D4.1 Evolutive Infrastructure Inventory - Report Nr 1

This first version of the Evolutive Infrastructure Inventory (Report Nr 1, Deliverable D4.1) provides a current inventory of edge AI resources, objects, artefacts, and concepts available in the dAIEDGE consortium. In the future, these resources will be made accessible by dAIEDGE’s “virtual center-of-excellence" and its virtual research labs (VLabs). These facilities aim at accelerating and deepening European research in highly distributed edge AI systems.
As a first gap analysis, the current inventory reveals a large diversity of edge AI hardware and software resources available within the consortium. These resourced comprise more than 20 kinds of software objects and about 10-15 hardware types. While a satisfying response ratio was achieved (66%), certain larger hardware and software partners in dAIEDGE have not yet been able to classify their resources and provide input to the inventory. A hypothesis for this is the complexity within these partners for conducting such an inventory survey at these partners, including many contacts points in the partners and legal issues such as licenses. Our aim is to improve the coverage of the inventory and provide the partners with dedicated support.
The deliverable D4.1 comprises two parts: first, a short document is describing the inventory as it is currently seen, the chosen categories of edge AI objects, and the methods and processes to acquire the data for the inventory. The second part is a database of the inventory and is implemented by an Excel file. The M06 version of the inventory is available as an Excel File (“Evolutive_Infrastructure_ Inventory_M06.xlsx”) upon request.
The classification of the inventory is based on an expected three-layer structure of the “virtual center-of-excellence" and the VLabs. These three layers are:
•an inner core of “edge AI services and objects, e.g. edge AI models, data sets, computational facilities, or edge AI hardware,
•a middle layer of “Management and Operational Services, e.g. authentication and authorization mechanisms or concepts for enforcing societal compliance, and
•an outer layer of “community building services”, e.g. AI marketplaces, software repositories or directories, or testbed federations.
Besides describing the classification, this document:
•details the methods used to acquire the data, i.e., the design of the applied questionnaire and the support to the consortium for answering it, e.g., to verify the validity of the data,
•provides statistics on available resources, e.g., to permit gap analysis and strategy development for the VLab design,
•points to the catalog of resources, i.e., the Excel database, which allows for searching for edge AI resources; thus enabling technical solutions for sharing and integration into a VLab,
•and a lists of contact points at dAIEDGE partners for continuous information collection and fast updates of the inventory as well as for a rapid verification of requirements and technical details.
•provide very first statistics and gap analysis.

This version (Report Nr 1) of the Evolutive Infrastructure Inventory (Deliverable D4.1) supports the internal communication and development activities in the dAIEDGE Network-of-Excellence project and is part of the milestone MS1 at month 06 (M06).
A refined and evolved version of the Evolutive Infrastructure Inventory (Report Nr 2, Deliverable D4.6, M12) will be available in month 12 and as part of milestone MS2. The more detailed version (D4.6) aims at a deeper and more comprehensive description of the requirements, identification of the gaps in resources (Objective O4.1), challenges and capabilities, and internal road mapping of the project for the VLabs.
The deliverable provides input data for task T4.2 (“Virtual research lab design” [M06-M12]), which will perform the actual design and technical implementation of the VLabs.