AIF/R’s goal is to create value for its stakeholders via establishment of a reliable and resilient platform for data sharing and analytics. The platform consists of a set of services and governing structures, which enables railway stakeholders, nationally and internationally, to provide and consume data and services securely.
The ongoing digitalisation and implementation of AI-technologies in railway is highly dependent on availability and accessibility of data for a geographically distributed system. AIF/R is facilitating this by providing a platform for data sharing. AIF/R is a set of smart cloud/edge-based data services that are aimed to accelerate digitalisation in railway. AIF/R’s services provide capabilities such as acquisition, inte-gration, transformation, and processing of railway related data across endpoints, e.g. authorities, industries, academia, and SME:s. AIF/R’s integrated services can be invoked on-premises or in multiple cloud-based environments. AIF/R architecture is built on loosely coupled storage and computing services, see Figure 1.
AIF/R provides digital pipelines between data providers and data consumers, as illustrated in Figure 1. Each pipeline represents a set of orchestrated activities aimed to extract, transfer, load, and process datasets between the provider and the consumer. AIF/R’s pipelines are configurable entities, which can utilise a palette of technologies for e.g. communication, storage, and processing, to enable context-adaptability and fulfil the users’ requirements. Selection of appro-priate technologies for each pipeline will be based on the context specific requirements such as requirements on scalability, authenti-cation, and authorisation. It is believed that a generic data factory for railway should be hosted as a neutral open platform, which is governed by a body with focus on research and innovation.
AIF/R will provide railway stakeholders, nationally and internally, with cloud/edgebased services and supporting governance processes that can be used to access various type of datasets. AIF/R will also provide a cyber-security platform including authentication mecha-nisms at various levels. Cyber security platform of AIF/R will man-age access to data based on data openness characteristic, such as 1) fully open data (requires no authentication); semi-open data (requires weak authentication), and noopen data (strong authentication). Furthermore, the data services in AIF/R provide various Applica-tion Programming interfaces (API) to facilitate dynamic and context based integration between back-ends and front-ends services. Some examples of these are Logic Apps, Services Bus, Event Grid, REST etc.
AIF/R will also provide computing capabilities for AI, beside capa-bilities for storage, networking, and communication. These capabili-ties are materialised in a set of interconnected and loosely coupled services, which be orchestrated dynamically to adapt to various contexts. The overarching architecture of AIF/R is illustrated in Figure 2. AIF/R’s services will fulfil stakeholders’ requirements on data access and on knowledge discovery. Furthermore, AIF/R will enable integration with other AI environments such as ‘AI Innova-tion Sweden’ and initiatives within existing ‘H2020’framework and the future framework of ‘Horizon Europe’.
AIF/R will artefact, which relies on four main pillars:
- A technology platform (AI Factory) – AIF/R’s technology plat-form is a service-oriented scalable environment for enabling capa-bilities for information logistics, including services data acquisition, data filtering, data quality, data transformation, cyber security, data processing, and visualisation.
- A digital governance platform (eGovernance) – AIF/R’s eGovern-ance platform deals with all activities necessary to enable govern-ance of digital assets in a heterogeneous technology environment with a large number of stakeholder.
- A communication platform – AIF/R’s communication platform will support and facilitate the achievement of the overall goals. Communications will be built on scientific results, which will be reviewed. A major outlet for communication is the research com-munity, in which articles in peer-reviewed journals will be of high importance.
- A coordinating platform – AIFF/R’s coordnitating platform is built upon an agile mind-set with related tools, to facilitate col-laboration between stakeholders and also to increase the efficiency of AIF/R’s DevOps (development and Operations) processes
Researchers: Ramin Karim (PL), Miguel Castano, Veronica Jägare, Jaya Kumari, Yang Zhu, Amit Patwardhan, Cecilia Glover, Kevin Karim, Emil Lindh, Laila Moussallik
Sponsor: Vinnova, JVTC, Trafikverket, Association of Swedish Train Operating Companies, Infranord, Norrtåg, Alstom, Bombardier, Damill, Omicold, Sweco, TransitioAI