WO2021179480A1 - 一种面向边缘云的多级时钟同步方法 - Google Patents

一种面向边缘云的多级时钟同步方法 Download PDF

Info

Publication number
WO2021179480A1
WO2021179480A1 PCT/CN2020/098659 CN2020098659W WO2021179480A1 WO 2021179480 A1 WO2021179480 A1 WO 2021179480A1 CN 2020098659 W CN2020098659 W CN 2020098659W WO 2021179480 A1 WO2021179480 A1 WO 2021179480A1
Authority
WO
WIPO (PCT)
Prior art keywords
clock
server
synchronization
client
network
Prior art date
Application number
PCT/CN2020/098659
Other languages
English (en)
French (fr)
Inventor
巩玉旺
陈乃阔
刘毅枫
Original Assignee
山东超越数控电子股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 山东超越数控电子股份有限公司 filed Critical 山东超越数控电子股份有限公司
Publication of WO2021179480A1 publication Critical patent/WO2021179480A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04J—MULTIPLEX COMMUNICATION
    • H04J3/00—Time-division multiplex systems
    • H04J3/02—Details
    • H04J3/06—Synchronising arrangements
    • H04J3/0635—Clock or time synchronisation in a network
    • H04J3/0638—Clock or time synchronisation among nodes; Internode synchronisation
    • H04J3/0658—Clock or time synchronisation among packet nodes
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04J—MULTIPLEX COMMUNICATION
    • H04J3/00—Time-division multiplex systems
    • H04J3/02—Details
    • H04J3/06—Synchronising arrangements
    • H04J3/0635—Clock or time synchronisation in a network
    • H04J3/0638—Clock or time synchronisation among nodes; Internode synchronisation
    • H04J3/0658—Clock or time synchronisation among packet nodes
    • H04J3/0661—Clock or time synchronisation among packet nodes using timestamps
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L69/00—Network arrangements, protocols or services independent of the application payload and not provided for in the other groups of this subclass
    • H04L69/16—Implementation or adaptation of Internet protocol [IP], of transmission control protocol [TCP] or of user datagram protocol [UDP]
    • H04L69/164—Adaptation or special uses of UDP protocol

