基于多核ARM的数控系统实时性研究

随着数控系统的不断发展,数控系统对精度和速度有了更高的要求,这都对当前普遍采用的单核处理器平台提出了严峻的挑战.基于同构的四核ARM处理器平台,在添加了实时补丁的Linux系统环境下,根据数控系统中硬实时任务、软实时任务、非实时任务并存的特点,把数控系统任务合理的分配到多核ARM上运行,并采用改进的RM算法进行调度,最后对新的任务调度算法进行实时性能测试.本调度模型能够解决了数控系统混合任务调度的问题,并能满足数控系统对实时性的要求.     

Resource Reservation in Dynamic Real-Time Systems

This paper focuses on the problem of providing efficient run-time support to multimedia applications in a real-time system, where different types of tasks (characterized by different criticality) can coexist. Whereas critical real-time tasks (hard tasks) are guaranteed based on worst-case execution times and minimum interarrival times, multimedia tasks are served based on mean parameters. A novel bandwidth reservation mechanism (the constant bandwidth server) allows real-time tasks to execute in a dynamic environment under a temporal protection mechanism, so that each task will never exceed a predefined bandwidth, independently of its actual requests. The paper also discusses how the proposed server can be used for handling aperiodic tasks efficiently and how a statistical analysis can be applied to perform a probabilistic guarantee of soft tasks. The performance of the proposed method is compared with that of similar service mechanisms (dynamic real-time servers and proportional share schedulers) through extensive simulation experiments.     

RT-ROS: A real-time ROS architecture on multi-core processors

ROS, an open-source robot operating system, is widely used and rapidly developed in the robotics community. However, running on Linux, ROS does not provide real-time guarantees, while real-time tasks are required in many robot applications such as robot motion control. This paper for the first time presents a real-time ROS architecture called RT-RTOS on multi-core processors. RT-ROS provides an integrated real-time/non-real-time task execution environment so real-time and non-real-time ROS nodes can be separately run on a real-time OS and Linux, respectively, with different processor cores. In such a way, real-time tasks can be supported by real-time ROS nodes on a real-time OS, while non-real-time ROS nodes on Linux can provide other functions of ROS. Furthermore, high performance is achieved by executing real-time ROS nodes and non-real-time ROS nodes on different processor cores. We have implemented RT-ROS on a dual-core processor and conducted various experiments with real robot applications. The experimental results show that RT-ROS can effectively provide real-time support for the ROS platform with high performance by exploring the multi-core architecture.     

基于异构多核处理器的H.264并行编码算法

H.264视频编码标准计算复杂度较高,难以完成高清视频的实时编码。为此,提出异构多核DM6467平台的H.264并行编码算法。综合DM6467内部各个硬件加速引擎的依赖关系和存储器特点,设计宏块级并行编码算法,通过分析多slice模式流水线的特点,以及数字信号处理器和ARM双核任务分配,提出合并流水线、核间负载均衡的优化方案。实验结果表明,优化后的编码器效率提高18%,能实现在DM6467平台上1080P的实时编码。     

Windows CE.NET实时性能的测试与研究

嵌入式操作系统的实时性能是衡量一个嵌入式操作系统的重要指标。目前有多种指标和测试方法可以测量一个操作系统的实时性能。本文主要关注于使用OSBench采测量Windows CE．NET任务调度器的实时性能，并分析了在不同系统硬件和不同系统负荷情况下实时性能的变化情况。得到了四种不同同步方式的时间特性，及它们之间的比较差异，也证明了windows CE．NET具有最好的实时性能。     

多任务并行IEEE1394协议栈设计

为满足嵌入式系统对高可靠性和实时性的要求,利用IEEE1394总线标准的软硬件特性,设计多任务并行IEEE1394协议栈,通过设置不同任务优先级获得不同服务质量的数据传输,从而使IEEE1394子系统任务和命令的执行更具实时性。在VxWorks平台上实现的基于该协议栈的数字视频解码与SBP2移动存储验证了该设计的可行性。     

异构平台实时任务的可用性提升容错调度算法

随着互联网+、云计算以及大数据等领域的迅速发展,异构平台成为部署科学计算、工业控制、云存储等关键应用的重要平台.由于平台内处理机性能及软硬件体系结构的异构性,异构平台表现出良好的可扩展性与高性价比.但是平台规模扩大和系统应用日趋复杂导致异构平台上实时任务的可调度性变差,系统可用性降低.针对此问题,提出了一种异构平台实时任务的可用性提升容错调度算法(availability improving fault-tolerant scheduling algorithm, AIFSAL).以处理器利用率和可用性成本为依据设计任务调度整体框架结构、处理机、任务以及调度模型;结合可用性成本设计算法并通过主副版本备份(primary/backup copy, PB)方法实现容错,任务副版本根据处理器利用率不同选择被动或重叠方式执行以减少系统冗余开销,提高可调度性,调度中无论任务主、副版本均优先选择可用性成本低的处理机以提高系统可用性;对任务分配情况和可调度性进行理论分析以证明AIFSAL的可行性.仿真实验与比较分析表明,AIFSAL较可用性约束(availability approached task scheduling algorithm, AATSAL)算法、单调速率扩展(task partition based fault-tolerant rate-monotonic, TPFTRM)算法以及最早完成时间(MinMin)算法在不降低可调度性的基础上有效地提升了系统可用性,减少了系统综合开销,综合性能提高显著.     

