3 PhD's in Model-Driven Design of Distributed Video Processing Systems for CPS (V36.3159)

3 PhD's in Model-Driven Design of Distributed Video Processing Systems for CPS

Aanstellingsfunctie(s)
Promovendus
Faculteiten
Faculteit Electrical Engineering
FTE
1,0
Sluitdatum
18/02/2018
Aanvraagnummer
V36.3159

Functieomschrijving

Eindhoven 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. In 2017, TU/e was ranked 15th in Europe in the Times Higher Educational World University ranking for Engineering and Technology. TU/e has around 3,000 employees and 2,300 PhD students (half of which international, representing about 70 nationalities). 

Electronic Systems group at the TU/e
The Electronic Systems (ES) group consists of seven full professors, one associate professor, eight assistant professors, several postdocs, about 30 PDEng and PhD candidates and support staff. The ES group is world-renowned for its design automation and embedded systems research. It is our ambition to provide a scientific basis for design trajectories of electronic systems, ranging from digital circuits to cyber-physical systems. The group is strongly involved in the electrical engineering bachelor and master programs of the TU/e, as well as in the automotive bachelor program and the embedded systems master program. The group has excellent infrastructure that includes individual computers, computer servers, state-of-the-art FPGA and GPU farms, sensor- and ad-hoc networking equipment, a cyber-physical systems lab, an electronics lab and a comprehensive range of electronic-design software. ES has strong collaborations with industry, research institutes and other universities. Eleven of its staff members have a second affiliation besides their TUE-ES affiliation. The ES group has been very successful in attracting funding for its research through national and international projects and collaborations (EU programs: H2020, ITEA, CATRENE, ECSEL, Artemis, Marie Curie; national programs: NWO, RVO, contract research), for a total budget of around 2M euro per year. The ES group is a multicultural team, with staff members of eight different nationalities and students from all over the world.

Project description
The envisioned research is part of the recently granted ECSEL project FitOptiVis, coordinated by Philips Healthcare. The objective of FitOptiVis is to develop an integral approach for smart integration of distributed image- and video-processing pipelines for cyber-physical systems (CPS) covering a reference architecture, supported by low-power, high-performance, smart devices, and by methods and tools for combined design-time and run-time multi-objective optimization within system and environment constraints.

In today’s CPS domains such as healthcare, high-tech equipment, and robotics, efficient image processing plays a crucial role. The challenge is to ensure low latency, high-performance, and energy efficiency. The trade-off among these conflicting design objectives is key in the design decisions. This calls for a new paradigm of combined model-driven design-time and runtime multi-objective optimization that is specifically targeted for distributed imaging and video-processing pipelines within CPS. 

The Electronic Systems group of the Electrical Engineering department at TU/e calls for applications for three PhD positions funded in the FitOptiVis project. 

PhD position 1: Model-Driven Design and Optimization of Reconfigurable Streaming Image Pipelines
Model-driven design methods allow automated synthesis of optimized streaming image pipelines. Formal application models such as timed dataflow models, real-time calculus or timed automata provide powerful analysis and synthesis techniques. CPS applications and architectures are highly dynamic and static solutions do not suffice. Run-time adaptation, reconfiguration and optimization are needed to accommodate the dynamic behavior. The run-time adaptation needs to consider multiple objectives.

The FitOptiVis project intends to address this challenge with a model-driven design method that spans both the design-time phase and the run-time phase of CPS. At design-time, reconfigurable components are developed together with compositional models that capture their configurations and the trade-offs they provide considering multiple objectives. Those component are composed into reconfigurable applications deployed onto the system’s virtual platforms. Configurations are optimized at run-time using the component models, optimizing system level quality and resource-management objectives.

The PhD student in this position will develop formal models and their semantics, and analysis, synthesis and optimization methods targeting both design and run-time to realize the FitOptiVis objectives. The position is supervised by Dr. Marc Geilen and Prof.dr.ir. Twan Basten. 

