Non-preemptive Scheduling Algorithms Research Articles

Energy-Aware Nonpreemptive Scheduling of Mixed-Criticality Real-Time Task Systems

Energy-aware real-time scheduling for mixed-criticality (MC) systems with different criticality levels has drawn many researchers’ attentions. However, most of the studies focus on the preemptive MC task model and few studies consider the nonpreemptive MC task model, in which all jobs cannot be preempted until completion. In this article, we address the energy minimization problem for MC systems with nonpreemptive dynamic priority scheduling. First, we develop schedulability test of nonpreemptive earliest deadline first (NP-EDF) in single processor MC systems. Second, we extend the results to nonpreemptive earliest deadline first with virtual deadline (NP-EDFVD), which is the first attempt for nonpreemptive dynamic priority scheduling in single processor MC systems. Third, the energy-aware nonpreemptive scheduling algorithm (EANPS) based on NP-EDFVD is proposed to solve the energy minimization problem for MC systems with nonpreemptive dynamic priority scheduling. Finally, an industrial use-case and extensive simulations are used to validate the performance of the proposed algorithm, and the experimental results show that the EANPS algorithm consumes average 25.72% less energy than that of NP-EDFVD.

On Non-Preemptive VM Scheduling in the Cloud

We study the problem of scheduling VMs (Virtual Machines) in a distributed server platform, motivated by cloud computing applications. The VMs arrive dynamically over time to the system, and require a certain amount of resources (e.g. memory, CPU, etc) for the duration of their service. To avoid costly preemptions, we consider non-preemptive scheduling: Each VM has to be assigned to a server which has enough residual capacity to accommodate it, and once a VM is assigned to a server, its service cannot be disrupted (preempted). Prior approaches to this problem either have high complexity, require synchronization among the servers, or yield queue sizes/delays which are excessively large. We propose a non-preemptive scheduling algorithm that resolves these issues. In general, given an approximation algorithm to Knapsack with approximation ratio r , our scheduling algorithm can provide rβ fraction of the throughput region for β < r. In the special case of a greedy approximation algorithm to Knapsack, we further show that this condition can be relaxed to β<1. The parameters β and r can be tuned to provide a tradeoff between achievable throughput, delay, and computational complexity of the scheduling algorithm. Finally extensive simulation results using both synthetic and real traffic traces are presented to verify the performance of our algorithm.

On Non-Preemptive VM Scheduling in the Cloud

We study the problem of scheduling VMs (Virtual Machines) in a distributed server platform, motivated by cloud computing applications. The VMs arrive dynamically over time to the system, and require a certain amount of resources (e.g. memory, CPU, etc) for the duration of their service. To avoid costly preemptions, we consider non-preemptive scheduling: Each VM has to be assigned to a server which has enough residual capacity to accommodate it, and once a VM is assigned to a server, its service cannot be disrupted (preempted). Prior approaches to this problem either have high complexity, require synchronization among the servers, or yield queue sizes/delays which are excessively large. We propose a non-preemptive scheduling algorithm that resolves these issues. In general, given an approximation algorithm to Knapsack with approximation ratio r , our scheduling algorithm can provide rβ fraction of the throughput region for β < r . In the special case of a greedy approximation algorithm to Knapsack, we further show that this condition can be relaxed to β<1 . The parameters β and r can be tuned to provide a trade-off between achievable throughput, delay, and computational complexity of the scheduling algorithm. Finally extensive simulation results using both synthetic and real traffic traces are presented to verify the performance of our algorithm.