Definitions

  • the invention relates to the field of cloud computing virtualization, in particular to a multi-level clock synchronization method facing edge cloud.
  • Edge cloud computing refers to the use of an open cloud platform that integrates network, computing, storage, and application core capabilities on the side close to the source of things or data to provide services nearby.
  • the edge cloud is composed of server nodes distributed in the same area, specifically processing service requests from users in the region, and quickly and flexibly providing cloud computing services to users.
  • the edge cloud generally runs in actual scenarios such as on-board vehicles and temporary command posts (such as disaster relief). In actual scenarios, there are generally multiple edge clouds in a certain area, and they are interconnected through the network.
  • the technical problem to be solved by the present invention is to provide a multi-level clock synchronization method for edge clouds, which can simultaneously realize clock synchronization between edge clouds, inside edge clouds, and between physical machines and virtual machines.
  • the technical solution adopted by the present invention is: a multi-level clock synchronization method for edge clouds.
  • the clock synchronization described in this method includes three levels, namely, clock synchronization between edge clouds and edge cloud synchronization.
  • the internal clock synchronization of the cloud, the synchronization of the clocks of the physical machine and the virtual machine, the central architecture between the edge cloud and the edge cloud is used for network timing.
  • Each edge cloud selects a server as the server for clock synchronization, and other internal nodes serve as Clock synchronization of the client, the client and the server send and receive network messages to realize the network timing, the edge cloud chooses a certain edge cloud server as the top-level clock source, and other edge cloud servers as the top-level clock source On the client side, the top-level clock source and the client side transmit and receive network messages to implement network timing. Finally, the physical machine synchronizes the clock to the virtual machine.
  • the clock synchronization process between the physical machine and the virtual machine is as follows: S11, periodically execute according to the detection cycle, first obtain the physical machine time and the virtual machine time; S12), calculate the clock error between the physical machine and the virtual machine, and determine the clock Whether the error exceeds the threshold of forced synchronization, if yes, perform forced synchronization, and then wait for the next detection cycle, if not, proceed to the next step; S13), determine whether the clock error exceeds the maximum allowable error threshold, if it does not exceed this threshold, then Do not synchronize, wait for the next detection cycle, otherwise proceed to the next step; S14), calculate the correction step DeltaTime/(N-1), DeltaTime is the clock error calculated in step S12, N is the maximum number of iterations, and then use the correction step Perform N-1 iteration synchronization for the cycle to correct the virtual machine time; S15), after loop iteration N-1 times, recalculate the clock error between the physical machine and the virtual machine self-check for the Nth time, and confirm whether the maximum
  • the client After the client receives the network message, it parses the correction time and sets it to the local; S29), the client encapsulates the local time after correcting the time Send to the server for confirmation; S210). After receiving the confirmation from the client, the server calculates the clock error. If it is less than the maximum allowable error threshold, the synchronization ends, otherwise the next round of synchronization starts.
  • the network timing is based on the UDP network protocol.
  • the forced synchronization and iterative synchronization between the physical machine and the virtual machine are realized through inter-process communication.
  • the server clock of each edge cloud comes from the Beidou device, the Internet clock, or the upper-level clock server.
  • the present invention can provide a complete clock synchronization scheme for edge cloud scenarios, including: synchronization between edge clouds, inside edge clouds, and physical machines and virtual machines.
  • the UDP protocol-based network timing method is adopted between the edge clouds and within the edge cloud, which can efficiently synchronize clocks through the UDP protocol, and achieve high-precision clock synchronization through the method of delay compensation.
  • Real-time synchronization between the physical machine and the virtual machine ensures the clock accuracy of the virtual machine.
  • Figure 1 is a diagram of the synchronization architecture of multi-level clocks
  • Figure 2 is a flowchart of clock synchronization between a physical machine and a virtual machine
  • FIG. 3 is a flowchart of network timing.
  • This embodiment 1 discloses a multi-level clock synchronization method for edge clouds. This method can simultaneously realize clock synchronization between edge clouds, inside edge clouds, and between physical machines and virtual machines.
  • Edge cloud A is selected as the top clock source in multiple edge clouds
  • server A1 is selected as the clock synchronization server in edge cloud A
  • A2 and A3 are used as the local clock.
  • Synchronization client, edge cloud B, C as the lower-level clock client, of which B1 and C1 as their internal clock synchronization server, where A1 is responsible for synchronizing to A2, A3 and lower-level B1 and C1, and then B1 is responsible for synchronization
  • the clock goes to the B2 and B3 clients.
  • C1 is responsible for synchronizing the clock to the C2 and C3 clients.
