Supervisors: Dr. Wilfried Steiner (email@example.com) Prof. Dipl.-Ing. Dr. Gerhard Fohler
Deterministic Virtualization for Mixed Criticality Industrial Applications
In Fog Computing, virtualization and time sensitive networking (TSN) are commonly accepted and presumed to act as key enabler of IIoT. TSN provides deterministic networking, whereas virtualization provides scalability, availability, reliability, and security. Network function virtualization (NFV) is a popular example of how virtualization can facilitate novel technologies. In case of NFV, a hypervisor today already enables the consolidation of multiple virtual network functions (VNF) on a commercial off-the-shelf multicore processor, effectively reducing operating expenses (OPEX), capital expenses (CAPEX), and time-to-value (TTV) for telecommunication service providers.
Virtualization could help manufacturing and producing industry to reduce their OPEX, CAPEX, and TTV as well, since hypervisors running on COTS MCPs enable novel use cases, such as control as a service (CaaS), dynamic configuration of production sites, and online optimization of production processes. However, for many use cases virtualization must meet additional, demanding non-functional requirements.
Objectives & Expected Results:
- Hypervisor prototype: Development of a hypervisor prototype that allows for the execution of hard real-time applications inside VMs while allowing dynamic management of best-effort VMs.
- Bounded communication: Development of a bounded communication mechanism for the hypervisor. The communication mechanism can either utilize hardware features such as Intel’s VT-d or be purely software-based thus device emulation that guarantees bounded response times.
- Formal methods: Provision of formal methods to show that the VMs dedicated to real-time applications meet their deadlines given the hypervisor prototype.
- “The Need for Deterministic Virtualization in the Industrial Internet of Things” — Jan Ruh, Wilfried Steiner, IoT-Fog ’19 Proceedings of the Workshop on Fog Computing