WO2023206600A1 - 一种面向主动网络遥感的低时延低开销路径部署方法 - Google Patents

一种面向主动网络遥感的低时延低开销路径部署方法 Download PDF

Info

Publication number
WO2023206600A1
WO2023206600A1 PCT/CN2022/091621 CN2022091621W WO2023206600A1 WO 2023206600 A1 WO2023206600 A1 WO 2023206600A1 CN 2022091621 W CN2022091621 W CN 2022091621W WO 2023206600 A1 WO2023206600 A1 WO 2023206600A1
Authority
WO
WIPO (PCT)
Prior art keywords
node
path
low
delay
nodes
Prior art date
Application number
PCT/CN2022/091621
Other languages
English (en)
French (fr)
Inventor
张华�
张鹏辉
Original Assignee
东南大学
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 东南大学 filed Critical 东南大学
Publication of WO2023206600A1 publication Critical patent/WO2023206600A1/zh

Links

Images

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00Arrangements for monitoring or testing data switching networks
    • H04L43/12Network monitoring probes
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/12Discovery or management of network topologies
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00Routing or path finding of packets in data switching networks
    • H04L45/34Source routing
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04QSELECTING
    • H04Q9/00Arrangements in telecontrol or telemetry systems for selectively calling a substation from a main station, in which substation desired apparatus is selected for applying a control signal thereto or for obtaining measured values therefrom
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02DCLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00Reducing energy consumption in communication networks
    • Y02D30/70Reducing energy consumption in communication networks in wireless communication networks