DPCA: Data Prioritization and Capacity Assignment in Wireless Sensor Networks

Scheduling different types of data packets, such as high or low priority data packets at the sender node, is important for reducing energy and capacity consumptions and end-to-end delay. Current scheduling schemes of wireless sensor networks use preemptive and non-preemptive scheduling algorithms, which incur relatively long end-to-end transmission delay and high processing overhead. Besides, they do not consider the path capacity, which represents the capacity of the network for transferring as much as sensory data to the sink node(s). Consequently, sensory data are routed to the sink node(s), whatever they are more or less, important for supporting domain applications. To remedy this issue, we propose a method, which differentiates between high and low priority when routing sensory data to the sink node(s). Specifically, the priority of sensory data is determined through a novel capacity assignment mechanism. When the network capacity, which depends on the capacity of routing paths, may not be sufficient for supporting the sensory data routing requirement, sensory data with a relatively high priority should be routed to the sink node(s), while that with a relatively low priority may be decreased or prohibited. Experimental evaluation has been conducted, and the result shows that congestion and packets dropping are reduced, when sensory data can be differentiated in their priority and the network traffic is relatively heavy.

Adaptive Open-Shop Scheduling for Optical Interconnection Networks

This paper deals with resource management in optical interconnection networks. It first proposes an optical resource management framework as a platform to develop and evaluate efficient solutions for multipoint-to-multipoint optical communication systems with a centralized controller. The paper then focuses on studying the optical resource scheduling (ORS) problem as a core element in the framework by applying the classical open-shop scheduling theory. The ORS problem can therefore be solved by adopting the existing preemptive and nonpreemptive open-shop scheduling algorithms. In an optical network with nonnegligible reconfiguration delay, a preemptive algorithm may incur high reconfiguration overhead resulting in worse performance compared to the nonpreemptive strategy. Motivated by this fact, this paper proposes an adaptive open-shop scheduling (AOS) algorithm that dynamically decides the optimal scheduling strategy according to traffic condition and system parameters, such as reconfiguration delay, nonpreemptive approximation ratio, and number of involved optical interfaces. The solution is assessed by means of an analytical model that allows to quantify the network performance in terms of packet delay and potential energy savings obtained by the sleep mode operation. As a possible application scenario, the inter- and intrarack optical interconnection networks in data centres are considered. Analytical results demonstrate how the proposed AOS outperforms the nonpreemptive and preemptive scheduling strategies for typical configurations used in data center networks. In addition, the reconfiguration delay and wake-up time of optical devices are identified as performance-determining factors.

Precautious-RM: a predictable non-preemptive scheduling algorithm for harmonic tasks

A major requirement of many real-time embedded systems is to have time-predictable interaction with the environment. More specifically, they need fixed or small sampling and I/O delays, and they cannot cope with large delay jitters. Non-preemptive execution is a known method to reduce the latter delay; however, the corresponding scheduling problem is NP-Hard for periodic tasks. In this paper, we present Precautious-RM as a predictable linear-time online non-preemptive scheduling algorithm for harmonic tasks which can also deal with the former delay, namely sampling delay. We derive conditions of optimality of Precautious-RM and show that satisfying those conditions, tight bounds for best- and worst-case response times of the tasks can be calculated in polynomial-time. More importantly, response time jitter of the tasks is analyzed and it is proven that under specific conditions, each task has either one or two values for response time, which leads to improving the predictability of the system interaction with the environment. Simulation results demonstrate efficiency of Precautious-RM in increasing accuracy of control applications.

Non- Preemptive Real Time Scheduling using Checkpointing Algorithm for Cloud Computing

This paper focuses on providing a solution for online real time services using non-preemptive scheduling algorithm in order to minimize the execution time of the migrated tasks. Earlier, a non-preemptive scheduling with task migration algorithm is used to minimize the penalty. Whenever a task misses its deadline, it will migrate the task to another virtual machine and starts its execution from the beginning. Therefore it increases the execution time of the migrated task. In order to overcome this problem, a non-preemptive real time scheduling using checkpointing algorithm is proposed to minimize the execution time of the migrated tasks and minimizes the penalty even better by earlier completion of migrated tasks. This improves the overall system performance. Our simulation results outperform the older approaches based on the similar model.