PhD position 2: Multi-Mode Embedded Control Systems for Reconfigurable Video Processing Platforms
Vision-in-the-loop feedback control systems are common in many CPS domains. The processing of video data imposes a long sensing delay in the loop, resulting in a degraded control performance. This often becomes a major bottleneck in the overall vision-based system design. The common practice is to design the system for the worst-case behavior leading a significant resource over-dimensioning. However, the worst-case is very rare and the processing often runs under one of several clearly identifiable finite number of modes (of course, one of them being the worst-case). Depending on the mode, the resource allocation to the video application can be dynamically reconfigured. This way, overall resource usage and dimensioning can be tightened. To realize this, on the one hand, novel design and performance analysis strategies for multi-mode control systems are required. On the other hand, precise models of the embedded implementation platform taking into consideration the artefacts of the underlying reconfiguration mechanism are needed. An important challenge is to effectively design and optimize the appropriate trade off between implementation cost (resource usage) and control performance. This PhD position will focus on the development of design, analysis and optimization techniques for such multi-mode embedded control systems. The position is supervised by Dr. Dip Goswami and Prof.dr.ir. Twan Basten. 

PhD position 3: Composable and Predictable Virtual Platform Reconfiguration
CPS are complex heterogeneous systems with heterogeneous, partially shared, resources and with dynamic stream data processing applications. Platform virtualization is a method to offer applications their private share of the shared resources without the need to address the interference from other applications sharing the resources. This way, separation of concerns is achieved that dramatically simplifies the development and verification of applications and makes their operation more predictable.

FitOptiVis exploits the virtual platform approach by defining a reference architecture in which system resources are offered to applications as virtual platforms that give applications their dedicated share of system resources with composable functionality and performance. Dynamic reconfiguration of applications and resources is indispensable for an efficient and performant CPS. This requires the possibility to dynamically reconfigure virtual platforms in a composable and predictable manner without disturbing the ongoing operation of applications.

The PhD student in this position will develop composable and predictable platform virtualization techniques and distributed composable and predictable reconfiguration methods that support the run-time reconfiguration and optimization strategies of FitOptivis. The student will develop and prototype such techniques on the CompSOC composable and predictable multiprocessor prototype platform. The position is supervised by Dr. Marc Geilen, Dr. Dip Goswami and Prof.dr. Kees Goossens.

Functie-eisen

We are looking for highly motivated candidates with good English proficiency. Candidates need to have obtained a relevant master degree (Computer Engineering, Informatics or Computer Sciences, Electrical Engineering - ideally with focus on Embedded Systems or Cyber Physical Systems) with excellent grades. 

The PhD positions have the following specific requirements:

  • Position 1: Strong analytical skills, affinity for formal models and semantics, algorithmic solutions, and software engineering practices. Good programming skills in C++.
  • Position 2: Strong background in mathematical modeling and optimization. Familiarity with feedback control systems and, preferably, performance analysis and modeling of embedded systems. Good programming skills in C and C++.
  • Position 3: Experience with embedded system development, HW design languages, FPGA development, Good programming skills in C and C++.

Arbeidsvoorwaarden

We offer a challenging job in a dynamic and ambitious university through a fixed-term appointment for a period of 4 years. The research must be concluded with the attainment of a Ph.D. degree. As an employee of the university you will receive a competitive salary as well as excellent employment conditions. The salary starts at € 2222.-per month (gross) in the first year, increasing up to € 2840.- per month (gross) in the last year. Moreover, an 8% holiday allowance and 8,3% end-of-year allowance is provided annually. Assistance for finding accommodation can be given. The university offers an attractive package of fringe benefits such as excellent technical infrastructure, child care, savings schemes, and excellent sports facilities.

TU/e also offers you the opportunity for personal development by developing your social and communication skills. We do this by offering every PhD student a series of courses in the PROOF program as an excellent addition to your scientific education.

Informatie en sollicitatie

Information

For more information about the advertised positions, please contact:

Prof. dr. ir. Twan Basten, a.a.basten@tue.nl

Prof. dr. Kees Goossens, k.g.w.goossens@tue.nl

Dr. Dip Goswami, d.goswami@tue.nl

Dr. ir. Marc Geilen, m.c.w.geilen@tue.nl

Ir. Víctor Sánchez, v.sanchez.martin@tue.nl

More information on employment conditions can be found here:  http://www.tue.nl/en/university/working-at-tue/working-conditions/. 

Application

If interested, please use 'apply now'-button at the top of this page.

Please submit at least the following information (only pdf files are accepted):

1. Cover letter, including availability (starting date), and statement of interests. The letter should mention which of the PhD position(s) you are interested in.

2. Curriculum vitae, including full education and employment histories, publication record, proof of proficiency in English, and the names of at least three references that can be contacted.

3. Scans of certificates showing BSc, MSc, and other courses followed, with grades and rankings.

4. Up to two selected publications (e.g. MSc thesis, conference paper) in English of which you are the first or main author.

 

Please keep in mind; you can upload only 5 documents up to 2 MB each!