Definitions

  • the present invention relates to the technical field of network traffic remote sensing, and in particular to a low-latency and low-cost path deployment method for active network remote sensing.
  • In-band telemetry technology (INT), as a typical representative of network telemetry technology, completes network measurements by querying and collecting device internal status information during data packet forwarding. During the process of network measurement, INT will continuously inject telemetry information into forwarded data packets, and send telemetry data information to the control plane at the last hop of the forwarding path for centralized storage and analysis.
  • INT In-band telemetry technology
  • a key disadvantage of INT is the huge overhead caused by network traffic awareness. As each switch on the path continues to add information, the telemetry information carried by the data packets also continues to increase.
  • an in-band telemetry scheme that actively embeds source routing into INT probes and allows specified probe packets to be routed through the network.
  • the advantage of active in-band sensing is that it can stably provide the controller with full network visibility.
  • detection flows need to be actively generated to complete traffic information collection of the entire network topology, the path deployment of each detection flow is an important factor affecting the performance of the sensing system.
  • the Euler path deployment algorithm commonly used at present can generate a minimum number of detection flows based on the number of odd vertices in the topological network to cover the entire detection network, it does not take into account the delay difference on each detection flow path, resulting in actual problems. The huge delay in the situation affects the centralized control of the controller.
  • the Euler path deployment scheme ignores the different types and amounts of telemetry data caused by different routing devices carrying different services, thus causing unnecessary overhead in the process of traffic sensing, leading to Network system performance degrades.
  • the purpose of the present invention is to provide a low-latency and low-cost path deployment method for active network remote sensing to solve the technical problems mentioned in the background art.
  • the present invention further improves the deployment of detection flow paths based on the active in-band remote sensing routing mechanism.
  • Today as the types of network services become increasingly complex, the types of information that network control needs to obtain are also becoming increasingly diverse. Different routes often need to provide different types of data to the control plane due to different services they undertake. Therefore, it is obviously inappropriate to directly carry out path planning without considering the amount of data added by different routes.
  • the present invention considers the impact of the amount of data added by different routes on the overall traffic sensing system; on the other hand, the present invention performs a balanced restriction process on the delay of each detection flow path to control the path delay problem. into the path planning problem.
  • a low-latency and low-overhead path deployment method for active network remote sensing includes the following steps:
  • Step S1 Construct a detection flow path for a network topology, and gradually traverse the amount of data that needs to be added to the detection data packet for each node in the network topology through the detection flow path, where the detection flow is selected based on the selection strategy. the starting node;
  • Step S2 Perform judgment to determine whether the starting node selected in step S1 has an uncovered physical link. If it exists, perform step S3. If it does not exist, return to step S1 to continue selecting the starting node;
  • Step S3 Select the next node of the starting node based on the corresponding undetected physical link information of the starting node;
  • Step S4 Perform path delay detection to determine whether to continue adding nodes. If the delay test conditions are met, continue to add nodes. If the delay test conditions are not met, return to step S1;
  • Step S5 Add nodes and update current node information and network coverage information
  • Step S6 Perform a test to check whether full coverage of the network topology is completed. If it is not completed, return to step S2. If it is completed, make a judgment to determine whether the termination condition is met. If so, stop and complete all path planning tasks.
  • the adjacency matrix is initialized according to the network topology status, which specifically includes:
  • step S1 during the path planning process, the following selection strategy is used to select the starting node of the detection flow, including:
  • b(i) represents the amount of data that node i needs to add to the detection packet
  • L j is the set of links covered by the jth path, specifically expressed as Among them, N j represents the number of nodes in the j-th path.
  • step S2 the following method is used to determine whether a certain node has an uncovered physical link, including:
  • the path f k is temporarily expressed as is the i-th node in the k-th path f k ;
  • step S3 the next node of the starting node is selected by the following method, including:
  • b(i) represents the amount of data that node i needs to insert into the data packet.
  • step S4 specifically includes:
  • the current delay T k of the path is calculated based on the current information.
  • the delay expression is expressed as:
  • t(i,j) represents the delay information between node i and node j
  • T max is the maximum delay value that the system can accept
  • step S1 If the delay test conditions are met, continue to add nodes. Otherwise, abandon the addition and perform step S1.
  • the information that needs to be updated after adding the node specifically includes: the information of the node passed by the path fk ; the current delay information Tk of the path fk ; the current coverage link set L′ of the remote sensing system .