Energy-Efficient Scheduling in Nonpreemptive Systems With Real-Time Constraints

In the past decade, the development of mobile and embedded systems has demanded energy efficiency for improving the lifetime of embedded devices. To avoid preemption overhead or ease timing verification, nonpreemptive scheduling has been deemed useful or necessary in meeting system timing requirements for certain applications built on embedded devices. In this paper, our aim is to design nonpreemptive scheduling algorithms that ensure timing correctness and optimize energy consumption on a processor with variable speeds. We propose a representative algorithm, ISA, which can produce lower speeds for a variety of nonpreemptive task sets than other comparable methods, and hence resulting in significant energy savings. When combined with a selective frequency-inheritance policy we design to efficiently determine if processor speedup can be disabled without jeopardizing any task deadlines, ISA can achieve even larger gains, up to 30% reduction in energy consumption. Finally, we propose a dynamic slack reclamation policy built on ISA, namely ISA-DR, which can result in additional energy savings when a task consumes less than its worst-case execution time.

Scheduling Hybrid WDM/TDM Passive Optical Networks With Nonzero Laser Tuning Time

Owing to the high bandwidth provisioning, hybrid wavelength division multiplexing/time division multiplexing (WDM/TDM) passive optical network (PON) is becoming an attractive future-proof access network solution. In hybrid WDM/TDM PON, tunable lasers are potential candidate light sources attributed to their multiwavelength provisioning capability and color-free property. Currently, the laser tuning time ranges from a few tens of nanoseconds to seconds, or even minutes, depending on the adopted technology. Different laser tuning time may introduce different network performance. To achieve small packet delay and ensure fairness, the schedule length for given optical network unit (ONU) requests is desired to be as short as possible. This paper illustrates contributions in four main aspects. First, we show that both preemptive and nonpreemptive scheduling problems with the objective of minimizing the schedule length are NP-hard when the laser tuning time is nonzero. Second, we present a heuristic preemptive scheduling algorithm with an approximation factor of at most 2 and a heuristic nonpreemptive scheduling algorithm with an approximation factor of at most 2-1/ <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">m</i> , where <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">m</i> is the number of wavelengths. Third, extensive simulations have been conducted, and simulation results show that our proposed algorithms, which consider laser tuning time, achieve significantly better performances as compared to algorithms that are directly derived from existing algorithms without considering laser tuning time. Fourth, since the scheduling in one cycle is related to that in the last cycle, we provide some discussions on the scheduling in multiple cycles.

Schedulability analysis for non-preemptive fixed-priority multiprocessor scheduling

Non-preemptive scheduling is usually considered inferior to preemptive scheduling for time critical systems, because the non-preemptive block would lead to poor task responsiveness. Although this is true in single-processor scheduling, we found by empirical simulation experiments that it is not necessarily the case in multiprocessor scheduling. Additionally, non-preemptive scheduling enjoys other benefits like lower implementation complexity and run-time overhead. So non-preemptive scheduling may be a better alternative compared to preemptive scheduling for a considerable part of real-time applications on multiprocessor/multi-core platforms. As the technical contribution, we study the schedulability analysis problem of global non-preemptive fixed-priority scheduling ( NP-FP ) on multiprocessors. We propose schedulability test conditions for NP-FP , building upon the “problem window analysis” by Baruah [8] for preemptive scheduling. We firstly derive a linear-time general schedulability test condition that works on not only NP-FP , but also any other work-conserving non-preemptive scheduling algorithm. Then we improve the analysis and present a test condition of quadratic time-complexity for NP-FP , which has significant performance improvement comparing to the first one. A notable advantage of our proposed test conditions is, while the test in [8] needs to enumerate for a large number of possible problem window sizes, our proposed test conditions only need to be conducted with a single problem window size, and thereby are significantly more efficient. Experiments with randomly generated task sets are conducted to evaluate the performance of the proposed test conditions.