  • each server synchronizes the clock to the virtual machine.
  • A1 server, A2 server, and A3 server are all physical machines running on virtual machines.
  • B1 server, B2 server, and B3 server are physical machines running on virtual machines
  • the C3 server is a physical machine on which a virtual machine runs.
  • the clock synchronization between the physical machine and the virtual machine proposed in this embodiment refers to the clock synchronization between the physical machine in the edge cloud and the virtual machine running on it. Since the clock accuracy of the virtual machine cannot be guaranteed, it must be performed by the physical machine.
  • Real-time synchronization there are two synchronization methods: the first group is forced synchronization, when the clock error of the virtual machine and the physical machine is found to exceed the threshold, it will be synchronized immediately; the second type is iterative synchronization, and the clock error is performed according to a fixed cycle Detection, such as detecting the clock error once every 10 seconds, if the clock error is less than the threshold value, no clock synchronization is performed, if it is greater than the threshold value, it is corrected, and the synchronization is stopped when the clock error is less than the threshold value.
  • Step1 Perform periodically according to the detection cycle.
  • the virtual machine calls the local API to obtain the local system time, and then obtains the physical machine time through the virtual channel;
  • Step2 Calculate the clock error DeltaTime between the physical machine and the virtual machine, and determine whether it exceeds the mandatory synchronization threshold (FTH), if so, perform the mandatory synchronization, and then wait for the next detection cycle, if not, continue to the next step;
  • FTH mandatory synchronization threshold
  • Step3 Determine whether the maximum allowable error threshold (MAXTH) is exceeded, if it does not exceed this threshold, then no synchronization will be performed, and the next detection cycle will be waited, otherwise iterative synchronization will start;
  • MAXTH maximum allowable error threshold
  • Step4 After the iterative synchronization starts, perform N-1 iterations according to the maximum number of iterations N, and each time the correction time step is DeltaTime/(N-1);
  • Step5 After looping for N-1 times, recalculate DeltaTime for the Nth time, and confirm whether it is within the MAXTH range. If it does not exceed this threshold, the iterative synchronization is ended, otherwise, continue to the next step;
  • Step6 Perform mandatory synchronization in the last iteration, and end the current round of clock synchronization.
  • the forced synchronization and iterative synchronization between the physical machine and the virtual machine are implemented through inter-process communication.
  • This implementation proposes a network timing method based on UDP protocol to achieve clock synchronization between edge clouds and within edge clouds.
  • the C/S architecture mode is adopted.
  • the server provides time to all clients once in a fixed period (for example, 1 minute). Due to the clock data There will be a certain delay in transmission through the network, especially between edge clouds, the delay may be relatively large, otherwise high-precision synchronization cannot be achieved.
  • the invention adopts the method of delay compensation to realize high-precision clock synchronization, that is, network detection is first performed through UDP message, the network delay is calculated, and finally the clock data is added to the network delay to serve as an accurate clock.
  • Step1 The server first initiates the process of detecting network delay
  • Step2 Encapsulate the local timestamp T1 into a UDP message and send it to the client;
  • Step3 After the client receives the server message, it records the current reception time T2 of the client;
  • Step4 Before the client sends back to the server, encapsulate T1, T2 and the sending time T3 into UDP packets for sending;
  • Step5 After the server receives the message feedback from the client, it records the receiving time T4;
  • Step7 Verify the rationality of the network delay.
  • the one-way network delay is at the millisecond level. Set the maximum one-way network delay to MAXND. If ND>MAXND, you need to re-probe. If ND ⁇ MAXND, the network delay Time valid
  • Step8 The server starts to execute the network timing process
  • Step9 The server modifies the time of the client: server local time + network delay, encapsulates it into a UDP message and sends it;
  • Step10 After the client receives the message, it parses out the correction time and sets it locally;
  • Step11 After the client corrects the time, it encapsulates the local time and sends it to the server for confirmation;
  • Step12 After the server receives the confirmation from the client, it calculates the clock error. If it is less than the maximum allowable error threshold (MAXTH), the synchronization ends, otherwise the next round of synchronization starts.
  • MAXTH maximum allowable error threshold
  • the server clock of each edge cloud comes from a Beidou device, an Internet clock, or an upper-level clock server.
  • This method can simultaneously realize clock synchronization between edge clouds, within edge clouds, and between physical machines and virtual machines, and provides a complete clock synchronization solution for edge cloud scenarios. And through delay compensation, real-time synchronization between the physical machine and the virtual machine to ensure the accuracy of clock synchronization.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Computer Security & Cryptography (AREA)
  • Electric Clocks (AREA)
  • Synchronisation In Digital Transmission Systems (AREA)

