Abstract
The ongoing digital transformation in manufacturing, driven by Industry 4.0 (I4.0), integrates intelligent and cooperative Cyber-Physical Systems (CPSs) to enhance productivity, flexibility, and product quality while enabling new business models. This shift is powered by the convergence of advanced communication technologies, cloud and edge computing, and the Industrial Internet of Things (IIoT). In particular, a very relevant and related subfield in network computing environments is next-generation industrial control: while current industrial control mainly relies on statically selected monolithic applications and wired technologies like fieldbus and industrial Ethernet for low-latency and high reliability, I4.0 CPS solutions are expected to employ distributed control functions based on dynamically composable microservices and wireless technologies like private 5G, WiFi7, IEEE 802.15.4e Time Slotted Channel Hopping (TSCH), etc. In this context, JOint ResoUrce Management in ReconfigurabLE I4.0 Factories (JOULE) has focused on investigating, modelling, designing, and assessing novel Quality of Service (QoS)-constrained management mechanisms and strategies for virtualized resources in industrial networked computing environments. JOULE has specifically worked on the dynamic (re)allocation/deployment/(re)configuration of compute, storage, and networking resources under per-application requirements on quality, e.g., time and reliability, addressing the following challenging research directions: 1. Unified abstractions and models for distributed heterogeneous virtualized resources, allowing us to express and map high-level QoS to actionable low-level management decisions; 2. Innovative algorithms and mechanisms for dynamic end-to-end reallocation and reconfiguration of virtualized (containerized) resources able to consider the above requirements; 3. Distributed QoS-constrained orchestration of industrial support/service components that efficiently use compute, storage, and networking virtualized resources; 4. Qualitative assessment and minimal viable demonstration of the proposed mechanisms/strategies in a manufacturing scenario entailing dynamically reconfigurable settings. Achieved Results As planned, JOULE has investigated, developed, and tested innovative solutions in several research sub-fields, with good academic and industrial impact. In particular, the JOULE partners have developed innovative control algorithms and mechanisms enabling dynamic, end-to-end reallocation and reconfiguration of virtualized and containerized compute, storage, and networking resources. These novel mechanisms have demonstrated to be able to react to workload variations and operational policies, optimizing the placement and lifecycle management of services across distributed infrastructures. These developments were closely coupled with the definition of the JOULE architecture, which provided the structural framework and integration principles guiding the design and deployment of the control mechanisms. In parallel, semantic modelling activities supported this process by providing a formalized representation of system entities, relationships, and policies. This enabled consistent interpretation of resources and services across components, facilitated interoperability, and enhanced the capabilities required for (semi)automated decision-making within the JOULE framework. In addition, about innovative distributed orchestration mechanisms running at heterogeneous and virtualized infrastructures spanning the Cloud-to-Things Computing (C2TC), we have designed and developed novel orchestration strategies with ensured predictable performance, efficient resource utilization, and service continuity under dynamic operational conditions. To achieve this, JOULE emphasized data affinity and locality principles, ensuring that computational tasks and data flows are orchestrated in proximity to where data are generated or most intensively used, thereby reducing communication overhead and latency. Furthermore, the orchestration logic incorporated intelligent off-loading and in-loading mechanisms, enabling the dynamic relocation of workloads between edge and cloud domains according to contextual QoS requirements, resource availability, and energy-efficiency objectives. This allowed JOULE solutions to adaptively balance computational pressure while maintaining deterministic performance for time-sensitive industrial applications. Moreover, about prototyping JOULE integrated solutions into a realistic manufacturing scenario, JOULE adopted an agile and iterative approach that made it possible to assess and validate the developed mechanisms first through simulation and emulation environments, enabling the evaluation of their stability, scalability, and responsiveness prior to deployment in real testbed conditions. Indeed, the most promising candidate solutions were integrated and evaluated within a real industrial testbed continuum environment, allowing the assessment of their performance, robustness, and operational viability under realistic deployment conditions. This approach fostered continuous improvement and tight coupling between design, validation, and experimentation, accelerating the maturation of the JOULE framework toward practical applicability. As a final remark, JOULE has achieved the expected set of innovative results in terms of relevant publications, software prototypes available for the community, and an integrated application solution for the manufacturing industry capable of respecting challenging requirements in terms of QoS level agreements. The relevance and international visibility of these results is also shown by the associated record of publications and by the related impact in the research community, as well as by the follow-up actions already planned as future work of collaboration among the participating units.
Project details
Unibo Team Leader: Paolo Bellavista
Unibo involved Department/s:
Dipartimento di Informatica - Scienza e Ingegneria
Coordinator:
ALMA MATER STUDIORUM - Università di Bologna(Italy)
Total Eu Contribution: Euro (EUR) 228.130,00
Total Unibo Contribution: Euro (EUR) 82.104,00
Project Duration in months: 24
Start Date:
28/09/2023
End Date:
28/09/2025