  • step S6 the termination condition is specifically: reaching the maximum running time limit.
  • the present invention designs a detection packet forwarding path planning scheme based on the information of each routing device telemetry project and uses the idea of a greedy selection algorithm, thereby reducing the overhead of the active in-band telemetry system.
  • the forwarding path length is controlled to achieve control of the detection system delay.
  • the in-band network telemetry system can obtain stable network-wide visibility, effectively reduce telemetry overhead and improve the latency performance of the telemetry system.
  • Figure 1 is a schematic structural diagram of the routing mechanism provided in Embodiment 1;
  • Figure 2 is a schematic flowchart of a low-latency, low-cost path deployment method for active network remote sensing provided in Embodiment 1.
  • this embodiment provides a low-latency and low-overhead path deployment method for active network remote sensing.
  • the routing mechanism adopted by this method is as shown in Figure 1. Specifically, in this mechanism, UDP data packets carry INT information, allocate a fixed size space as an SR label to store the port number to control the forwarding path of the data packet, and then allocate a variable size space as an INT label to store network remote sensing information.
  • the active in-band traffic awareness solution solves the problem of uncontrollable paths through source routing (SR) technology. That is, in this embodiment, the routing mechanism requires three logical routers: an INT generator, an INT forwarder, and an INT collector.
  • the INT generator is responsible for generating probe packets at the starting node of the probe flow that are forwarded along a specific routing path.
  • the INT forwarder is responsible for adding local traffic information to the detection packet and forwarding the packet to the next port according to the packet transmission path.
  • the INT collector is generally the last node on the detection flow path. It is responsible for forwarding the data packets to the control plane for analysis and processing after the detection flow completes the collection of traffic information.
  • the routing mechanism adopted in this embodiment is that it can generate a detection flow covering the entire network topology, thereby stably obtaining network status information of the entire network. This is a very important advantage for traffic awareness in today's large-scale networks.
  • Step S1 Initialize the adjacency matrix according to the network topology state
  • step S1 includes:
  • V ⁇ i
  • i ⁇ V is the physical node with serial number i in the monitoring network
  • E ⁇ e(i,j)
  • i,j ⁇ V ⁇ is the set of direct physical links of nodes, where e(i,j) ⁇ E is the physical link state between node i and node j. If there is a connected physical link between node i and node j, then e(i,j) exists; otherwise e(i,j) does not exist.
  • the adjacency matrix that stores graph information can be initialized as:
  • Step S2 Create a new detection flow path, gradually traverse the amount of data that needs to be added to the detection packets of each node in the network topology, and select the starting node of the detection flow according to the selection strategy;
  • the following selection strategy is used to select the starting node of the detection flow, including:
  • the selection strategy can be expressed as:
  • b(i) represents the amount of data that node i needs to add to the detection packet
  • L j is the set of links covered by the jth path, specifically expressed as Among them, N j represents the number of nodes in the j-th path.
  • Step S3 If all physical links of the current node have been covered, perform step S2;
  • the current node The conditions for the existence of physical links that have not been covered can be specifically described as:
  • the path f k can be temporarily expressed as is the i-th node in the k-th path f k .
  • the set of connected nodes can be expressed by formula (4):
  • Step S4 If the current node has undetected physical link information based on the current node, select the next node to be added based on the selection strategy;
  • the selection strategy can be described as: b(i) represents the amount of data that node i needs to insert into the data packet.
  • Step S5 Perform path delay detection to determine whether to continue adding nodes. If not, perform step S2;
  • path delay detection needs to be performed.
  • the specific steps for path delay detection are as follows:
  • the delay expression can be expressed as:
  • t(i,j) represents the delay information between node i and node j.
  • T max is the maximum delay value that the system can accept.
  • step S2 If the delay test conditions are met, you can continue to add nodes Otherwise, abandon the addition and perform step S2.
  • Step S6 Add nodes and update current node information and network coverage information
  • the information that needs to be updated after adding a node specifically includes: a. Information about the nodes that path f k passes through; b. Current delay information T k of path f k ; c , the current coverage link set L′ of the remote sensing system.
  • Step S7 Check whether full coverage of the network topology is completed. If not, execute step S3;
  • full network coverage judgment needs to be performed.
  • Step S8 If the termination conditions are met, stop and complete all path planning tasks.
  • the termination condition for completing path planning is:
  • the network remote sensing system completes full network coverage.
  • the maximum running time limit is reached. For situations where the network is too large or too complex to provide full network coverage, avoid wasting unnecessary resources.
  • the present invention proposes a low-latency and low-cost path deployment method for active network remote sensing.
  • This method can generate a detection flow path deployment plan for an active in-band remote sensing system with high efficiency.
  • This method can be used in Effectively reduce the delay and overhead of the remote sensing system while ensuring telemetry visibility.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