Abstract

本发明公开一种面向边缘云的多级时钟同步方法，能够实现边缘云之间、边缘云内部、物理机与虚拟机之间的三级时钟同步，其中边缘云之间与边缘云内部采用中心架构进行网络授时，每个边缘云内部选择一台服务器作为时钟同步的服务端，内部其他节点作为时钟同步的客户端，边缘云之间选择某个边缘云的服务端作为顶级时钟源，其他边缘云的服务端作为顶级时钟源的客户端，客户端与服务端之间进行网络报文收发实现网络授时，最后，由物理机将时钟同步到虚拟机。本发明能够同时实现边缘云之间、边缘云内部以及物理机与虚拟机之间的时钟同步。

Description

一种面向边缘云的多级时钟同步方法 技术领域
本发明涉及云计算虚拟化领域，具体涉及一种面向边缘云的多级时钟同步方法。
背景技术
边缘云计算是指在靠近物或数据源头的一侧，采用网络、计算、存储、应用核心能力为一体的开放云平台，就近提供服务。边缘云由分布在同一地区的服务器节点组成，具体处理本地区用户的服务请求，迅速并弹性地向用户提供云计算服务。边缘云一般运行于车载、临时指挥所(如：救灾)等实际场景下。在实际场景下，一般一定区域内会存在多个边缘云，之间通过网络进行互联互通。对于业务或应用软件来说，比如救灾指挥所，消息和指令需在多个边缘云之间进行准确传递，要实现协同工作，就必然要解决时钟同步问题，否则可能会造成消息和指令的不准确性，影响指挥所的协同调度工作。
边缘云场景下的时钟同步，目前存在三个问题：第一，如何实现边缘云之间的时钟同步；第二，如何实现边缘云的内部时钟同步；第三，如何实现物理机和虚拟机之间的时钟同步。
发明内容
本发明要解决的技术问题是提供一种面向边缘云的多级时钟同步方法，能够同时实现边缘云之间、边缘云内部以及物理机与虚拟机之间的时钟同步。
为了解决所述技术问题，本发明采用的技术方案是：一种面向边缘云的多级时钟同步方法，本方法所述的时钟同步包括三个级别，分别为边缘云之间的时钟同步、边缘云内部的时钟同步、物理机和虚拟机的时钟同步，边缘云之间与边缘云内部采用中心架构进行网络授时，每个边缘云内部选择一台服务器作为时钟同步的服务端，内部其他节点作为时钟同步的客户端，客户端与服务端之间进行网络报文收发实现网络授时，边缘云之间选择某个边缘云的服务端作为顶级时钟源，其他边缘云的服务端作为顶级时钟源的客户端，顶级时钟源与其客户端之间进行网络报文收发实现网络授时，最后，由物理机将时钟同步到虚拟机。
进一步的，物理机与虚拟机之间的时钟同步过程为：S11、按照检测周期定期执行，首先获取物理机时间和虚拟机时间；S12)、计算物理机与虚拟机的时钟误差，判断该时钟误差是否超过强制同步的阈值，若是，则进行强制同步，然后等待下一检测周期，若否，则执行下一步；S13)、判断时钟误差是否超过最大允许误差阈值，若未超过此阈值，则不进行同步，等待下一检测周期，否则执行下一步；S14)、计算修正步长DeltaTime/(N-1)， DeltaTime为步骤S12计算的时钟误差，N为最大迭代次数，然后以修正步长为周期进行N-1次迭代同步从而修正虚拟机时间；S15)、循环迭代N-1次后，第N次重新计算物理机与虚拟机自检的时钟误差，并确认是否达到最大允许误差阈值以内，若未超过此阈值，则结束迭代同步，否则强制同步物理机与虚拟机的时间，并结束本轮时钟同步。
