Hyper-Distributed Artificial Intelligence Platform for Network Resources Automation and Management Towards More Efficient Data Processing Applications

Duration: 01/04/2024 – 31/03/2027   |   Total Funding: € 4,668,912.50

Challenge & Objectives

The challenges addressed by the Hyper-AI project stem from the significant advancements in digital information technology since the 1980s, particularly the explosive growth of the internet from a network of personal computers to a vast mobile/cellular network encompassing billions of smartphones and portable electronics. The expansion of the Internet of Things has allowed the connection of billions of objects and smart devices, facilitating seamless integration across various domains.

Despite these developments, challenges arise due to the heterogeneous nature of connected devices in terms of size, weight, data fidelity, processing, sensing, storage, and communication capabilities. Moore’s law predicts exponential growth in device capacity, increasing the potential for deploying highly tactile data-driven applications using distributed resources. However, this also necessitates new network forms and abstractions to effectively utilise and coordinate computational resources across the network continuum.

Digital technologies are expected to accelerate progress toward Sustainable Development Goals, mitigate downward trends, and reduce carbon emissions. The diversity of digital applications enables the replication and transferability of functionalities across different domains. The availability of vast amounts of data, both open/public and closed/confidential, presents an opportunity for dynamic data collection mechanisms and cognitive algorithms, enabling interoperable, situation-aware decision-making and operational autonomous optimisation.

High-fidelity applications in strategic domains such as healthcare, agriculture, mobility, automotive, manufacturing, and green energy demand effective management of networked resources for optimal performance. To address uncertainties and stochasticity, dynamic swarming and pooling approaches inspired by natural paradigms are crucial in computing research domains.

The Hyper-AI project aims to address these challenges by enabling hyper-distributed execution of demanding data processing applications. It introduces the concept of smart virtual computing entities (nodes) from Cloud, Edge, and Internet of Things infrastructures. Hyper-AI focuses on intensive data-processing applications and implements computing swarms as autonomous, self-organised networks of smart nodes. These networks offer diverse resources and can dynamically connect, interact, and cooperate. Hyper-AI proposes semantic representation concepts for heterogeneous resources’ abstraction in a homogeneous way across the entire network infrastructure.

The main orchestration and control concept of HYPER-AI is inspired by autonomic systems which employ swarmed computing schemes. Its objective is to simplify smart multi-node deployment scenarios through AI-driven self-configuration, self-healing, self-optimising, and self-protecting mechanisms at runtime. The project suggests flexible integration of resources at the edge, core cloud, and along big data processing and communication channels for energy, time, and cost-efficient execution of dynamic and data-driven application workflows.

1

Design an open architecture for HYPER-AI, promoting collaborative development and bringing computation closer to the edge with decentralised AI. It aims to deliver a scalable mesh architecture, incorporating a multi-objective optimisation framework.

2

Ensure dynamic cognitive decision-making in HYPER-AI throughout the application lifecycle. Enable continuous incremental learning for perception entities, optimising Quality-of-Service (QoS) and Quality-of-Experience (QoE) by strategically offloading processing operations to edge resources.

3

Develop autonomic system entities for Smart-Nodes’ autonomy and rapid coordination during application design and runtime. HYPER-AI delivers optimisation tools, creating autonomous Smart-Nodes with dynamic pooling of network resources based on real-time situation awareness.

4

Safeguard the openness of HYPER-AI outcomes, promoting social innovation and engaging external stakeholders. Adopting open science practices, the project stimulates an actively engaged international community aligned with EU Open Research principles.

5

Demonstrate HYPER-AI’s usability, performance, and interoperability in various use cases. The project focuses on improving application deployment time, reconfiguration efficiency, and network coverage.

6

Emphasise wide communication, efficient dissemination, and vertical exploitation of HYPER-AI results. Propose novel business models, study market uptake, and investigate disruptive models for incentivised and secure Edge resource sharing, contributing to the European data economy.

eBOS Role in the Project

In the HYPER-AI project, EBOS plays a crucial role in various aspects. eBOS will contribute to managing the registration and lifecycle of nodes within the Computing Continuum, ensuring smooth resource engagement. Additionally, EBOS will contribute in developing open connectors, enabling transparent management and interoperation of computing swarms across diverse hardware and software environments. As a leader, eBOS will integrate technological enablers and architectural modules, ensuring interoperability and modularity to accommodate future service requirements. Finally, eBOS will leverage its expertise gained from developing the EO4EU platform, funded by the European Commission’s Horizon Europe initiative. Drawing from this experience, EBOS will contribute to the front-end development of HYPER-AI, programming platform, facilitating the specification and orchestration of deployment workflows for HYPER-AI applications. This synergy between projects underscores EBOS’s commitment to advancing innovative solutions across diverse domains, enhancing the accessibility and usability of cutting-edge technologies.