引用本文
  • 董博南,杨秋松,李明树.Iot-EDF:基于 Unikernel 的物联网任务调度方法[J].信息安全学报,已采用    [点击复制]
  • DongBonan,YangQiuSong,LiMingShu.Iot-EDF: Unikernel-based Task Scheduling Method for the Internet of Things[J].Journal of Cyber Security,Accept   [点击复制]
【打印本页】 【下载PDF全文】 查看/发表评论下载PDF阅读器关闭

过刊浏览    高级检索

本文已被:浏览 100次   下载 0  
Iot-EDF:基于 Unikernel 的物联网任务调度方法
董博南, 杨秋松, 李明树
0
(中国科学院软件研究所)
摘要:
Unikernel作为虚拟化领域的前沿技术,在物联网环境中因其启动快速和低资源消耗以及高安全性的特点而被广泛应用。然而目前Unikernel缺乏根据不同任务特点所设计的动态调度机制,用以保障物联网大规模任务调度场景下的工作效率。为解决这一问题,首先总结了物联网环境下Unikernel的任务特点,在此基础上,提出一种新的基于Unikernel的物联网环境下的调度机制Iot-EDF及其数学模型,该机制通过划分任务的重要性和截止时间,不仅有效的提高了Unikernel在物联网环境中的任务调度效率,避免了重要数据丢失的问题,还能通过数学模型预测任务执行的总体情况,降低任务执行失败的风险。同时,基于网络时间协议,提出一种适用于Unikernel物联网场景下的时钟同步方法,通过调整从不同地点发送的数据包中的时间戳来生成时差表,以解决全局时钟同步的问题,从而确保Iot-EDF调度机制中对截止时间计算的准确。最后,对Iot-EDF调度机制在OSv Unikernel上进行了实现和验证,实验结果表明,Iot-EDF在降低内存开销的情况下,可以有效提升OSv Unikernel任务的执行成功率,相比于未经改动的OSv Unikernel,对重要任务的完成率提升达21%。此外,还模拟实际应用场景,对系统的吞吐量进行测试,与原始的OSv Unikernel相比,吞吐量提升了30%,进一步证明了Iot-EDF调度机制的有效性。
关键词:  物联网  Unikernel  调度机制  时钟同步  任务完成率
DOI:
投稿时间:2024-02-20修订日期:2024-06-24
基金项目:中国科学院战略性先导科技专项基金资助项目(No.XDA-Y01-01, No.XDC02010600)
Iot-EDF: Unikernel-based Task Scheduling Method for the Internet of Things
DongBonan, YangQiuSong, LiMingShu
(Institute of Software Chinese Academy of Sciences)
Abstract:
Unikernel, recognized as a leading-edge technology in the field of virtualization, is widely utilized in the Internet of Things(IoT) environment due to its rapid startup, low resource consumption, and high security attributes. However, it is currently observed that Unikernel lacks a dynamic scheduling mechanism designed according to the unique charac-teristics of different tasks, which is crucial for ensuring efficiency in large-scale IoT task scheduling scenarios. To address this issue, the task characteristics of Unikernel in the IoT environment are first summarized. Subsequently, without altering the structural features of Unikernel, a new scheduling mechanism for the Unikernel-based IoT setting, named IoT-EDF, along with its mathematical model, is proposed. This mechanism, by categorizing tasks based on their importance and deadlines, not only significantly enhances the scheduling efficiency within the IoT environment but also prevents the loss of critical data. Moreover, it enables the overall situation of task execution to be predicted through the mathematical model, thereby reducing the risks associated with task execution failures. Simultaneously, based on the Network Time Protocol(NTP), a clock synchronization method suitable for the Unikernel IoT scenario is proposed. This method, without compromising the lightweight characteristics of Unikernel, adjusts the timestamps in data packets sent from various locations and generates a time difference table. By doing so, the issue of global clock synchronization is addressed, thereby ensuring the accuracy of deadline calculations within the IoT-EDF scheduling mechanism. Finally, the IoT-EDF scheduling mechanism is implemented and validated on the OSv Unikernel. Exper-imental results indicate that IoT-EDF can effectively enhance the success rate of task execution on OSv Unikernel while reducing memory overhead. Compared to the unmodified OSv Unikernel, the completion rate for critical tasks is increased by 21%. Moreover, the throughput of the system is tested in simulated real-world scenarios, demonstrat-ing a 30% improvement compared to the original OSv Unikernel, further proving the effectiveness of the IoT-EDF scheduling mechanism in Unikernel under IoT scenarios.
Key words:  Internet of Things  unikernel  scheduling mechanism  clock synchronization  task completion rate