进一步的，网络授时的过程为：S21)、服务端首先发起探测网络延时流程，将本地时间戳T1封装成网络报文发送到客户端；S22)、客户端接收到服务端报文后，记录客户端的当前接收时间T2；S23)、客户端发回服务端之前，将T1、T2和发送时间T3封装成网络报文进行发送；S24)、服务端接收到客户端反馈的消息后，记录接收时间T4；S25)、计算网络延时ND＝(T2-T1+T4-T3)/2；S26)、对网络延时进行合理性验证，设置单向网络延时最大值为MAXND，若ND>MAXND，则需重新进行探测，若ND<MAXND，则网络延时有效，执行下一步；S27)、服务端开始执行网络授时流程，将客户端修正时间封装成网络报文并发送至客户端，客户端修正时间为服务器本地时间和网络延时的和；S28)、客户端接收到网络报文后，解析出修正时间并设置到本地；S29)、客户端修正时间后将本地时间封装后发送至服务端进行确认；S210)、服务端接收到客户端发来的确认之后，计算时钟误差，若小于最大允许误差阈值，则结束同步，否则开始下一轮同步。
进一步的，网络授时基于UDP网络协议进行。
进一步的，物理机与虚拟机之间的强制同步和迭代同步通过进程间通信实现。
进一步的，每个边缘云的服务端时钟来自于北斗设备、互联网时钟或者上一级时钟服务器。
本发明的有益效果：本发明能够为边缘云场景提供完整的时钟同步方案，包括：边缘云之间、边缘云内部以及物理机和虚拟机的同步。边缘云之间与边缘云内部采用基于UDP协议的网络授时方法，能够通过UDP协议进行高效地时钟同步，并通过延时补偿的方法实现高精度时钟同步。物理机与虚拟机之间进行实时同步，从而保证虚拟机的时钟精度。
附图说明
图1为多级时钟的同步架构图；
图2为物理机和虚拟机时钟同步的流程图；
图3为网络授时的流程图。
具体实施方式
下面结合附图与具体实施例对本发明作进一步的说明。
实施例1
本实施例1公开一种面向边缘云的多级时钟同步方法，本方法能够同时实现边缘云之间、边缘云内部以及物理机与虚拟机之间的时钟同步。
如图1所示，为多级时钟的同步架构图，在多个边缘云中选择边缘云A作为顶级时钟源，边缘云A中选择A1服务器作为时钟同步服务端，A2、A3作为本级时钟同步客户端，边缘云B、C作为下级时钟客户端，其中B1、C1作为各自内部的时钟同步服务端，其中A1负责同步到本级的A2、A3以及下级的B1和C1，然后B1负责同步时钟到B2、B3客户端，C1负责同步时钟到C2、C3客户端，最后各服务器再将时钟同步到虚拟机上。图1中，A1服务器、A2服务器、A3服务器均是运行在其上虚拟机的物理机，同样，B1服务器、B2服务器、B3服务器是运行在其上虚拟机的物理机，C1服务器、C2服务器、C3服务器是运行在其上虚拟机的物理机。
本实施例提出的物理机和虚拟机之间的时钟同步是指边缘云中的物理机和运行其上的虚拟机之间的时钟同步，由于虚拟机的时钟精度无法保障，必须由物理机进行实时同步，同步方式分为两种：第一组为强制同步，当发现虚拟机的时钟和物理机的时钟误差超过阈值时，则立即同步；第二种为迭代同步，按照固定周期进行时钟误差检测，比如10秒钟检测一次时钟误差，若小于阈值则不进行时钟同步，若大于阈值则进行修正，知道时钟误差小于阈值停止同步。
