Response Time Analysis Techniques Research Articles

Generalized and Scalable Offset-Based Response Time Analysis of Fixed Priority Systems

Abstract Offset-based response time analysis techniques obtain tight worst-case response time (WCRT) bounds by accounting release time dependencies between tasks. The maximum response time variation of a task or a message (i.e., the difference between the worst-case and best-case response time) is used to compute the end-to-end delays of distributed systems (Palencia and Harbour, 1998), (Tindell and Clark, 1994). Hence, the WCRT evaluation plays an important role in determining tight end-to-end delays of distributed systems. In real-time theory, there exist two approaches to compute the WCRT of a task: the classical response time analysis (RTA) approach and the modular performance analysis with the Real-Time Calculus (MPA-RTC). MPA-RTC has its roots in Network Calculus (NC). MPA-RTC offers more powerful abstraction than RTA based techniques and allows composition in terms of tasks, event streams, and resource sharing which makes it a strong candidate to analyze distributed systems (Wandeler, 2006), (Perathoner, 2011). However, one of the key limitations of the MPA-RTC is its inability to handle offset dependencies between tasks, whereas the classical RTA techniques can handle them. In this paper, we propose a method to consider offset dependencies between tasks using an MPA-RTC framework for a fixed priority scheduler in a uniprocessor system. Hence, our approach leverages the advantages of the expressive MPA-RTC framework model. We propose novel heuristics and approximation to reduce the inherent complexity of an offset-based RTA. We quantitatively evaluate the effectiveness of our approach to the state-of-the-art classical offset-based RTA techniques.

Response-Time Analysis of Multipath Flows in Hierarchically-Scheduled Time-Partitioned Distributed Real-Time Systems

Modern industrial cyberphisical systems exhibit increasingly complex execution patterns like multipath end-to-end flows, that force the real-time community to extend the schedulability analysis methods to include these patterns. Only then it is possible to ensure that applications meet their deadlines even in the worst-case scenario. As a driving motivation, we present a real industrial application with safety requirements, that needs to be re-factored in order to leverage the features of new execution paradigms such as time partitioning. In this context we develop a new response-time analysis technique that provides the capacity of obtaining the worst-case response time of multipath flows in time-partitioned hierarchical schedulers and also in general fixed-priority (FP) real-time systems. We show that the results obtained with the new analysis reduce the pessimism of the currently used holistic analysis approach.

The concept of Maximal Unschedulable Deadline Assignment for optimization in fixed-priority scheduled real-time systems

This paper considers the problem of design optimization for real-time systems scheduled with fixed priority, where task priority assignment is part of the decision variables, and the timing constraints and/or objective function linearly depend on the exact value of task response times (such as end-to-end deadline constraints). The complexity of response time analysis techniques makes it difficult to leverage existing optimization frameworks and scale to large designs. Instead, we propose an efficient optimization framework that is three orders of magnitude (1000 times) faster than Integer Linear Programming (ILP) while providing solutions with the same quality. The framework centers around three novel ideas: (1) an efficient algorithm that finds a schedulable task priority assignment for minimizing the average worst-case response time; (2) the concept of Maximal Unschedulable Deadline Assignment (MUDA) that abstracts the schedulability conditions, i.e., a set of maximal virtual deadline assignments such that the system is unschedulable; and (3) a new optimization procedure that leverages the concept of MUDA and the efficient algorithm to compute it.

A Unified Framework for Period and Priority Optimization in Distributed Hard Real-Time Systems

Modern embedded systems, such as automotive, are physically distributed with an increasing number of microcontrollers and buses. They support complex functions such as active safety and autonomous driving features with a high degree of data dependencies. The most common configuration uses periodic activation of tasks and messages coupled with priority-based scheduling. Selecting task and message parameters so that end-to-end deadlines are met can be very challenging, since such deadlines are enforced across a set of microcontrollers and buses. In this paper, we address the problem of optimal selection of task and message activation periods and priorities. Existing approaches cannot scale to large designs and have to settle to optimize period or priority separately, largely due to the complexity of response time analysis techniques. Instead, we present a new, unified framework that simultaneously optimizes period and priority assignment. It avoids the pitfalls of existing approaches by abstracting the response time calculation with the new concept of maximal unschedulable period and deadline assignment. We demonstrate with two industrial case studies that our approach runs magnitudes faster than existing approach on period optimization, while providing substantially better solutions.

