-LD, and JSON-Schema among others. The goal is to have proper alignment and conversion rules between GML/XML, JSON, and RDF, given that future WFS may produce any of these.
The MoPoQ work is complemented with a LiDAR Point Cloud study, which looks holistically into all elements of LiDAR point cloud data handling.
As part of CITE, we will, for the first time, look into the development of tests for secured clients. Given that security is required for most installations today, having compliant client applications is key. On the service side, we will develop tests and a reference implementation for the Defence Profile of OGC’s Catalogue Service for the Web 2.0.
The final demonstration of Testbed 14 is planned for early December, 2018, and we will certainly have tons of results to be discussed at the December Technical Committee meeting in Charlotte, NC, where we hand over all the results to the OGC Standards Program. If you cannot wait until then and want further information, please contact sserich [at] opengeospatial.org (Scott Serich) for administrative aspects, or isimonis [at] opengeospatial.org (myself) for everything technical.
Dr. Ingo Simonis is Director of Innovation and Science of the OGC. Since Testbed 12, he serves as lead architect for all testbeds.