如图2所示，物理机和虚拟机的时钟同步的具体过程为：
Step1：按照检测周期定期执行，首先虚拟机调用本地API获取本地系统时间，然后通过虚拟通道获取物理机时间；
Step2：计算物理机与虚拟机的时钟误差DeltaTime，判断是否超过强制同步的阈值(FTH)，若是则进行强制同步，然后等待下一检测周期，若否则继续执行下一步；
Step3：判断是否超过最大允许误差阈值(MAXTH)，若未超过此阈值，则不进行同步，等待下一检测周期，否则开始迭代同步；
Step4：迭代同步开始后，根据最大迭代次数N，进行N-1次迭代，每次修正时间步长为DeltaTime/(N-1)；
Step5：循环迭代N-1次后，第N次重新计算DeltaTime，并确认是否达到MAXTH范围以内，若未超过此阈值，则结束迭代同步，否则继续执行下一步；
Step6：最后一次迭代进行强制同步，并结束本轮时钟同步。
本实施例中，物理机与虚拟机之间的强制同步和迭代同步通过进程间通信实现。
本实施提出基于UDP协议的网络授时方法实现边缘云之间和边缘云内部的时钟同 步，采用C/S架构模式，服务端按照固定周期(比如1分钟)给所有客户端授时一次，由于时钟数据通过网络传输会有一定的延时，特别是在边缘云之间，可能延迟相对较大，否则无法达到高精度的同步。本发明采用延时补偿的方法来实现高精度时钟同步，即先通过UDP报文进行网络探测，计算网络延时，最后再将时钟数据加上网络延时来作为准确的时钟。
如图3所示，网络授时的具体过程为：
Step1：服务端首先发起探测网络延时流程；
Step2：将本地时间戳T1封装成UDP报文发送到客户端；
Step3：客户端接收到服务端报文后，记录客户端的当前接收时间T2；
Step4：客户端发回服务端之前，将T1、T2和发送时间T3封装成UDP报文进行发送；
Step5：服务端接收到客户端反馈的消息后，记录接收时间T4；
Step6：计算网络延时ND＝(T2-T1+T4-T3)/2；
Step7：对网络延时进行合理性验证，一般单向网络延迟为毫秒级别，设置单向网络延时最大值为MAXND，若ND>MAXND，则需重新进行探测，若ND<MAXND,则网络延时有效；
Step8：服务端开始执行网络授时流程；
Step9：服务端将客户端修正时间：服务端本地时间+网络延时，封装成UDP报文并发送；
Step10：客户端接收到报文之后，解析出修正时间，并设置到本地；
Step11：客户端修正时间后将本地时间封装后发送至服务端进行确认；
Step12：服务端接收到客户端发来的确认之后，计算时钟误差，若小于最大允许误差阈值(MAXTH)，则同步结束，否则开始下一轮同步。
本实施例中，每个边缘云的服务端时钟来自于北斗设备、互联网时钟或者上一级时钟服务器。
本方法能够同时实现边缘云之间、边缘云内部、物理机与虚拟机之间的时钟同步，为边缘云场景提供完整的时钟同步方案。并且通过延时补偿、物理机与虚拟机之间实时同步的方式保证时钟同步的精度。
以上描述的仅是本发明的基本原理和优选实施例，本领域技术人员根据本发明做出的改进和替换，属于本发明的保护范围。