A hybrid performance analysis technique for distributed real-time embedded systems

It remains a challenging problem to tightly estimate the worst-case response time of an application in a distributed embedded system, especially when there are dependencies between tasks. Recently, a holistic worst-case response time analysis approach called scheduling time bound analysis has been proposed to find a tight upper bound of the worst-case response times of applications specified by a set of task graphs. Since it assumes that the starting offsets of applications are known and fixed, it fails to make a tight estimation despite increased computation time when the starting offsets are dynamic. To overcome this problem, we propose a novel conservative performance analysis, called hybrid performance analysis, combining the response time analysis technique and the scheduling time bound analysis technique to compute a tighter bound faster. The proposed scheme is proven to be conservative formally. Through extensive experiments with real-life benchmarks and synthetic examples, the superior performance of our proposed approach compared with previous methods is confirmed.

On the ineffectiveness of 1/m-based interference bounds in the analysis of global EDF and FIFO scheduling

Enormous efforts have been spent in the derivation of sufficient schedulability tests for popular global schedulers such as global fixed-priority (G-FP) and global earliest-deadline first (G-EDF). Among all the proposals, response-time analysis techniques are established as the most popular ones and have been widely adopted by several authors in subsequent researches. Such tests are based on interference bounds that have been derived by exploiting the work-conserving property of such schedulers, and are typically expressed with a 1 / m multiplier in the interference equation (with m being the number of available processors). This paper shows that the popular response-time analysis for G-EDF is ineffective with respect to partitioned EDF (P-EDF) scheduling. That is, it is proven that every task set that is deemed schedulable by such an analysis is also schedulable under P-EDF by adopting a trivial partitioning algorithm. Furthermore, an analogous result is proven for global first-in first-out (G-FIFO) scheduling when analyzed with a 1 / m-based interference bound. Finally, experimental results are presented to compare sufficient tests for G-EDF and G-FIFO with the corresponding partitioned schemes under a vast collection of partitioning heuristics. The results show that the considered tests for global scheduling are always outperformed by those for partitioned scheduling exhibiting a significant performance gap.

On flexible and robust parameter assignment for periodic real-time components

In order to increase the flexibility of design process, we consider a component-based system in which some components may be changed, repaired, or upgraded. We assume that the worst-case execution time (WCET) of each task that is implemented by a component is known. Our first goal is to find the smallest set of periods that make the task set schedulable by EDF or by RM. We call these periods safe periods because then any period assignment which is larger than the safe periods will be schedulable. Having safe periods, system designer will be able to use any optimization criteria of his choice to assign periods of the tasks without a need to incorporate the response time analysis techniques, which are usually NP-Hard problems, into his optimization constraints. Instead, the schedulability constraint will be reduced to verifying that the assigned periods are larger than the safe periods. Our solution for safe periods for the RM, is based on finding the smallest set of harmonic periods with utilization 1. Here we use our recently developed polynomial-time approximation algorithm with bounded error of 2 for finding these safe periods for RM. As the second part of our contribution, we consider the robustness property. First we explain how to find safe periods such that a certain level of robustness (based on potential changes of each component) is guaranteed. Doing this, if one of the components changes in the future, other components can still run using their previous periods, and hence, we isolate the effect of future changes from parameters of the other components. Finally, we obtain the robustness factor as a function of the available spare capacity of the system, i.e., the unused utilization. We determine to what extend the WCET of any of the components can be increased without violating the safe periods.