本发明公开了一种面向主动网络遥感的低时延低开销路径部署方法,包括:创建探测流路径,并根据初始节点选择策略确定探测流起点;随后根据路径时延要求设定节点添加策略,并基于添加节点更新路径信息;实现全网络覆盖后,完成全部路径部署并输出路径结果。本发明提出的方法可以以较高的效率生成主动带内遥感系统的探测流路径部署方案,利用此方法可以在保证遥测可见性的前提下有效地降低遥感系统的时延和开销。

Description

一种面向主动网络遥感的低时延低开销路径部署方法 技术领域
本发明涉及网络流量遥感技术领域,特别是涉及一种面向主动网络遥感的低时延低开销路径部署方法。
背景技术
由于超大规模数据中心网络的出现,以及部署的各个种类服务不断增加,细粒度的、全网络范围的可见性对进行大规模数据中心网和骨干网的拥塞控制、故障排除和容量规划等工作至关重要。然而传统的流量感知方案需要在控制平面与数据平面之间频繁交互,更由于设备能力的限制,传统的流量感知方案已经无法满足目前大规模数据中心网络对流量感知细粒度、低延迟、高覆盖范围的要求。
如今,网络遥测技术已经发展成为了用来收集网络数据的热门技术。相较于传统的流量感知技术,网络遥测技术具有更好的网络可见性和延展性。带内遥测技术(INT)作为网络遥测技术的典型代表,它通过在数据包转发的过程中查询并收集设备内部状态信息来完成网络测量。INT在进行网络测量的过程中会不断向转发的数据包中注入遥测信息,并在转发路径的最后一跳发送遥测数据信息到控制平面进行集中存储和分析。INT有一个关键性的缺点是网络流量感知所带来的巨大开销。随着路径上各个交换机不断地添加信息,数据包所携带的遥测信息也不断增加,特别是在转发路径较长的情况下,数据包携带的数据量随着不断转发而增大,给后面交换机之间的转发带来了很大的开销。同时,随着软件定义网络的不断发展,控制层面越来越需要获取整个网络拓扑的可见性,获得实时的整体流量状态信息对于进行网络控制至关重要。
在主动带内感知方案中,一种主动将源路由嵌入到INT探测中,并允许指定探测包通过网络路由的带内遥测方案。主动带内感知的优势在于可以为控制器稳定提供全网络的可见性,但是由于需要主动生成探测流来完成整个网络拓扑的流量信息采集,各个探测流的路径部署是影响感知系统性能的重要因素。目前通常采用的欧拉路径部署算法,虽然可以根据拓扑网络奇数顶点的个数生成最小数量的探测流来覆盖整个探测网络,但是由于没有考虑各个探测流路径上的时延差距,从而导致在实际情况中巨大的时延影响了控制器的集中控制。此外,欧拉路径部署方案在部署的过程中忽略了不同路由设备由于搭载的服务不同而导致的被遥测数据种类和数据量不同,从而在流量感知的过程中造成了不必要的开销,导致了网络系统的性能下降。
发明内容
有鉴于此,本发明的目的在于提供一种面向主动网络遥感的低时延低开销路径部署方法,用以解决背景技术中提及的技术问题。本发明在主动带内遥感路由机制的基础上进一步完善了对探测流路径的部署。在网络服务种类日益复杂的今天,网络控制需要获得的信息种类也日益繁多,不同的路由往往因为承担服务的不同而需要向控制平面提供不同种类的数据。因此,不考虑不同路由所需添加的数据量直接进行路径规划显然是不合时宜的。一方面,本发明考虑了不同路由所需添加的数据量对整体流量感知系统带来的影响;另一方面,本发明对各个探测流路径的时延做了均衡限制处理将路径时延控制问题带入了路径规划问题中。
为了实现上述目的,本发明采用如下技术方案:
一种面向主动网络遥感的低时延低开销路径部署方法,所述方法包括如下步骤:
步骤S1、针对一网络拓扑,构建一探测流路径,通过该探测流路径对该网络拓扑中各个节点,其需添加入探测数据包的数据量进行逐步遍历,其中,基于选择策略选择该探测流的起始节点;
步骤S2、执行判断,判断步骤S1中选择的起始节点,其是否存在未覆盖的物理链路,若存在,则执行步骤S3,若不存在,则回到步骤S1继续选择起始节点;
步骤S3、基于该起始节点其对应的未检测物理链路信息,选择该起始节点的下一节点;
步骤S4、执行路径时延检测以判断是否继续添加节点,若满足时延检验条件,则继续添加节点,若不满足时延检验条件,则回到步骤S1中;
步骤S5、添加节点,并更新当前节点信息和网络覆盖信息;
步骤S6、执行检测,检测是否完成网络拓扑全覆盖,若没有完成,则回到步骤S2,若已完成,则进行判断,判断是否满足终止条件,若满足,则停止并完成所有路径规划任务。
进一步的,在构建探测流路径之前,先根据网络拓扑状态初始化邻接矩阵,其具体包括:
将探测网络定义为一个无向物理图,用G=(V,E)表示;其中,V={i|i=1,…,n}为物理节点的集合,i∈V为监测网络中序号为i物理节点;E={e(i,j)|i,j∈V}为节点直接物理链路的集合,其中e(i,j)∈E为节点i和节点j之间的物理链路状态;
如果节点i和节点j之间有连通的物理链路,则e(i,j)存在;否则e(i,j)不存在;存储图信息的邻接矩阵初始化为:
Figure PCTCN2022091621-appb-000001
进一步的,在步骤S1中,在路径规划的过程中,采用如下的选择策略来选择探测流的起始节点,包括:
假设部署第k条路径f k时,
Figure PCTCN2022091621-appb-000002
为路径f k中第一个节点的序号,则
Figure PCTCN2022091621-appb-000003
的选择策略表述为:
Figure PCTCN2022091621-appb-000004
在公式(2)中,b(i)表示节点i需要向探测数据包中添加的数据量;
其中,该
Figure PCTCN2022091621-appb-000005
还必须拥有一条不在系统已覆盖链路集合L'的物理链路,则系统已覆盖链路集合L'表示为:
Figure PCTCN2022091621-appb-000006
在公式(3)中,L j是第 j条路径覆盖的链路集合,具体表示为
Figure PCTCN2022091621-appb-000007
其中,N j表示第j条路径中的节点数量。
进一步的,在所述步骤S2中,通过如下的方法来判断某一节点其是否存在未覆盖的物理链路,包括:
假设部署第k条路径f k时,已向路径中添加节点