Claims (6)

  1. 一种面向边缘云的多级时钟同步方法，其特征在于：本方法所述的时钟同步包括三个级别，分别为边缘云之间的时钟同步、边缘云内部的时钟同步、物理机和虚拟机的时钟同步，边缘云之间与边缘云内部采用中心架构进行网络授时，每个边缘云内部选择一台服务器作为时钟同步的服务端，内部其他节点作为时钟同步的客户端，客户端与服务端之间进行网络报文收发实现网络授时，边缘云之间选择某个边缘云的服务端作为顶级时钟源，其他边缘云的服务端作为顶级时钟源的客户端，顶级时钟源与其客户端之间进行网络报文收发实现网络授时，最后，由物理机将时钟同步到虚拟机。
  2. 根据权利要求1所述的面向边缘云的多级时钟同步方法，其特征在于：物理机与虚拟机之间的时钟同步过程为：S11、按照检测周期定期执行，首先获取物理机时间和虚拟机时间；S12)、计算物理机与虚拟机的时钟误差，判断该时钟误差是否超过强制同步的阈值，若是，则进行强制同步，然后等待下一检测周期，若否，则执行下一步；S13)、判断时钟误差是否超过最大允许误差阈值，若未超过此阈值，则不进行同步，等待下一检测周期，否则执行下一步；S14)、计算修正步长DeltaTime/(N-1)，DeltaTime为步骤S12计算的时钟误差，N为最大迭代次数，然后以修正步长为周期进行N-1次迭代同步从而修正虚拟机时间；S15)、循环迭代N-1次后，第N次重新计算物理机与虚拟机自检的时钟误差，并确认是否达到最大允许误差阈值以内，若未超过此阈值，则结束迭代同步，否则强制同步物理机与虚拟机的时间，并结束本轮时钟同步。
  3. 根据权利要求1所述的面向边缘云的多级时钟同步方法，其特征在于：网络授时的过程为：S21)、服务端首先发起探测网络延时流程，将本地时间戳T1封装成网络报文发送到客户端；S22)、客户端接收到服务端报文后，记录客户端的当前接收时间T2；S23)、客户端发回服务端之前，将T1、T2和发送时间T3封装成网络报文进行发送；S24)、服务端接收到客户端反馈的消息后，记录接收时间T4；S25)、计算网络延时ND＝(T2-T1+T4-T3)/2；S26)、对网络延时进行合理性验证，设置单向网络延时最大值为MAXND，若ND>MAXND，则需重新进行探测，若ND<MAXND，则网络延时有效，执行下一步；S27)、服务端开始执行网络授时流程，将客户端修正时间封装成网络报文并发送至客户端，客户端修正时间为服务器本地时间和网络延时的和；S28)、客户端接收到网络报文后，解析出修正时间并设置到本地；S29)、客户端修正时间后将本地时间封装后发送至服务端进行确认；S210)、服务端接收到客户端发来的确认之后，计算时钟误差，若小于最大允许误差阈值，则结束同步，否则开始下一轮同步。
  4. 根据权利要求1或3所述的面向边缘云的多级时钟同步方法，其特征在于：网络授时基 于UDP网络协议进行。
  5. 根据权利要求2所述的面向边缘云的多级时钟同步方法，其特征在于：物理机与虚拟机之间的强制同步和迭代同步通过进程间通信实现。
  6. 根据权利要求1所述的面向边缘云的多级时钟同步方法，其特征在于：每个边缘云的服务端时钟来自于北斗设备、互联网时钟或者上一级时钟服务器。
PCT/CN2020/098659 2020-03-09 2020-06-29 一种面向边缘云的多级时钟同步方法 WO2021179480A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010157183.3 2020-03-09
CN202010157183.3A CN111294135B (zh) 2020-03-09 2020-03-09 一种面向边缘云的多级时钟同步方法

Publications (1)

Publication Number Publication Date
WO2021179480A1 true WO2021179480A1 (zh) 2021-09-16

Family

ID=71028630

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/098659 WO2021179480A1 (zh) 2020-03-09 2020-06-29 一种面向边缘云的多级时钟同步方法

Country Status (2)

Country Link
CN (1) CN111294135B (zh)
WO (1) WO2021179480A1 (zh)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111294135B (zh) * 2020-03-09 2021-06-08 山东超越数控电子股份有限公司 一种面向边缘云的多级时钟同步方法
CN111901417B (zh) * 2020-07-28 2022-08-23 超越科技股份有限公司 一种基于车载边缘云环境的时间同步系统和实现方法
CN111835449A (zh) * 2020-07-30 2020-10-27 山东超越数控电子股份有限公司 一种基于ptp协议的局域网高精度时钟同步方法及装置
CN113055149B (zh) * 2021-02-20 2022-09-06 郑州中科集成电路与系统应用研究院 一种射频收发机级联系统下的时间同步和频率同步方法
CN112904932A (zh) * 2021-05-08 2021-06-04 鹏城实验室 时钟同步方法、板卡、计算机存储介质及终端设备
CN113691342A (zh) * 2021-09-13 2021-11-23 上海云轴信息科技有限公司 一种用于虚拟化平台时间同步的方法及系统

