3 PhD and 1 Postdoc in distributed safety-critical Cyber Physical Systems

University of Technology (TU/e, https://www.tue.nl/en/) is one of Europe’s top technological universities, situated at the heart of a most innovative high-tech region, with a wealth of collaborations with industry and academic institutes. Since 2009, the TU/e is Europe’s best-performing university in terms of research cooperation with industry (according to CWTS Leiden Ranking). TU/e has around 3,000 employees and 2,300 PhD students (half of which international, representing about 70 nationalities).

Project description

The envisioned research is part of the recently granted ECSEL Joint Technology Undertaking (JTU) project TRANSACT, coordinated by Philips. The goal of TRANSACT is to develop a universal distributed solution architecture for the transformation of safety-critical cyber-physical systems from local, stand-alone systems into safe and secure distributed solutions.
To that end TRANSACT will leverage edge and cloud eco-systems to lower CPS’ cost, increase their pace of updates, and improve applications and solution-oriented services.

The Electronic Systems group of the Electrical Engineering department at TU/e calls for applications for three PhD positions and one Postdoctoral position funded in the TRANSACT project. The focus of the work will be on model-based approaches to analyze and manage timing performance in distributed edge/cloud platforms for safety-critical applications.

PhD position 1: Model inference

Model-based methods based on mathematical application models such as scenario-aware dataflow models provide powerful performance analysis and automated management techniques. The TRANSACT project copes with industrial distributed cyber-physical systems for which such models are not readily available. The goal of this project is therefore to infer stochastic models from the logging data obtained from the CPS. These models will be used as a basis in the other PhD projects to develop performance analysis and performance management techniques. The focus will be on concurrent models such as scenario-aware dataflow models, message sequence charts or activity models This project will i) identify the required models-of-computations (MoCs) based on industrial use-cases, ii) formalize them with respect to syntax and semantics, and iii) develop automated inference techniques (possibly based on Machine Learning) to obtain model instances in the different MoCs.

The position is supervised by prof.dr.ir. Jeroen Voeten and dr. Mitra Nasri.

PhD position 2: Performance analysis

​This project concerns the development of performance-analysis techniques, based on the concurrent models obtained in PhD position 1. Focus will be on timing properties and resource usage, for trend and anomaly detection, root-cause analysis, what-if analysis and design-space exploration. We consider safety-critical applications, which may have to deal with significant timing variations, due to the heterogenous distributed nature of the edge/cloud platform. To deal with these timing variations, we will develop techniques that combine worst-case analysis with stochastic analysis. Computationally tractable variants of these techniques will form the basis of the online performance management techniques of PhD position 3.

The position is supervised by dr.ir. Marc Geilen, prof.dr.ir. Jeroen Voeten and prof.dr.ir. Twan Basten.

PhD position 3: Performance management

This project deals with online performance management techniques that prevent the CPS from violating safety or performance requirements. Systems are characterized with models inferred with techniques developed in PhD position 1. The management techniques involve online performance monitoring and prediction techniques (based on the work of PhD position 2), fallback scenario’s replacing high quality, but less certain operations (e.g., high-resolution imaging algorithms deployed in the cloud) by guaranteed but lower-quality services (e.g., low-resolution counterparts deployed in the edge), and techniques to re-deploy applications when resources are congested. The management techniques should ensure predictable and reliable quality of service amidst diverse and partially unreliable resources.

The position is supervised by dr.ir. Marc Geilen, prof.dr.ir. Twan Basten and dr. Mitra Nasri.

​​​​​​​Postdoc position: integral role

The postdoc in this project will have an integral role and will actively collaborate with the three PhD candidates listed above and with external project partners. Goal of this postdoctoral project is to develop a method to guarantee safety and performance in distributed edge/cloud platforms for safety-critical CPS, based on the model inference, performance analysis and performance management techniques developed in the PhD projects. This includes the demonstration and validation of the method on the TRANSACT use cases.