Figure PCTCN2022091621-appb-000008
此时路径f k暂时表示为
Figure PCTCN2022091621-appb-000009
是第k条路径f k中第i个节点;
使用集合
Figure PCTCN2022091621-appb-000010
表示与节点
Figure PCTCN2022091621-appb-000011
相连节点的集合,则用公式(4)表示:
Figure PCTCN2022091621-appb-000012
从集合
Figure PCTCN2022091621-appb-000013
中选择下一个节点
Figure PCTCN2022091621-appb-000014
并且节点
Figure PCTCN2022091621-appb-000015
与节点
Figure PCTCN2022091621-appb-000016
之间 的链路不能被集合L'覆盖,则表示为:
Figure PCTCN2022091621-appb-000017
进一步的,在所述步骤S3中,通过如下的方法选择该起始节点的下一节点,包括:
在节点选择的过程中,路径
Figure PCTCN2022091621-appb-000018
下一节点
Figure PCTCN2022091621-appb-000019
选择策略具体步骤如下:
首先,通过当前节点
Figure PCTCN2022091621-appb-000020
获取集合
Figure PCTCN2022091621-appb-000021
并根据是否满足
Figure PCTCN2022091621-appb-000022
排除已经被覆盖的链路所连接的节点;
然后,选择集合
Figure PCTCN2022091621-appb-000023
剩余节点中需要向数据包中添加数据量最小的节点作为目标,选择策略描述为:
Figure PCTCN2022091621-appb-000024
b(i)表示节点i所需向数据包中插入的数据量。
进一步的,所述步骤S4具体包括:
首先,根据路径
Figure PCTCN2022091621-appb-000025
的当前信息计算得路径的当前时延T k,时延表达式表示为:
Figure PCTCN2022091621-appb-000026
在公式(5)中,t(i,j)表示节点i和节点j之间的时延信息;
然后,获取所选定节点
Figure PCTCN2022091621-appb-000027
与节点
Figure PCTCN2022091621-appb-000028
之间的时延
Figure PCTCN2022091621-appb-000029
时延检验公式如下所示:
Figure PCTCN2022091621-appb-000030
在公式(6)中,T max是系统能够接受的最大时延值;
如果满足时延检验条件,则继续添加节点
Figure PCTCN2022091621-appb-000031
否则,放弃添加并执行步骤S1。
进一步的,在所述步骤S5中,添加节点后所需更新的信息具体包括:路径f k经过节点的信息;路径f k当前的时延信息T k;遥感系统当前的覆盖链路集合L′。
进一步的,在所述步骤S6中,所述网络拓扑全覆盖,其判断条件为:E=L′,即遥感系统覆盖的链路集合L′与网络拓扑的物理链路集合E相同。
进一步的,在所述步骤S6中,所述终止条件,具体为:达到最大运行时间限制。
本发明的有益效果是:
本发明基于各路由设备遥测项目的信息,利用贪心选择算法的思想设计了探测包转发路径规划方案,降低了主动带内遥测系统的开销。同时基于各路径的时延信息,控制了转发路径长度以实现了对探测系统时延的控制。带内网络遥测系统可以获得稳定的全网可见性,有效降低遥测的开销并提升遥测系统的时延性能。
附图说明
图1为实施例1中提供的路由机制的结构示意图;
图2为实施例1中提供的一种面向主动网络遥感的低时延低开销路径部署方法的流程示意图。
具体实施方式
为使本发明实施例的目的、技术方案和优点更加清楚,下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本发明保护的范围。
实施例1
参见图1和图2,本实施例提供一种面向主动网络遥感的低时延低开销路径部署方法,该方法采用的路由机制,如图1所示,具体的说,在该机制中,使用UDP数据包来携带INT信息,分配固定大小的空间作为SR标签用来存储端口号以便控制数据包的转发路径,然后分配不定大小的空间作为INT标签用来存储网络遥感信息。
主动带内流量感知方案通过源路由(SR)技术解决了路径不可控的问题。也即是,在本实施例中路由机制需要INT生成器、INT转发器和INT收集器三种逻辑路由器来共同实现。INT生成器负责在探测流的起始节点处生成沿特定路由路径转发的探测数据包。INT转发器负责将本地流量信息添加入探测数据包中,并根据数据包传输路径转发数据包至下一端口。INT收集器一般是探测流路径上的最后一个节点,他负责在探测流完成流量信息收集工作后将数据包转发给控制平面进行分析处理。
具体的说,本实施例采用的该路由机制,其最大的优点在于可以生成覆盖整个网络拓扑的检测流,从而稳定地获取全网络的网络状态信息。这对于如今大规模网络的流量感知是非常重要的优势。
本实施例方法具体流程如图2所示,其包括如下的步骤:
步骤S1、根据网络拓扑状态初始化邻接矩阵;
具体的说,在本实施例中,该步骤S1包括:
将探测网络定义为一个无向物理图,用G=(V,E)表示。其中,V={i|i=1,…,n}为物理节点的集合,i∈V为监测网络中序号为i物理节点;E={e(i,j)|i,j∈V}为节点直接物理链路的集合,其中e(i,j)∈E为节点i和节点j之间的物理链路状态。如果节点i和节点j之间有连通的物理链路,则e(i,j)存在;否则e(i,j)不存在。存储图信息的邻接矩阵可以初始化为:
Figure PCTCN2022091621-appb-000032
步骤S2、新建一条探测流路径,并对网络拓扑各个节点的所需添加入探测数据包的数据量进行逐步遍历,并根据选择策略选择探测流的起始节点;