A fine-grained response time analysis technique in heterogeneous environments

It is crucial for the network operators and Internet service providers (ISPs) to determine the reasons that cause large response time fluctuations. In this paper, we consider passive measurements from heterogeneous environment (ADSL, FTTH and 3G/3G+ access technologies) of an European ISP ‘Orange’. Through experimental analysis of real traces, the need of a fine-grained traffic analysis technique is demonstrated. We show that finding the root causes of the observed poor performance using simple metrics such as response time, RTT and packet loss is difficult. In view of this fact, the different factors that play a role in determining the resulting response time are described through examples. Then, a breakdown method that drills down into the passively observed TCP connections is proposed. The method decomposes the end-to-end response time into many time periods and maps each one to a specific parameter or a physical phenomenon. Thus, the impact of not only the network parameters but also the application configuration and user behavior is captured. The resulting time periods are given as input to a clustering algorithm in order to group together transfers with similar performance holding traffic of different application protocols over different access technologies. As a result, the contribution of each participant in the performance bottleneck is identified. The proposed technique is validated through extensive simulations and real passively measured traces and it is compared to other works. Exemplifying the technique on real traces from Internet and enterprise traffic is introduced and discussed to demonstrate the power of the approach and its simplicity. In contrast to some existing tools, ISPs and enterprise administrators do not need to modify their network architecture or to install a new software or a plugin at the client or at the server side in order to use our technique. In addition, data sampling is not used. This is particularly important in order to keep data consistency and to detect metrics peaks. Last, our tool deals with both long and short TCP connections.

Improved Carry-in Workload Estimation for Global Multiprocessor Scheduling

As an important and fundamental tool for analyzing the schedulability of a real-time task set on the multiprocessor platform, response time analysis (RTA) has been researched for several years on both Global Fixed Priority (G-FP) and Global Earliest Deadline First (G-EDF) scheduling. This paper proposes a new analysis that improves over current state-of-the-art RTA methods for both G-FP and G-EDF scheduling, by reducing their pessimism. The key observation is that when estimating the carry-in workload, all the existing RTA techniques depend on the worst case scenario in which the carry-in job should execute as late as possible and just finishes execution before its worst case response time (WCRT). But the carry-in workload calculated under this assumption may be over-estimated, and thus the accuracy of the response time analysis may be impacted. To address this problem, we first propose a new method to estimate the carry-in workload more precisely. The proposed method does not depend on any specific scheduling algorithm and can be used for both G-FP and G-EDF scheduling. We then propose a general RTA algorithm that can improve most existing RTA tests by incorporating our carry-in estimation method. To further improve the execution efficiency, we also introduce an optimization technique for our RTA tests. Experiments with randomly generated task sets are conducted and the results show that, compared with the state-of-the-art technologies, the proposed tests exhibit considerable performance improvements, up to 9 and 7.8 percent under G-FP and G-EDF scheduling respectively, in terms of schedulability test precision.

New response time analysis for global EDF on a multiprocessor platform

Time-predictability is the most important requirement for a real-time system, and researchers have therefore paid attention to the duration between the arrival and completion of a real-time task, called response time. RTA (Response Time Analysis) studies, however, rely on the same technique, yielding room for further improvement, especially regarding multiprocessor platforms. For this paper, we investigated the properties of an existing utilization-based schedulability analysis for global EDF (Earliest Deadline First) on a multiprocessor platform, and developed a new RTA technique based on the corresponding properties, which calculates the response times of tasks in task sets deemed schedulable by the existing analysis. We demonstrated through simulations that our proposed RTA technique not only calculates response times that are less pessimistic than those of the existing approach, but also successfully derives response times that cannot be obtained by the existing approach.

