Showing papers by "Sanjoy Baruah published in 2011"

Response-Time Analysis for Mixed Criticality Systems

Abstract: Many safety-critical embedded systems are subject to certification requirements. However, only a subset of the functionality of the system may be safety-critical and hence subject to certification, the rest of the functionality is non safety-critical and does not need to be certified, or is certified to a lower level. The resulting mixed criticality system offers challenges both for static schedulability analysis and run-time monitoring. This paper considers a novel implementation scheme for fixed priority uniprocessor scheduling of mixed criticality systems. The scheme requires that jobs have their execution times monitored (as is usually the case in high integrity systems). An optimal priority assignment scheme is derived and sufficient response-time analysis is provided. The new scheme formally dominates those previously published. Evaluations illustrate the benefits of the scheme.

Mixed-criticality scheduling of sporadic task systems

Abstract: We consider the scheduling of mixed-criticality task systems, that is, systems where each task to be scheduled has multiple levels of worst-case execution time estimates. We design a scheduling algorithm, EDF-VD, whose effectiveness we analyze using the processor speedup metric: we show that any 2-level task system that is schedulable on a unit-speed processor is correctly scheduled by EDF-VD using speed φ here φ 2 criticality levels. We finally consider 2-level instances on m identical machines. We prove speedup bounds for scheduling an independent collection of jobs and for the partitioned scheduling of a 2-level task system.

Certification-Cognizant Time-Triggered Scheduling of Mixed-Criticality Systems

Abstract: In many modern embedded platforms, safety-critical functionalities that must be certified correct to very high levels of assurance co-exist with less critical software that are not subject to certification requirements. Recent research in real-time scheduling theory has yielded some promising techniques for meeting the dual goals of (i) being able to certify the safety-critical functionalities under very conservative assumptions, and (ii) ensuring high utilization of platform resources under less pessimistic assumptions. This research has centered on an event-triggered/ priority-driven approach to scheduling. However current practice in many safety-critical domains, including (the safety-critical components of) automotive and avionics systems and factory automation, favors a time-triggered approach. In such time-triggered systems, non-interference of safety-critical components by non-critical ones is ensured by strict isolation between components of different criticalities, although such isolation facilitates the certification of the safety-critical functionalities, it can cause very low resource utilization. The research reported in this document is, to our knowledge, the first to study time-triggered scheduling from the perspective of both ensuring certifiability of high-criticality functionalities, and obtaining high resource utilization as in (i) and (ii) above. We present algorithms for time-triggered scheduling of mixed-criticality systems that offers resource utilization guarantees similar to those of event-triggered scheduling. Since the time-triggered approach currently seems to find greater acceptability with certification authorities, it is hoped that this research will hasten the adoption of these results in building embedded systems that are subject to mandatory certification.

Implementing mixed criticality systems in Ada

Abstract: Many safety-critical embedded systems are subject to certification requirements. However, only a subset of the functionality of the system may be safety-critical and hence subject to certification; the rest of the functionality is non safety-critical and does not need to be certified, or is certified to a lower level. The resulting mixed criticality system offers challenges both for static schedulability analysis and run-time monitoring. This paper considers both of these issues and indicates how mixed criticality applications can be implemented in Ada. In particular, code is produced to illustrate how the necessary run-time mode changes can be supported. This support makes use of a number of the new features introduced into Ada 2005.

Timing faults and mixed criticality systems

Abstract: Many safety-critical embedded systems are subject to certification requirements. However, only a subset of the functionality of the system may be safety-critical and hence subject to certification; the rest of the functionality is non safety-critical and does not need to be certified, or is certified to a lower level. The resulting mixed criticality system offers challenges both for static analysis and run-time monitoring. This paper is concerned with timing failures and how they can arise and be tolerated. The main causes of these errors are faults in the estimation of worst-case execution times (WCETs). For different levels of criticality, different forms of static analysis for WCET are employed. This give rise to a novel implementation scheme for the fixed priority uniprocessor scheduling of mixed criticality systems. The scheme requires that jobs have their execution times monitored (as is usually the case in high integrity systems). This results in higher levels of schedulability than previously published.

A Lookup-Table Driven Approach to Partitioned Scheduling

Abstract: The partitioned preemptive EDF scheduling of implicit-deadline sporadic task systems on an identical multiprocessor platform is considered. Lookup tables, at any selected degree of accuracy, are pre-computed for the multiprocessor platform. By using these lookup tables, task partitioning can be performed in time polynomial in the representation of the task system being partitioned. Although the partitioning will not in general be optimal, the degree of deviation from optimality is bounded according to the degree of accuracy selected during the pre-computation of the lookup tables.