Enhanced Utility Accrual Scheduling Algorithms for Adaptive Real Time System

Problem statement: This study proposed two utility accrual real time scheduling algorithms named as Preemptive Utility Accrual Scheduling (PUAS) and Non-preemptive Utility Accrual Scheduling (NUAS) algorithms. These algorithms addressed the unnecessary abortion problem that was identified in the existing algorithm known as General Utility Scheduling (GUS). It is observed that GUS is inefficient for independent task model because it simply aborts any task that currently executing a resource with lower utility when a new task with higher utility requests the resource. The scheduling optimality criteria are based on maximizing accrued utility accumulated from execution of all tasks in the system. These criteria are named as Utility Accrual (UA). The UA scheduling algorithms are design for adaptive real time system environment where deadline misses are tolerable and do not have great consequences to the system. Approach: We eliminated the scheduling decision to abort a task in GUS and proposed to preempt a task instead of being aborted if the task is preemptive able. We compared the performances of these algorithms by using discrete event simulation. Results: The proposed PUAS algorithm achieved the highest accrued utility for the entire load range. This is followed by the NUAS and GUS algorithms. Conclusion: Simulation results revealed that the proposed algorithms were more efficient than the existing algorithm, producing with higher accrued utility ratio and less abortion ratio making it more suitable and efficient for real time application domain.

A non-preemptive scheduling algorithm for soft real-time systems

Real-time systems are often designed using preemptive scheduling and worst-case execution time estimates to guarantee the execution of high priority tasks. There is, however, an interest in exploring non-preemptive scheduling models for real-time systems, particularly for soft real-time multimedia applications. In this paper, we propose a new algorithm that uses multiple scheduling strategies for efficient non-preemptive scheduling of tasks. Our goal is to improve the success ratio of the well-known Earliest Deadline First (EDF) approach when the load on the system is very high and to improve the overall performance in both underloaded and overloaded conditions. Our approach, known as group-EDF (gEDF) is based on dynamic grouping of tasks with deadlines that are very close to each other, and using Shortest Job First (SJF) technique to schedule tasks within the group. We will present results comparing gEDF with other real-time algorithms including, EDF, Best-effort, and Guarantee, by using randomly generated tasks with varying execution times, release times, deadlines and tolerance to missing deadlines, under varying workloads. We believe that grouping tasks dynamically with similar deadlines and utilizing a secondary criteria, such as minimizing the total execution time (or other metrics such as power or resource availability) for scheduling tasks within a group, can lead to new and more efficient real-time scheduling algorithms.

Segmentation-based nonpreemptive channel scheduling algorithms for optical burst-switched networks

One of the key components in the design of optical burst-switched nodes is the development of channel scheduling algorithms that can efficiently handle data burst contentions. Traditional scheduling techniques use approaches such as wavelength conversion and buffering to resolve burst contention. In this paper, we propose nonpreemptive scheduling algorithms that use burst segmentation to resolve burst contentions. We propose two segmentation-based scheduling algorithms, namely, nonpreemptive minimum overlapping channel (NP-MOC) and NP-MOC with void filling (NP-MOC-VF), which can significantly reduce the loss experienced in an optical burst-switched network. We further reduce packet loss by combining burst segmentation and fiber delay lines (FDLs) to resolve contentions during channel scheduling. We propose two types of scheduling algorithms that are classified based on the placement of the FDL buffers in the optical burst-switched node. These algorithms are referred to as delay-first or segment-first algorithms. The scheduling algorithms with burst segmentation and FDLs are investigated through extensive simulations. The simulation results show that the proposed algorithms can effectively reduce the packet-loss probability compared to existing scheduling techniques. The delay-first algorithms are suitable for applications that have higher delay tolerance and strict loss constraints, while the segment-first algorithms are suitable for applications with higher loss tolerance and strict delay constraints.