On the convergence of the holistic analysis for EDF distributed systems

Dynamic scheduling techniques, and EDF (Earliest Deadline First) in particular, have demonstrated their ability to increase the schedulability of real time systems compared to fixed-priority scheduling. In distributed systems, the scheduling policies of the processing nodes tend to be the same as in stand-alone systems and, although few EDF networks exist, it is foreseen that dynamic scheduling will gradually develop into real-time networks. There are some response time analysis techniques for EDF scheduled distributed systems, mostly derived from the holistic analysis developed by Spuri. A major factor influencing the response time is the release jitter of each task, which is the maximum variation suffered by the release time of the task jobs. The convergence of the holistic analysis in the context of EDF distributed systems with shared resources had not been studied until now. There is a circular dependency between the task release jitter values, response times and the preemption level ceilings of shared resources. In this paper we present an extension of Spuri's algorithm and we demonstrate that its iterative formulas are non-decreasing, even in the presence of shared resources. This result enables us to assert that the new algorithm converges towards a solution for the response times of the tasks and messages in a distributed system.1A more extensive version of this work is available as a technical report in: http://hdl.handle.net/10902/5309.1

A scheme for slot allocation of the FlexRay Static Segment based on response time analysis

In the last decade, the FlexRay communication protocol has been promoted as a standard for dependable in-vehicular communications. In the FlexRay protocol, the communication timeline is organized as a sequence of four segments, whereas the Static Segment assigns a set of static slots for the transmission of synchronous messages. In this paper, we address the following problems: “How to efficiently transmit periodic messages in the Static Segment without requiring their periods to be multiples of, or to be synchronized with the FlexRay Communication Cycle?” “Is it possible to guarantee that periodic messages are transferred before their deadlines, without imposing such strict synchronization?” Unlike traditional approaches that use linear-programming based techniques, we evaluate the minimum number of allocated slots using traditional Response Time Analysis (RTA). The use of RTA techniques allows us to consider the timing requirements associated to each of the asynchronous message streams. Unlike other approaches, the RTA-based technique proposed in this paper: (a) is able to deal with message stream sets where periods are not multiple of the FlexRay cycle duration and (b) does not require the strict synchronization between tasks/signals at the application layer and slots at the FlexRay communication controller. The proposed slot allocation scheme may be of high practical interest when considering the interconnection of FlexRay/CAN in-vehicular communication systems, allowing the remapping of existing CAN message streams to FlexRay.

Migrating from Per-Job Analysis to Per-Resource Analysis for Tighter Bounds of End-to-End Response Times

As the software complexity drastically increases for multiresource real-time systems, industries have great needs for analytically validating real-time behaviors of their complex software systems. Possible candidates for such analytic validations are the end-to-end response time analysis techniques that can analytically find the worst-case response times of real-time transactions over multiple resources. The existing techniques, however, exhibit severe overestimation when real-time transactions visit the same resource multiple times, which we call a multiple visit problem. To address the problem, this paper proposes a novel analysis that completely changes its analysis viewpoint from classical per-job basis-aggregation of per-job response times-to per-resource basis-aggregation of per-resource total delays. Our experiments show that the proposed analysis can find significantly tighter bounds of end-to-end response times compared with the existing per-job-based analysis.

An efficient response-time analysis for real-time transactions with fixed priority assignment

In the general context of tasks with offsets (general transactions), only exponential methods are known to calculate the exact worst-case response time (WCRT) of a task. The known pseudo-polynomial techniques give an upper bound of the WCRT. In this paper, we present a new worst-case response-time analysis technique (mixed method) for transactions scheduled by fixed priorities, and executing on a uniprocessor system. This new analysis technique gives a better (i.e. lower) pseudo-polynomial upper bound of the WCRT. The main idea is to combine the principle of exact calculation and the principle of approximation calculation, in order to decrease the pessimism of WCRT analysis, thus allowing improving the upper bound of the response time provided while preserving a pseudo-polynomial complexity. Then we define the Accumulative Monotonic property on which a necessary condition of feasibility is discussed. We also propose, to speed up the exact and mixed analysis, the dominated candidate task concept that allows reducing significantly the number of critical instants to study in an analysis.