The Partitioned EDF Scheduling of Sporadic Task Systems

Abstract: The partitioned scheduling of sporadic task systems on identical multiprocessors is considered. This is known to be intractable (NP-hard in the strong sense). A polynomial-time approximation scheme (PTAS) is proposed for sporadic task systems satisfying the additional constraint that for each of the three parameters -- worst-case execution time, relative deadline, and period -- that characterize sporadic tasks, the ratio of the largest value to the smallest value is bounded from above by a constant.

Partitioned Real-time Scheduling on Heterogeneous Shared-Memory Multiprocessors

Abstract: We consider several real-time scheduling problems on heterogeneous multiprocessor platforms, in which the different processors share a common memory pool. These include (i)~scheduling a collection of implicit-deadline sporadic tasks with the objective of meeting all deadlines, and (ii)~scheduling a collection of independent jobs with the objective of minimizing the make span of the schedule. Both these problems are intractable (NP-hard). For each, we derive polynomial-time algorithms for solving them approximately, and show that these algorithms have bounded deviation from optimal behavior. We also consider the problem of determining how much common memory a platform needs in order to be able to accommodate a specified real-time workload.

Sensitivity analysis of arbitrary deadline real-time systems with EDF scheduling

Abstract: The correctness of a real-time system depends on not only the system's output but also on the time at which results are produced. A hard real-time system is required to complete its operations before all its timing deadlines. For a given task set it is useful to know what changes can be made to a task that will result in a system that is borderline schedulable. It is also beneficial in an engineering context to know the minimum speed of a processor that will deliver a schedulable system. We address the following sensitivity analysis (parameter computations) for EDF-scheduled systems on a uniprocessor: task execution times, speed of the processor, task periods and task relative deadlines. We prove that an optimal (minimum or maximum) system parameter can be determined by a single run of the Quick convergence Processor demand Analysis (QPA) algorithm. This algorithm provides efficient and exact sensitivity analysis for arbitrary deadline real-time systems. We also improve the implementation of this sensitivity analysis by using various starting values for the algorithms. The approaches developed for task parameter computations are therefore as efficient as QPA, and are easily incorporated into a system design support tool.

Efficient computation of response time bounds for preemptive uniprocessor deadline monotonic scheduling

Abstract: The deadline-monotonic (DM) scheduling of sporadic task systems upon a preemptive uniprocessor is considered. A technique is derived for determining upper bounds on the response time of the jobs of each task, when a constrained-deadline sporadic task system is scheduled. This technique yields a generalization to a load-based sufficient schedulability condition for DM, the generalization being the added ability to account for blocking in the presence of non-preemptable serially re-usable resources.

Real-Time Divisible Load Theory: Incorporating Computation Costs

Abstract: We extend the current state of the art in real-time divisible load theory (RT-DLT), by considering the problems of scheduling a real-time divisible job on computing clusters in which different processing nodes have different computing capabilities, as well as different costs associated with executing on them. We seek to minimize the cost of executing a job while also meeting its deadline.

Proceedings of the ninth ACM international conference on Embedded software

Abstract: Embedded systems play a critically important role in the modern world; the complexity of such systems is increasing at a breathtaking pace. EMSOFT, the International Conference on Embedded Software, is the flagship conference sponsored by ACM SIGBED, the Special Interest Group on Embedded Systems. In this role, EMSOFT seeks to bring together, on an annual basis, researchers and developers from academia, industry, and government to advance the science and the practice of embedded software development. It is the opinion of the international panel of experts comprising the EMSOFT technical program committee this year, that the papers included in these proceedings are representative of some of the finest research performed last year in the general domain of embedded software. This year 111 submissions were received and were reviewed by the program committee. After a very rigorous (and sometimes contentious) review process, a total of 27 papers were selected for presentation at EMSOFT and inclusion in the proceedings. In addition to these peer-reviewed papers, EMSOFT this year introduces an exciting innovation: roadmap presentations. Five leading experts in domains of particular interest to EMSOFT have been invited to make presentations during the conference, that introduces these domains to a wider audience and provides an overview of ongoing and upcoming research challenges. We believe that these roadmap presentations provide additional direction and focus to the conference program. Extended abstracts of the roadmap presentations appear in these proceedings. Two invited sessions, one focused on automotive control and the other dealing with software certification, round off the conference program; papers from the participants of these invited sessions also appear in these proceedings. We believe that the papers published in these proceedings represent an accurate snapshot of the efforts being made by our research community to provide firm theoretical and engineering foundations to the discipline of embedded software development, and to come up with new techniques, tools, and methodologies for the analysis, design, and implementation of software for controlling increasingly complex embedded systems. We hope that this selection does some justice to both the breadth and the technical depth that together make the study of embedded software so very fascinating from both a scientific and an engineering perspective.