A Cognitive Fractal and Secure EDGE based on an unique Open-Safe-Reliable-Low Power Hardware Platform Node

Projektinformationen

FRACTAL

ID Finanzhilfevereinbarung: 877056

DOI

10.3030/877056

Projekt abgeschlossen

EK-Unterschriftsdatum 19 Mai 2020

Startdatum 1 September 2020

Enddatum 31 August 2023

Finanziert unter

• INDUSTRIAL LEADERSHIP - Leadership in enabling and industrial technologies - Information and Communication Technologies (ICT)

Gesamtkosten

€ 15 727 138,75

EU-Beitrag

€ 4 708 841,69

Koordiniert durch

IKERLAN S. COOP
Spain

Periodic Reporting for period 3 - FRACTAL (A Cognitive Fractal and Secure EDGE based on an unique Open-Safe-Reliable-Low Power Hardware Platform Node)

Berichtszeitraum: 2022-09-01 bis 2023-08-31

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

The objective of this research activity is to create a reliable computing node that will create a Cognitive Edge under industry standards which becomes a building block of scalable Internet of Things (from Low Computing to High Computing Edge Nodes).

The strategic objectives:

O1: Design and Implement an Open-Safe-Reliable Platform to Build Cognitive Edge Nodes of Variable Complexity

O2: Guarantee extra-functional properties (dependability, security, timeliness and energy-efficiency) of FRACTAL nodes and systems built using FRACTAL nodes (i.e. FRACTAL systems).

O3: Evaluate and validate the analytics approach by means of AI to help the identification of the largest set of working conditions still preserving safe and secure operational behaviors.

O4: To integrate fractal communication and remote management features into FRACTAL nodes.

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

During project period 1, several key developments were done. The foundational specifications of the platform were established and documented in deliverable D2.1. A methodological framework for employing FRACTAL's innovations was also detailed in deliverable D2.2. Adjustments were made to the Hardware layer of the Fractal Node to tailor it for partner-specific requirements. Efforts were made to standardize chosen platforms among partners for smoother integration. A comprehensive analysis of the current state of Network-On-Chip (NoC), security, low power services, and AI service scheduling was conducted. An open-source NoC, named LISNoC, was developed, supporting time-triggered messaging and source-based routing. Additionally, studies into low-power services, Edge AI, and open-source edge applications were conducted.

During the second project period, the FRACTAL Solution's conceptual blueprint was established, encompassing three layers: Edge HW Platform, Edge SW Platform, and Cloud SW Platform. The Edge HW Platform incorporated novel accelerators, safety features, and operating systems, among other components. To enhance safety, security, and power efficiency, various components and fault tolerance techniques were developed. The Edge SW Platform focused on a software framework based on microservices, complemented by a validation methodology. An study about the AI methods and tools to be used for use case applications have been done, starting from the functional and non-functional requirements of the use cases, to define which are the methods and the tools to be employed.

During the third period, development of components were finalized, and the architecture of which they are all part, the FRACTAL System, was refined and updated. This is a Modular Open Reference Architecture. It is an open modular reference architecture that can be configured according to the needs of the application in which the FRACTAL nodes are to be used. This means that depending on the features of the application, the components to be used and integrated in it are selected, following the defined building process. The interactive Fractal System included in the project website shows information about the of the selected components, their characteristics, installation and configuration (https://fractal-project.eu/about/fractal-system/).

Also, during the third period, the implementation of the 8 Uses Cases of the project were finalized, completing the development of the AI algorithms involved and integrating the components needed by each one and validating them.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

The overall result of the project is a Modular Open Reference Architecture configurable to different applications, based on their features following a building process defined (https://fractal-project.eu/about/fractal-system/)

The main achievements reached in the project in each of the fields in which work has been carried out in the project are:

Hardware platforms:​

Making technologies available in a usable platform.​

Putting together technologies that, before, could only be found in different platforms​

Enable novel services on top that did not exist in similar platforms before.

Safety, Security and Low Power Technologies​

Hardware Implementation of FRACTAL adaptive architecture to low-power and safety (Hierarchical Adaptive Time-triggered Multicore Architecture [HATMA]): HATMA is a framework that enables adaptation in a system of FRACTAL nodes while considering the trade-off between safety, security, low-power, environmental scenarios, and user-defined scenarios. It utilizes the capability to switch schedules that are pre-computed based on objective functions that consider multiple scenarios.

In order to enable the HATMA operation, the network on-chip technology is extended. We developed an Adaptive Time-Triggered Network on Chip (ATTNoC) to facilitate the HATMA operation.

AI-Based Metascheduler (Architecture to safely deploy a dependable schedule for safety-critical application, AI-based model to deploy time-critical schedules, run-time framework to enable adaptation in time-critical applications)

AI & Safe, autonomous decisions:​

Theoretical study on FRACTAL AI: AI for edge and AI on edge.​

Cloud Platform for supporting Edge Nodes in resource-intensive tasks (ML training, Models and Data storage and versioning, etc.)​

A novel RODS (Resilient On-demand Distributed Systems) platform for edge orchestration ​

Containerized ML model server for Edge deployments: Training, inference, storage and model management.

Edge SW architecture​

Same open architecture implementation on low, mid & high-end edge nodes​

A complete set of software tools (Data Ingestion, Data Storage, etc.) to orchestrate the Edge Nodes. ​

Validation methodology for edge computing software​

Stand-alone communications subsystem 

The implementation of 8 real world Use Cases, in which different components of the FRACTAL System have been integrated, taking into account their specific features, has served for showing the functioning of the FRACTAL System adapted to each scenario and have validated it.

The FRACTAL Project is a key project in solving computing challenges in the European Union and it has created extensive, accessible known-how for the digital decade. So, these results, as they contribute to the development of Edge computing, bring real-time, high-bandwidth, low-latency access to latency-dependent applications, distributed at the edge of the network.

FRACTAL Cognitive Edge Nodes make products smarter, more interconnected. Interdependent, collaborative and autonomous by bringing computing closer to the devices and users and allowing more efficient use of data and energy, . This improves sustainability in all of its aspects economic, social and environmental.

Big Picture

Updated Big picture