RM及其扩展可调度性判定算法性能分析

可调度性判定是实时调度算法的关键问题.单调速率算法RM(rate monotonic)及其扩展是应用广泛的实时调度算法,大量文献讨论了实时任务在这些算法下的可调度性判定,给出了相应的判定算法.但迄今为止,对这些判定算法的性能分析都是理论上的定性分析或者只是少数几种判定算法之间的简单比较,这不利于实时系统的开发.归纳了RM及其扩展的可调度性判定算法,通过测试平台,系统地测试和分析了各算法的性能和适用场合,讨论了各种条件和实现方式对算法性能和可调度性的影响.     

Deadline and activation time assignment for partitioned real-time application on multiprocessor reservations

Providing temporal isolation between critical activities has been an important design criterion in real-time open systems, which can be achieved using resource reservation techniques. As an abstraction of reservation servers, virtual processor is often used to represent a portion of computing power available on a physical platform while hiding the implementation details. In this paper, we present a general framework of partitioning an application comprised of hard real-time tasks with precedence constraints onto multiple virtual processors in consideration of communication latencies between tasks. A novel method is proposed for assigning deadlines and activation times to tasks such that tasks partitioned onto different virtual processors can be analyzed separately using well-established theories for uniprocessor. Extensive simulations have been performed and the results have shown that, compared to existing algorithms, the proposed method achieves better performance in terms of minimizing both total bandwidth and the maximum individual bandwidth.     

Improving adaptivity and fairness of processing real-time tasks with QoS requirements on clusters through dynamic scheduling

In this paper, we consider the problem of scheduling a set of independent real-time tasks with QoS requirements on a cluster, in which the adaptivity and fairness are the two important performance metrics. Thereby, we propose a novel scheduling strategy named AFSS that can guarantee: 1) excellent adaptivity, i.e., more real-time tasks can be accepted when the system is in heavy workload, and real-time tasks have high QoS levels when the system is lightly loaded; 2) fairness, i.e., real-time tasks have fair QoS levels to hold smooth processing quality. The GS and LA algorithms in AFSS are discussed and analyzed. Further, several theorems are given to prove the effectiveness of AFSS.     

截止时限为关键参数的混合关键级实时任务调度研究

混合关键级系统 (mixed criticality system)的研究源于在单一平台上执行多个重要级不同的功能,当前混合关键级调度研究,一般考虑高关键级任务在不同关键级运行模式表现为或周期或计算时间不同,即以任务释放时间间隔和最差情况执行时间为关键参数.但截至时限(dendline)也是实时任务的重要时间参数,尤其是对硬实时任务,以截止时限作为关键参数进行相应混合关键级调度的研究尚是空白.针对此问题,以截止时限作为关键参数,以响应时间分析为基础,对单处理器平台中的混和关键级任务的可调度性进行分析,并提出了一种预提升关键级的调度算法RCLA(raising critical-level in advance),在低关键级运行模式下,高关键级低优先级任务有限度地提前抢占低关键级高优先级任务的执行,既确保了满足高关键级任务在不同关键级运行模式下的截止时限,也让尽可能多的任务可以被调度执行,使得计算资源得以充分利用.仿真结果验证了RCLA算法的有效性.     

无人机平台下的行人与车辆目标实时检测

在无人机图像中快速准确地检测行人和车辆是一项有意义但又极具挑战的任务,其广泛应用于军事侦察、交通管制以及偏远地区救援等任务中。然而,由于无人机属于小型移动设备,其内存和计算能力非常有限,使得如何保证其检测实时性一直是难题。针对SSD算法模型过大、运行内存占用量过高、很难在无人机设备上运行的问题,精心设计了轻量级的基准网络,通过削减原始网络的通道数目以及卷积数目来降低网络的参数量;针对无人机场景下目标小、场景复杂等问题,提出轻量级感受野模块来增强网络特征表示能力,并结合上下文信息来进一步提高小型目标的检测精度。实验结果表明,提出的方法在基于无人机的行人与车辆目标检测任务上有较高的准确性和实时性。     

