Visiting Cyber–Physical Systems Group, University of California Los Angeles (November 2014–January 2015) Amir Aminifar Department of Computer and Information Science, Linköping University, Sweden amir.aminifar@liu.se I. I NTRODUCTION I have been working on embedded control systems since 2010. Along with our research on analysis and design of embedded control systems, I visited the cyberphysical systems group1 of the university of California Los Angeles (UCLA) for a period of three months. Professor P. Tabuada, head of the cyber-physical systems group in UCLA, is one of the main experts in the area of eventtriggered control [1] and self-triggered [2]. Before this visit, in my Ph.D. studies, all our results were based on the traditional periodic control scheme. The advantages of self-triggered control attracted our attention as a potentially rich area of research. The self-triggered control is a new paradigm for control engineers which provides the same performance as the traditional periodic control, while consumes considerably less resources. The essence is that the self-triggered controller executes only when it is needed. Although the paradigm has been proved to be effective, the real-time scheduling problem of such controllers is still an open issue. In the embedded systems group (ESLAB), we are involved in the analysis and design of embedded real-time systems, while the cyberphysical systems group in UCLA are experts in the area of self-triggered control. The main goal of this visit was to exchange ideas and conduct research in the area of selftriggered control, i.e., to facilitate real-time scheduling of the self-triggered controllers. II. C YBER –P HYSICAL S YSTEMS G ROUP AND E LECTRICAL E NGINEERING D EPARTMENT The cyber–physical systems group of UCLA is led by professor P. Tabuada. There are two post-doctoral fellows, six doctoral students, and a few master students. Besides self-triggered and event-triggered control, other main research areas are controller synthesis and security in cyber-physical systems. They have weekly meetings where everybody talks about what (s)he has been doing and what is the main problem (s)he is dealing with. Occasionally, there are visitors who give talks about their research. I gave two talks while I was there, on our work in ESLAB on control–scheduling co-design and other realtime topics. It was interesting to get their feedback and questions. Something I noticed is that they usually have more visitors and collaboration in the electrical engineering de1 www.cyphylab.ee.ucla.edu partment of UCLA than in our department. This might be due to the fact that the electrical engineering department of UCLA has ranked one worldwide by Microsoft Academic Search based on H-index over ten years. 2 Although we have SAS seminars in the division, but I believe I noticed more interesting talks and collaboration in UCLA. Another interesting thing was the fact that they had many events to prepare their senior students for the future, i.e., in finding jobs and applying for funding. In addition, they also had a career week where the companies come to university to recruit students. III. R ESEARCH R ESULTS For the first couple of weeks, I have been reading a few papers related to self-triggered control and then we discussed several directions for this visit. By the end of the first month, the research direction in this visit was clear. The ultimate goal was to do schedulability analysis for self-triggered controllers running on a shared platform, possibly in presence of other hard real-time tasks. The first task, however, is to identify the self-triggered scheme which facilitates this goal. The first result in this visit was motivated by the following observation. The traditional self-triggered controllers often work in a greedy fashion, i.e., they compute the latest time that the controller should be executed in order for the plant to remain stable. However, after this step, due to the greedy nature of the algorithm, the control task might be required to execute very frequently to guarantee stability, such that it defeats the purpose of the self-triggered control, i.e., less resource usage. This is important from schedulability point of view because if we consider this greedy case in our analysis, then the interference from the self-triggered controller is considerably more than what happens in reality, since always the worst-case scenario is considered. And essentially, this leads to a very pessimistic analysis method. To remedy this, our proposed design will look into a finite horizon in the future and avoid such catastrophic cases. Preliminary studies show that our proposed self-triggered controller outperforms the previous results roughly by an order of magnitude, in terms of the number executions of the self-triggered controller. 2 http://www.engineer.ucla.edu/newsroom/featured-news/archive/ 2012/electrical-engineering-ranked-no.-1-and-school-ranked-no. -4-by-microsoft-academic-search (a) Half Dome (b) Top Figure 1. Yosemite national part VII. ACKNOWLEDGEMENTS I would like to acknowledge professor P. Tabuada at UCLA for hosting my visit and embedded systems group and CUGS (Swedish National Graduate School in Computer Science) for funding this visit. R EFERENCES Figure 2. Page museum at the La Brea Tar Pits In addition, we have set the basis for response-time analysis of self-triggered controllers, which is considered to be an open problem. The plan for the future is to work on this idea. IV. R ECOMMENDATION I think it is important to realize that three months visit is in fact very short and the overhead of paperwork related to visa/university and finding accommodation and planning the trip might be considerable. I also recommend to do such visits earlier in your studies, maybe in the third year of your studies. Keep in mind that it is great if there is a student in the host group who is working on the topic you are interested in. V. A DVENTURE AND S IGHTSEEING The adventure part of this visit was going to a winter hiking trip in Yosemite national part (see Figure 1) with UCLA outdoor activity group for a weekend in January. Beside the amazing nature, I got to meet many nice students. Sleeping in tents, star gazing, and hiking to waterfalls were some of the highlights of this trip. I also spent one Saturday in Page museum at the La Brea Tar Pits, one of the most famous fossil localities in the world (see Figure 2). VI. C ONCLUSION During this visit, not only did I enjoy working on an interesting topic and broaden my knowledge, but also I got to establish contacts for future collaboration and experienced working in another research environment. [1] P. Tabuada, “Event-triggered real-time scheduling of stabilizing control tasks,” IEEE Transactions on Automatic Control, vol. 52, no. 9, pp. 1680–1685, 2007. [2] A. Anta and P. Tabuada, “To sample or not to sample: Selftriggered control for nonlinear systems,” IEEE Transactions on Automatic Control, vol. 55, no. 9, pp. 2030–2042, Sept 2010.