Neil C. Audsley

Applying new scheduling theory to static priority pre-emptive scheduling

The paper presents exact schedulability analyses for real-time systems scheduled at runtime with a static priority pre-emptive dispatcher. The tasks to be scheduled are allowed to experience internal blocking (from other tasks with which they share resources) and (with certain restrictions) to release jitter, such as waiting for a message to arrive. The analysis presented is more general than that previously published and subsumes, for example, techniques based on the Rate Monotonic approach. In addition to presenting the relevant theory, an existing avionics case study is described and analysed. The predictions that follow from this analysis are seen to be in close agreement with the behaviour exhibited during simulation studies.

Hard Real-Time Scheduling: The Deadline-Monotonic Approach

Abstract The scheduling of processes to meet deadlines is a difficult problem often simplified by placing severe restrictions upon the timing characteristics of individual processes. One restriction often introduced is that processes must have deadline equal to period. This paper investigates schedulability tests for sets of periodic processes whose deadlines are permitted to be less than their period. Such a relaxation enables sporadic processes to be directly incorporated without alteration to the process model. Following an introduction oudining the constraints associated with existing scheduling approaches and associated schedulability tests, the deadline-monotonic approach is introduced. New schedulability tests are derived which vary in computational complexity. The tests are shown to be directly applicable to the scheduling of sporadic processes.

Fixed priority pre-emptive scheduling: an historical perspective

From its roots in job-shop scheduling, research into fixed priority pre-emptive scheduling theory has progressed from the artificial constraints and simplistic assumptions used in early work to a sufficient level of maturity that it is being increasingly used in the implementation of real-time systems. It is therefore appropriate that within this special issue we provide an historical perspective on the development of fixed priority pre-emptive scheduling.

On priority asignment in fixed priority scheduling

In static priority pre-emptive scheduling there are two areas of related work: priority assignment and feasibility analysis [1]. Given a priority ordering over a task set, feasibility analysis determines whether all task deadlines will be met at run-time. In synchronous periodic systems, all tasks are periodic and are initially released simultaneously. Indeed, a simultaneous release of all tasks occurs at intervals defined by the least common multiple (LCM) of task periods. For synchronous periodic systems, Leung and Whitehead [2] show thedeadline-monotonicpriority ordering policy is optimal where task deadlines do not exceed their (respective) periods. Tasks are assigned a priority inversely proportional to their deadlines— hence the task with the lowest (i.e., shortest) deadline is assigned the highest priority. 1 Optimality implies that if a feasible priority ordering over a task set exists, then a deadline-monotonic priority ordering is also feasible. In general, the complexity of (sufficient and necessary) feasibility analysis is NP-hard [4]. For deadline-

STRESS: a simulator for hard real-time systems

The STRESS environment is a collection of CASE tools for analysing and simulating the behaviour of hard real‐time safety‐critical applications. It is primarily intended as a means by which various scheduling and resource management algorithms can be evaluated, but can also be used to study the general behaviour of applications and real‐time kernels. This paper describes the structure of the STRESS language and its environment, and gives examples of its use.

The End Of The Line For Static Cyclic Scheduling

One common way of constructing hard real-time systems is to use a number of periodic and sporadic tasks assigned static priorities and dispatched at run-time according to the preemptive priority scheduling algorithm. Most analysis for such systems attempts to find the worst-case response time for each task by assuming that the worst-case scheduling point is when all tasks in the system are released simultaneously. Often, however, a given set of hard real-time tasks will have offset constraints: a number of tasks sharing the same periodic behaviour will be constrained to execute at fixed offsets in time relative to each other. In this situation the assumption of a simultaneous release of all tasks can lead to pessimistic scheduling results. In this paper we derive good response time bounds for tasks with offset information, giving an optimal priority ordering algorithm.

Deadline monotonic scheduling theory and application

Abstract Scheduling theories are now sufficiently mature that a genuine engineering approach to the construction of hard real-time systems is possible. In this paper we discuss the application of Deadline Monotonic Scheduling Theory (DMST). This theory is an extension of the more familiar approach based on rate monotonic priority assignment. The model presented can accomodate periodic and sporadic processes, different levels of criticality, process interaction and blocking, precedence constrained processes and multi-deadline processes. It is particularly well integrated with the use of Immediate Priority Ceiling Inheritance for control over process blocking. A basic pseudo-polynomial schedulability test is outlined and then supplemented by the introduction of offsets to control jitter, and period transformation to enable critical (hard) processes to be “protected” during potential transient overloads. These mathematical techniques derived within DMST can help designers experiment with alternative formulations and prove essential properties of systems before they are deployed.

T-CREST

Real-time systems need time-predictable platforms to allow static analysis of the worst-case execution time (WCET). Standard multi-core processors are optimized for the average case and are hardly analyzable. Within the T-CREST project we propose novel solutions for time-predictable multi-core architectures that are optimized for the WCET instead of the average-case execution time. The resulting time-predictable resources (processors, interconnect, memory arbiter, and memory controller) and tools (compiler, WCET analysis) are designed to ease WCET analysis and to optimize WCET performance. Compared to other processors the WCET performance is outstanding.The T-CREST platform is evaluated with two industrial use cases. An application from the avionic domain demonstrates that tasks executing on different cores do not interfere with respect to their WCET. A signal processing application from the railway domain shows that the WCET can be reduced for computation-intensive tasks when distributing the tasks on several cores and using the network-on-chip for communication. With three cores the WCET is improved by a factor of 1.8 and with 15 cores by a factor of 5.7.The T-CREST project is the result of a collaborative research and development project executed by eight partners from academia and industry. The European Commission funded T-CREST.

Analysing APEX applications

The next generation of civil aircraft may be produced using Integrated Modular Avionics (IMA). A component of IMA is APEX, a standard operating system interface. This supports a two-level scheduling scheme consisting of fixed priority scheduling within a statically generated cyclic schedule. This paper illustrates how APEX applications can be analysed for their response times and shows that there is potential for a large amount of release jitter.

Flexible scheduling for adaptable real-time systems

Complex real time systems, such as those envisaged for autonomous vehicle control, are expected to exhibit: adaptive and dynamic behaviour, resilience to software/hardware failures and graceful degradation, under conditions of overload. Two objectives need to be met before such properties can be realised. First, critical services must be guaranteed to provide results of a minimum acceptable quality and reliability by their deadlines. Second, the utility of the system needs to be maximised. We present an approach to meeting the above objectives. This approach combines the benefits of both fixed priority preemptive and best effort scheduling: offline analysis is used to guarantee that critical timing requirements will be met, whilst at run time, a simple adaptive threshold policy arbitrates between competing optional components, enhancing the system utility obtained.