Linux系统中多核实时调度平台的设计

为使各种实时调度算法能够在Linux环境中得到运行和验证,设计一种Linux系统中的公共多核调度平台。该平台由调度模块和跟踪测试模块组成,调度模块实现系统调用、调度处理以及同步功能,跟踪测试模块实现日志和跟踪功能。测试结果表明,该平台可正确显示调度算法的运行结果。     

基于Windows CE.NET的数控系统实时性能测试与分析

Windows CE.NET自问世以来,得到了越来越广泛的应用.本文在对Windows CE.NET的特点、体系结构及实时性能进行综述的基础上,介绍了数控系统对实时性能的要求以及在基于Windows CE.NET的数控系统中控制实时任务的方法.然后,为了验证Windows CE.NET是否适合作为数控系统的开发平台,本文重点测试并分析了了基于Windows CE.NET数控系统的实时性能.     

Experimental evaluation of slack management in real-time control systems: Coordinated vs. self-triggered approach

Effective slack management, i.e., management of unused computing resources, for real-time control tasks mandates to redistribute the available resources between controllers as a function of the state of the controlled plants. Slack can be allocated to control tasks to alter their rate of progress via e.g., the controllers’ period, in order to adapt their behavior to changes in the computing platform and in the environment. This paper presents an experimental evaluation of two representative slack redistribution policies for multitasking real-time control systems: “coordinated” vs. “self-triggered”. In the coordinated policy a resource manager is responsible for modifying each control task progress. Alternatively, in the self-triggered policy, each control task decides its progress. The demands that each policy poses to the computing platform are analyzed and different operating system architectures providing flexibility and adaptivity are discussed. A proof-of-concept implementation including the real-time control of three double integrator plants in the form of electronic circuits is presented, and a complete performance analysis is reported.     

基于熵权的传感反应网络任务分派算法

针对无线传感反应网络中事件频发区域内执行器节点能耗过大问题,提出一种基于嫡权的任务分派算法。利用拍卖机制,建立基于嫡权的执行代价评估模型,并根据网络的实际情况,将任务元分派建模成平衡或者非平衡整数规划问题,使任务并发执行。仿真实验表明,网络实时性和能耗均衡性都得到了改善。     

基于INtime的六自由度运动平台实时测控系统

在六自由度运动平台的测控系统中,既需要满足控制的实时性和较高的定时精度,又需要有较强的图形图像交互功能。针对现有测控系统采用的上下位机分开实现方式带来的缺点,设计了一种基于INtime的六自由度运动平台实时测控系统。该测控系统通过INtime进程对数据采集和控制板卡的直接操作实现实时性,同时在Windows进程中并行运行非实时任务,能够在一台工业控制计算机上实现测控系统的要求。实际运行测试结果显示,该系统具有高实时性能,满足实时控制的要求,图形渲染流畅,验证了该方案的可行性和有效性。     

基于Flink的任务调度策略

针对大数据流式计算平台拓扑中因各关键节点上任务间不同类型的通信方式导致的通信开销较大问题,提出一种Flink环境下的任务调度策略。通过各任务间数据流大小确定拓扑边权重,将有向无环图转化为拓扑关键路径模型,在保证关键路径上节点负载差异较小的同时,最小化关键任务的节点间通信开销。实验结果表明,该算法与Flink平台现有的任务调度策略相比,在WordCount和TwitterSentiment作业执行过程中计算平均时延降低了13.09%,有效提升了系统性能。     

面向移动平台人脸检测的FaceYoLo算法

针对移动平台上人脸检测实时性不强的问题,提出了一种基于深度学习的FaceYoLo实时人脸检测算法。首先,在YoLov3检测算法的基础上,加入快速消化卷积层（RDCL）缩小输入空间,然后加入多尺度卷积层（MSCL）丰富不同检测尺度的感受野,最后加入中心损失和致密化策略加强模型的泛化能力和鲁棒性。实验结果表明,在GPU上测试时,该算法较YoLov3算法在速度上提高至原来的8倍,每幅图像的处理速度可达0.002 8 s;精度提高了2.1个百分点;在Android平台上测试时,该算法较最好的MobileNet模型在检测速率上从5 frame/s提升到10 frame/s。通过实验结果可知,该算法能有效提高人脸检测在移动平台上的实时性能。     

实时控制任务的模糊反馈调度系统研究

分析了实时控制任务的控制性能在不同控制阶段与处理器利用率需求间的关系,提出一种实时控制任务的模糊反馈调度系统.模糊控制器通过监测实时控制任务的误差及其变化率,查询模糊决策表,动态决定任务的优先级,反馈调度器根据优先级分配任务的利用率.仿真结果表明，在计算资源有限时，该方法能有效改善实时控制任务的控制性能.