Cecilia Ekelin

Evaluation of Search Heuristics for Embedded System Scheduling Problems

In this paper we consider the problem of optimal task allocation and scheduling in embedded real-time systems. This problem is far from trivial due to the wide range of complex constraints that typically appear in this type of systems. We therefore address this problem using constraint programming due to its expressive, yet powerful features. Our work includes an evaluation of different search heuristics, such as variable-value orderings and symmetry exclusion, for this particular problem domain. It is shown that by using search configurations appropriate for the problem, the average search complexity can be reduced by as much as an order of magnitude.

Exact admission-control for integrated aperiodic and periodic tasks

Admission controllers are used to prevent overload in systems with dynamically arriving tasks. Typically, these admission controllers are based on sufficient (but not necessary) capacity bounds in order to maintain a low computational complexity. In this paper we present how exact admission-control for aperiodic tasks can be efficiently obtained. Our first result is an admission controller for purely aperiodic task sets where the test has the same runtime complexity as utilization-based tests. Our second result is an extension of the previous controller for a baseload of periodic tasks. The runtime complexity of this test is lower than for any known exact admission-controller.

The event-triggered and time-triggered medium-access methods

The processes of accessing a shared communication media have been extensively researched in the dependability and real-time area. For embedded systems, the primary approaches have revolved around the event-triggered and the time-triggered paradigms. In this paper, our goal is to objectively and quantitatively outline the capabilities and limitations of each of these paradigms. The event-triggered approach is commonly perceived as providing high flexibility, while the time-triggered approach is expected to provide for a higher degree of predictable communication access to the media. We have quantified the spread of their differences, and provide a summary discussion about suggested best usage for each approach. The focus of our work is on the response times of the communication system, and also on the schedulability of the communication system in collaboration with tasks in the nodes.

A lower-bound algorithm for minimizing network communication in real-time systems

In this paper, we propose a pseudo-polynomial-time lower-bound algorithm for the problem of assigning and scheduling real-time tasks in a distributed system such that the network communication is minimized The key feature of our algorithm is translating the task assignment problem into the so called k-cut problem of a graph, which is known to be solvable in polynomial time for fixed k. Experiments show that the lower bound computed by our algorithm in fact is optimal in up to 89% of the cases and increases the speed of an overall optimization algorithm by a factor of two on average.

A Lower-Bound Algorithm for Load Balancing in Real-Time Systems

We study the problem of finding a safe and tight lower-bound on the load-balancing objective often found in real-time systems. Our approach involves the formulation of the Multiple Bounded Change-Making Problem which we efficiently solve by using a new symmetry-breaking algorithm. An experimental evaluation shows that the computed lower-bound is optimal in more than 70% of the cases and is able to find more than four times as many decidedly optimal solutions.