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110261917A1 (en) * 2010-04-21 2011-10-27 Lsi Corporation Time Synchronization Using Packet-Layer and Physical-Layer Protocols
CN106909197A (zh) * 2017-02-28 2017-06-30 郑州云海信息技术有限公司 一种虚拟化主机时间管理方法及虚拟化主机系统
CN111294135A (zh) * 2020-03-09 2020-06-16 山东超越数控电子股份有限公司 一种面向边缘云的多级时钟同步方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10623240B2 (en) * 2016-08-25 2020-04-14 Intel Corporation IoT solution sizing
US10630529B2 (en) * 2016-12-29 2020-04-21 Kodiak Networks, Inc. System and method for push-to-talk (PTT) in mobile edge computing (MEC)
EP3566421B1 (en) * 2017-01-09 2021-12-29 Nokia Technologies Oy Method, computer programme and apparatus for coordinated content delivery in multicast / broadcast networks
CN109981771A (zh) * 2019-03-22 2019-07-05 长安大学 一种基于5g技术的车路通信测试系统及测试方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110261917A1 (en) * 2010-04-21 2011-10-27 Lsi Corporation Time Synchronization Using Packet-Layer and Physical-Layer Protocols
CN106909197A (zh) * 2017-02-28 2017-06-30 郑州云海信息技术有限公司 一种虚拟化主机时间管理方法及虚拟化主机系统
CN111294135A (zh) * 2020-03-09 2020-06-16 山东超越数控电子股份有限公司 一种面向边缘云的多级时钟同步方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
LOGHIN DUMITREL; RAMAPANTULU LAVANYA; TEO YONG MENG: "On Understanding Time, Energy and Cost Performance of Wimpy Heterogeneous Systems for Edge Computing", 2017 IEEE INTERNATIONAL CONFERENCE ON EDGE COMPUTING (EDGE), 25 June 2017 (2017-06-25), pages 1 - 8, XP033151574, DOI: 10.1109/IEEE.EDGE.2017.10 *

Also Published As

Publication number Publication date
CN111294135B (zh) 2021-06-08
CN111294135A (zh) 2020-06-16

Similar Documents

Publication Publication Date Title
WO2021179480A1 (zh) 一种面向边缘云的多级时钟同步方法
US10158444B1 (en) Event-driven precision time transfer
US9252902B2 (en) Precision timing in a data over cable service interface specification (DOCSIS) system
US10341083B2 (en) System and methods for network synchronization
CA2767499C (en) A method of time synchronization of free running nodes in an avionics network
CN103929293B (zh) 非对称延迟的时间同步方法及系统
US9614819B2 (en) Method and apparatus for securing clock synchronization in a network
US9756153B2 (en) Method for improving accuracy in computation of one-way transfer time for network time synchronization
JP5518191B2 (ja) 光伝送網が時刻同期プロトコルをキャリングする方法及びシステム
US8644352B1 (en) System and method for accurate time sampling in presence of output delay
US20140247839A1 (en) Time synchronization in distributed network testing equipment
US20180220385A1 (en) Synchronizing time among two or more devices
WO2014117489A1 (zh) 时钟同步方法及设备
WO2017054554A1 (zh) 一种时钟同步方法、装置及通信设备
EP3738275B1 (en) Method and arrangement for deterministic delivery of data traffic over wireless connection
CN108155982A (zh) 一种时间戳处理方法及设备
US20210400610A1 (en) Time synchronization in integrated 5g wireless and time-sensitive networking systems
CN102195736A (zh) 无源光网络中时间同步的处理方法、系统及olt
WO2021004630A1 (en) Technique for determining a packet delay budget
WO2023098683A1 (zh) 一种配电终端实时时钟同步方法、同步系统及配电终端
US20190036591A1 (en) Link acquisition in wireless communication systems
WO2018120549A1 (zh) 以太网无源光网络时间戳处理方法和装置、存储介质
CN101631015A (zh) 网络授时时间标签标定方法
CN109379160A (zh) 一种时钟校准服务器及方法
TW201616839A (zh) 頻率校正方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20924953

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20924953

Country of ref document: EP

Kind code of ref document: A1