具体的说,在本实施例中,在路径规划的过程中,采用如下的选择策略来选择探测流的起始节点,包括:
假设部署第k条路径f k时,
Figure PCTCN2022091621-appb-000033
为路径f k中第一个节点的序号,则
Figure PCTCN2022091621-appb-000034
的选择策略可以表述为:
Figure PCTCN2022091621-appb-000035
在公式(2)中,b(i)表示节点i需要向探测数据包中添加的数据量;
需要注意的是,
Figure PCTCN2022091621-appb-000036
还必须拥有一条不在系统已覆盖链路集合L'的物理链路,其中,系统已覆盖链路集合L'可以表示为:
Figure PCTCN2022091621-appb-000037
在公式(3)中,L j是第j条路径覆盖的链路集合,具体表示为
Figure PCTCN2022091621-appb-000038
其中,N j表示第j条路径中的节点数量。
步骤S3、若当前节点的所有物理链路均已经被覆盖,则执行步骤S2;
具体的说,在本实施例中,当前节点
Figure PCTCN2022091621-appb-000039
需存在物理链路尚未被覆盖的条件可以具体描述为:
假设部署第k条路径f k时,已向路径中添加节点
Figure PCTCN2022091621-appb-000040
此时路径f k暂时可以表示为
Figure PCTCN2022091621-appb-000041
是第k条路径f k中第i个节点。为了方便描述,使用集合
Figure PCTCN2022091621-appb-000042
表示与节点
Figure PCTCN2022091621-appb-000043
相连节点的集合,可以用公式(4)表示:
Figure PCTCN2022091621-appb-000044
从集合
Figure PCTCN2022091621-appb-000045
中选择下一个节点
Figure PCTCN2022091621-appb-000046
并且节点
Figure PCTCN2022091621-appb-000047
与节点
Figure PCTCN2022091621-appb-000048
之间的链路不能被集合L'覆盖,则表示为:
Figure PCTCN2022091621-appb-000049
步骤S4、若当前节点的根据当前节点的未检测物理链路信息,依据选择策略,选择即将添加下一节点;
在节点选择的过程中,路径
Figure PCTCN2022091621-appb-000050
下一节点
Figure PCTCN2022091621-appb-000051
选择策略具体步骤如下:
首先,通过当前节点
Figure PCTCN2022091621-appb-000052
获取集合
Figure PCTCN2022091621-appb-000053
并根据是否满足
Figure PCTCN2022091621-appb-000054
排除已经被覆盖的链路所连接的节点。
然后,选择集合
Figure PCTCN2022091621-appb-000055
剩余节点中需要向数据包中添加数据量最小的节点作为目标,选择策略可以描述为:
Figure PCTCN2022091621-appb-000056
b(i)表示节点i所需向数据包中插入的数据量。
步骤S5、执行路径时延检测以判断是否继续添加节点,若不满足则执行步骤S2;
具体的说,为了保证路径时延不超过系统最大实验限制,需要进行路径时延检测。路径时延检测具体步骤如下:
首先,根据路径
Figure PCTCN2022091621-appb-000057
的当前信息计算得路径的当前时延T k,时延表达式可以表示为:
Figure PCTCN2022091621-appb-000058
在公式(5)中,t(i,j)表示节点i和节点j之间的时延信息。
然后,获取所选定节点
Figure PCTCN2022091621-appb-000059
与节点
Figure PCTCN2022091621-appb-000060
之间的时延
Figure PCTCN2022091621-appb-000061
时延检验公式如下所示:
Figure PCTCN2022091621-appb-000062
在公式(6)中,T max是系统可以接受的最大时延值。
如果满足时延检验条件,则可以继续添加节点
Figure PCTCN2022091621-appb-000063
否则,放弃添加并执行步骤S2。
步骤S6、添加节点,并更新当前节点信息和网络覆盖信息;
具体的说,在本实施例中,在该步骤中,添加节点后所需更新的信息具体包括:a、路径f k经过节点的信息;b、路径f k当前的时延信息T k;c、遥感系统当前的覆盖链路集合L′。
步骤S7、检测是否完成网络拓扑全覆盖,若不满足则执行步骤S3;
具体的说,在本实施例中,为了满足网络遥测系统全网络覆盖的要求需要进行全网络覆盖判断。全网络覆盖的具体判断条件可以描述为E=L′,即遥感系统覆盖的链路集合L′与网络拓扑的物理链路集合E相同。
步骤S8、如果满足终止条件,停止并完成所有路径规划任务。
具体的说,在本实施例中,完成路径规划的终止条件为:
1、网络遥感系统完成全网络覆盖。遥感系统覆盖的链路集合L′与网络拓扑的物理链路集合E相同,E=L′。
2、达到最大运行时间限制。针对网络规模过大或较为复杂无法完成全网络覆盖的情况,避免浪费不必要的资源。
综上所述,本发明提出一种面向主动网络遥感的低时延低开销路径部署方法,该方法可以以较高的效率生成主动带内遥感系统的探测流路径部署方案,利用此方法可以在保证遥测可见性的前提下有效地降低遥感系统的时延和开销。
本发明未详述之处,均为本领域技术人员的公知技术。
以上详细描述了本发明的较佳具体实施例。应当理解,本领域的普通技术人员无需创造性劳动就可以根据本发明的构思作出诸多修改和变化。因此,凡本技术领域中技术人员依本发明的构思在现有技术的基础上通过逻辑分析、推理或者有限的实验可以得到的技术方案,皆应在由权利要求书所确定的保护范围内。

