D3.2 - Report on Edge AI functional and non- functional requirements

The deliverable D3.2, “Report on Edge AI functional and non-functional requirements” presents a comprehensive exploration/investigation of functional and non-functional requirements (including ethical and legal considerations) of edge AI systems, providing insights into the holistic view of these systems across hardware, software, the AI technology stack, and data applied to AI-based technologies and applications across the micro-, deep-, and meta-edge continuum. The deliverable is currently being extended to include more details and definitions and will be published as a monograph in this year by River Publishers, Denmark. The publication will be the Open Access and available both online and in printing format. The publication is a detailed written study focused on functional and non-functional requirements for edge AI systems to provide an overview of current state of play, concepts, definitions, taxonomy and gaps in the field.
Edge AI systems are characterised by several properties, such as functionality, performance, cost, dependability, and trustworthiness.
This deliverable interlinks the definition of functional and non-functional requirements of edge AI systems with the concepts of edge AI system dependability and trustworthiness.
Dependability is fundamental to the performance of edge AI systems and reflects their operational requirements. In real-time edge AI systems, dependability is the ability to provide services that can be trusted within a time period. The dependability of an edge AI system reflects the degree of trustworthiness in the system. Trustworthiness is the ability of an edge AI system to meet functional and non-functional requirements in a verifiable way, meaning that it can be checked for correctness by a person or tool.
Establishing trustworthiness in edge AI involves collaboration across technical, ethical, and regulatory domains. It is a continuous effort that requires defining the functional and non-functional requirements, their key performance indicators (KPIs), measures, monitoring, ongoing assessment, transparency, and alignment with societal values and needs.