Abstract:

Wireless sensor network systems and applications interact with the  physical world. In order to fulfil requirements such as timeliness, efficiency, and robustness, real-time properties have proven to be essential in many computer systems, and are beneficial in sensor networks as well. Wireless sensing and control systems are typically build up out of many, networked devices with very limited resources. Therefore, they need a small operating system (OS) kernel and scheduling methods that are lightweight and flexible enough for their environment. Such an OS must also be precise enough to give sufficient guarantees for quality of service and/or real-time guarantees.

In this tutorial on real-time systems for networked sensing and control we first introduce the basic background and scheduling methods of real-time system design and programming. Then we concentrate of possible mechanisms to maintain a high quality RTOS  with minimal means. Such an OS should be able to even deal with shared resources while yet RT behaviour can be guaranteed. We show that these  attractive properties are based on deadline inheritance (DI) which is used to refine the behaviour of the most popular scheduling methods such as  earliest deadline first (EDF), deadline monotonic (DM) and rate monotonic (RM).

Then we will focus on the typical requirements for wireless sensing and control. We will show that dynamics and adaptability are two crucial issues to be considered, and we will show how a data centric real-time operating system fits into these requirements. The proposed methods avoid gratuitous task switching while its programming overhead as well as runtime overhead is very low. We will illustrate the elegance of the underlying theory and we will shortly discuss the implementation of these scheduling techniques in DCOS.

Pierre Jansen: Pierre Jansen has performed research on computer architecture and real-time operating systems within industrial and academic contexts. He was responsible for the design of the high performance network of an early multiprocessor and he holds a list of patents in this field. Since 1982 he is working with the University of Twente, where he built an experimental MIMD computer with his group. Within this project, his interests gradually shifted from hard to software and he became involved in operating systems support for embedded systems. From his hand is a list of articles in the field of real-time scheduling. Currently his interest is in specification and implementation of lightweigth architectures, in which timing aspects play an essential role.