Claims (9)

  1. 一种面向主动网络遥感的低时延低开销路径部署方法,其特征在于,所述方法包括如下步骤:
    步骤S1、针对一网络拓扑,构建一探测流路径,通过该探测流路径对该网络拓扑中各个节点,其需添加入探测数据包的数据量进行逐步遍历,其中,基于选择策略选择该探测流的起始节点;
    步骤S2、执行判断,判断步骤S1中选择的起始节点,其是否存在未覆盖的物理链路,若存在,则执行步骤S3,若不存在,则回到步骤S1继续选择起始节点;
    步骤S3、基于该起始节点其对应的未检测物理链路信息,选择该起始节点的下一节点;
    步骤S4、执行路径时延检测以判断是否继续添加节点,若满足时延检验条件,则继续添加节点,若不满足时延检验条件,则回到步骤S1中;
    步骤S5、添加节点,并更新当前节点信息和网络覆盖信息;
    步骤S6、执行检测,检测是否完成网络拓扑全覆盖,若没有完成,则回到步骤S2,若已完成,则进行判断,判断是否满足终止条件,若满足,则停止并完成所有路径规划任务。
  2. 根据权利要求1所述的一种面向主动网络遥感的低时延低开销路径部署方法,其特征在于,在构建探测流路径之前,先根据网络拓扑状态初始化邻接矩阵,其具体包括:
    将探测网络定义为一个无向物理图,用G=(V,E)表示;其中,V={i|i=1,…,n}为物理节点的集合,i∈V为监测网络中序号为i物理节点;E={e(i,j)|i,j∈V}为节点直接物理链路的集合,其中e(i,j)∈E为节点i和节点j之间的物理链路状态;
    如果节点i和节点j之间有连通的物理链路,则e(i,j)存在;否则e(i,j)不存在;存储图信息的邻接矩阵初始化为:
    Figure PCTCN2022091621-appb-100001
  3. 根据权利要求2所述的一种面向主动网络遥感的低时延低开销路径部署方法,其特征在于,在步骤S1中,在路径规划的过程中,采用如下的选择策略来选择探测流的起始节点, 包括:
    假设部署第k条路径f k时,
    Figure PCTCN2022091621-appb-100002
    为路径f k中第一个节点的序号,则
    Figure PCTCN2022091621-appb-100003
    的选择策略表述为:
    Figure PCTCN2022091621-appb-100004
    在公式(2)中,b(i)表示节点i需要向探测数据包中添加的数据量;
    其中,该
    Figure PCTCN2022091621-appb-100005
    还必须拥有一条不在系统已覆盖链路集合L'的物理链路,则系统已覆盖链路集合L'表示为:
    Figure PCTCN2022091621-appb-100006
    在公式(3)中,L j是第 j条路径覆盖的链路集合,具体表示为
    Figure PCTCN2022091621-appb-100007
    其中,N j表示第j条路径中的节点数量。
  4. 根据权利要求3所述的一种面向主动网络遥感的低时延低开销路径部署方法,其特征在于,在所述步骤S2中,通过如下的方法来判断某一节点其是否存在未覆盖的物理链路,包括:
    假设部署第k条路径f k时,已向路径中添加节点
    Figure PCTCN2022091621-appb-100008
    此时路径f k暂时表示为
    Figure PCTCN2022091621-appb-100009
    是第k条路径f k中第i个节点;
    使用集合
    Figure PCTCN2022091621-appb-100010
    表示与节点
    Figure PCTCN2022091621-appb-100011
    相连节点的集合,则用公式(4)表示:
    Figure PCTCN2022091621-appb-100012
    从集合
    Figure PCTCN2022091621-appb-100013
    中选择下一个节点
    Figure PCTCN2022091621-appb-100014
    并且节点
    Figure PCTCN2022091621-appb-100015
    与节点
    Figure PCTCN2022091621-appb-100016
    之间的链路不能被集合L'覆盖,则表示为:
    Figure PCTCN2022091621-appb-100017
  5. 根据权利要求4所述的一种面向主动网络遥感的低时延低开销路径部署方法,其特征在于,在所述步骤S3中,通过如下的方法选择该起始节点的下一节点,包括:
    在节点选择的过程中,路径
    Figure PCTCN2022091621-appb-100018
    下一节点
    Figure PCTCN2022091621-appb-100019
    选择策略具体步骤如 下:
    首先,通过当前节点
    Figure PCTCN2022091621-appb-100020
    获取集合
    Figure PCTCN2022091621-appb-100021
    并根据是否满足
    Figure PCTCN2022091621-appb-100022
    排除已经被覆盖的链路所连接的节点;
    然后,选择集合
    Figure PCTCN2022091621-appb-100023
    剩余节点中需要向数据包中添加数据量最小的节点作为目标,选择策略描述为:
    Figure PCTCN2022091621-appb-100024
    b(i)表示节点i所需向数据包中插入的数据量。
  6. 根据权利要求5所述的一种面向主动网络遥感的低时延低开销路径部署方法,其特征在于,所述步骤S4具体包括:
    首先,根据路径
    Figure PCTCN2022091621-appb-100025
    的当前信息计算得路径的当前时延T k,时延表达式表示为:
    Figure PCTCN2022091621-appb-100026
    在公式(5)中,t(i,j)表示节点i和节点j之间的时延信息;
    然后,获取所选定节点
    Figure PCTCN2022091621-appb-100027
    与节点
    Figure PCTCN2022091621-appb-100028
    之间的时延
    Figure PCTCN2022091621-appb-100029
    时延检验公式如下所示:
    Figure PCTCN2022091621-appb-100030
    在公式(6)中,T max是系统能够接受的最大时延值;
    如果满足时延检验条件,则继续添加节点
    Figure PCTCN2022091621-appb-100031
    否则,放弃添加并执行步骤S1。
  7. 根据权利要求6所述的一种面向主动网络遥感的低时延低开销路径部署方法,其特征在于,在所述步骤S5中,添加节点后所需更新的信息具体包括:路径f k经过节点的信息;路径f k当前的时延信息T k;遥感系统当前的覆盖链路集合L′。
  8. 根据权利要求7所述的一种面向主动网络遥感的低时延低开销路径部署方法,其特征在于,在所述步骤S6中,所述网络拓扑全覆盖,其判断条件为:E=L′,即遥感系统覆盖的链路集合L′与网络拓扑的物理链路集合E相同。
  9. 根据权利要求8所述的一种面向主动网络遥感的低时延低开销路径部署方法,其特征在于,在所述步骤S6中,所述终止条件,具体为:达到最大运行时间限制。
PCT/CN2022/091621 2022-04-28 2022-05-09 一种面向主动网络遥感的低时延低开销路径部署方法 WO2023206600A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210463410.4 2022-04-28
CN202210463410.4A CN114844812B (zh) 2022-04-28 2022-04-28 一种面向主动网络遥感的低时延低开销路径部署方法