Highly predictable execution support for critical applications with HARETICK kernel

In this paper, the problem of providing a fully predictable execution environment for critical and hard real-time applications on embedded and DSP-based platforms is studied from the viewpoint of system architecture and operation. We introduce a set of homogenous models for time, signals and tasks, which will further serve as a basis for describing the architecture and operation of a particular hard real-time kernel – “HARETICK”. The kernel provides support for concurrent operation of hard real-time tasks (the HRT execution environment), using non-preemptive scheduling algorithms, along with soft real-time tasks (the SRT environment), using classical, preemptive, priority-based scheduling algorithms. A set of applications has been developed to test the correct operation of the HARETICK kernel according to the theoretical models and to evaluate its abilities to provide high predictability of execution for critical applications. Some of the main testing results are also discussed in the paper.

Efficient overloading techniques for primary-backup scheduling in real-time systems

In real-time systems, tasks have deadlines to be met despite the presence of faults. Primary-Backup (PB) scheme is one of the most important schemes that has been employed for fault-tolerant scheduling of real-time tasks, wherein each task has two versions and the versions are scheduled on two different processors with time exclusion. There have been techniques proposed for improving schedulability of the PB-based scheduling, of which Backup–Backup (BB) overloading is among the most popular ones. In this technique two or more backups can share/overlap in time on a processor. In this paper, we propose two new techniques that accommodate more tasks and/or tolerate faults effectively. In the first technique, called dynamic grouping, the processors are dynamically grouped into logical groups in order to achieve efficient overloading of tasks on resources, thereby improving the schedulability and the reliability of the system. In the second technique, called PB overloading, the primary of a task can share/overlap in time with the backup of another task on a processor. The intuition is that, for a primary (or backup), the PB-overloading can assign an earlier start time than that of the BB-overloading, thereby increasing the schedulability. We conduct schedulability and reliability analysis of the proposed techniques through simulation and analytical studies. Our studies show that dynamic grouping improves the schedulability more than static grouping, and offers graceful degradation with increasing faults. Also, PB-overloading improves the schedulability more than BB-overloading, and offers reliability comparable to that of BB-overloading. The proposed techniques are generic that they can be incorporated into many fault-tolerant non-preemptive scheduling algorithms.

Efficient overloading techniques for primary-backup scheduling in real-time systems

In real-time systems, tasks have deadlines to be met despite the presence of faults. Primary-Backup (PB) scheme is one of the most important schemes that has been employed for fault-tolerant scheduling of real-time tasks, wherein each task has two versions and the versions are scheduled on two different processors with time exclusion. There have been techniques proposed for improving schedulability of the PB-based scheduling, of which Backup–Backup (BB) overloading is among the most popular ones. In this technique two or more backups can share/overlap in time on a processor. In this paper, we propose two new techniques that accommodate more tasks and/or tolerate faults effectively. In the first technique, called dynamic grouping, the processors are dynamically grouped into logical groups in order to achieve efficient overloading of tasks on resources, thereby improving the schedulability and the reliability of the system. In the second technique, called PB overloading, the primary of a task can share/overlap in time with the backup of another task on a processor. The intuition is that, for a primary (or backup), the PB-overloading can assign an earlier start time than that of the BB-overloading, thereby increasing the schedulability. We conduct schedulability and reliability analysis of the proposed techniques through simulation and analytical studies. Our studies show that dynamic grouping improves the schedulability more than static grouping, and offers graceful degradation with increasing faults. Also, PB-overloading improves the schedulability more than BB-overloading, and offers reliability comparable to that of BB-overloading. The proposed techniques are generic that they can be incorporated into many fault-tolerant non-preemptive scheduling algorithms.

Approximation algorithms for general parallel task scheduling

A general parallel task scheduling problem is considered. A task can be processed in parallel on one of several alternative subsets of processors. The processing time of the task depends on the subset of processors assigned to the task. We first show the hardness of approximating the problem for both preemptive and nonpreemptive cases in the general setting. Next we focus on linear array network of m processors. We give an approximation algorithm of ratio O(log m) for nonpreemptive scheduling, and another algorithm of ratio 2 for preemptive scheduling. Finally, we give a nonpreemptive scheduling algorithm of ratio O(log 2 m) for m× m two-dimensional meshes.