Publications (1)

Publication Number Publication Date
WO2023206600A1 true WO2023206600A1 (zh) 2023-11-02

Family

ID=82567940

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/091621 WO2023206600A1 (zh) 2022-04-28 2022-05-09 一种面向主动网络遥感的低时延低开销路径部署方法

Country Status (2)

Country Link
CN (1) CN114844812B (zh)
WO (1) WO2023206600A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117835291A (zh) * 2024-03-04 2024-04-05 济南光路科技有限公司 一种基于物联网的数据管理系统及方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116032778B (zh) * 2023-03-29 2023-06-06 南京大学 云数据中心智能化监测应用任务部署方法、装置和系统
CN116232956B (zh) * 2023-05-06 2023-07-18 国网智能电网研究院有限公司 网络时延带内遥测方法、装置、电子设备及存储介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5838663A (en) * 1995-07-24 1998-11-17 Lucent Technologies Inc. Method for admission control and routing by allocating network resources in network nodes
CN109347741A (zh) * 2018-08-01 2019-02-15 北京邮电大学 基于带内网络遥测技术的全网路径最优化遍历方法及装置
CN113810225A (zh) * 2021-09-03 2021-12-17 中科南京信息高铁研究院 Sdn网络的带内网络遥测探测路径规划方法及系统

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200195553A1 (en) * 2018-12-17 2020-06-18 Netsia, Inc. System and method for measuring performance of virtual network functions
CN113347059B (zh) * 2021-05-24 2022-06-17 北京邮电大学 基于固定探针位置的带内网络遥测最优探测路径规划方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5838663A (en) * 1995-07-24 1998-11-17 Lucent Technologies Inc. Method for admission control and routing by allocating network resources in network nodes
CN109347741A (zh) * 2018-08-01 2019-02-15 北京邮电大学 基于带内网络遥测技术的全网路径最优化遍历方法及装置
CN113810225A (zh) * 2021-09-03 2021-12-17 中科南京信息高铁研究院 Sdn网络的带内网络遥测探测路径规划方法及系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
LIU ZHENGZHENG, JUN BI, YU ZHOU, YANGYANG WANG, YUNSENXIAO LIN: "Paradigm for proactive telemetry based on P4", JOURNAL ON COMMUNICATIONS, vol. 39, no. Z1, 30 September 2018 (2018-09-30), pages 162 - 169, XP055883641, ISSN: 1000-436x, DOI: 10.11959/j.issn.1000−436x.2018181 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117835291A (zh) * 2024-03-04 2024-04-05 济南光路科技有限公司 一种基于物联网的数据管理系统及方法
CN117835291B (zh) * 2024-03-04 2024-05-10 济南光路科技有限公司 一种基于物联网的数据管理系统及方法

Also Published As

Publication number Publication date
CN114844812B (zh) 2024-01-30
CN114844812A (zh) 2022-08-02

Similar Documents

Publication Publication Date Title
WO2023206600A1 (zh) 一种面向主动网络遥感的低时延低开销路径部署方法
CN108600102B (zh) 一种基于智慧协同网络的柔性数据传输系统
CN108199924B (zh) 基于带内网络遥测的全网流量可视化方法及装置
CN104320278B (zh) 基于软件定义网络sdn的广域网络实现方法及设备
CN105393511B (zh) 一种交换机模式切换方法、设备及系统
JPH1065733A (ja) 高速ルーティング制御方式
CN110275437B (zh) Sdn网络流量优势监控节点动态选择系统及其方法
CN106330749A (zh) 无回路多端部网络拓扑中的类别感知的负载平衡
CN105516025B (zh) 端到端的路径控制和数据传输方法、OpenFlow控制器和交换机
CN113347059B (zh) 基于固定探针位置的带内网络遥测最优探测路径规划方法
CN105553855B (zh) 动态调整底层网络生成树拓扑结构的方法及系统
CN116170327A (zh) 基于图神经网络和强化学习的分段路由网络增量部署方法
CN101471879A (zh) 一种层次化有序地址分组网络的路径控制系统和方法
Khoobbakht et al. Hybrid flow-rule placement method of proactive and reactive in SDNs
JPWO2015093478A1 (ja) ネットワークシステム、制御装置、制御方法およびプログラム
Ke et al. Routing strategy for SDN large flow based on deep reinforcement learning
Lin et al. Proactive multipath routing with a predictive mechanism in software‐defined networks
Rohitaksha et al. Routing in hybrid software defined networking using hop-count approach
Segara et al. Route discovery to avoid congestion in software defined networks
Wang et al. A Q-Learning based Routing Optimization Model in a Software Defined Network
Bano et al. A comparative analysis of hybrid routing schemes for SDN based wireless mesh networks
CN107483354A (zh) 基于sdn的网络阻塞解决方法及系统
Yi et al. Maximizing network utilization for SDN based on WiseAnt colony optimization
Zhu et al. MIFO: Multi-path Interdomain Forwarding
Hung et al. Scalable topology-based flow entry management in data center

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: 22939443

Country of ref document: EP

Kind code of ref document: A1