WO2015003414A1 - 一种基于多参数认知的快速网络拓扑推断方法 - Google Patents

一种基于多参数认知的快速网络拓扑推断方法 Download PDF

Info

Publication number
WO2015003414A1
WO2015003414A1 PCT/CN2013/080309 CN2013080309W WO2015003414A1 WO 2015003414 A1 WO2015003414 A1 WO 2015003414A1 CN 2013080309 W CN2013080309 W CN 2013080309W WO 2015003414 A1 WO2015003414 A1 WO 2015003414A1
Authority
WO
WIPO (PCT)
Prior art keywords
node
nodes
leaf
similarity
topology
Prior art date
Application number
PCT/CN2013/080309
Other languages
English (en)
French (fr)
Inventor
徐友云
李来扛
李宁
谢威
张冬梅
Original Assignee
中国人民解放军理工大学
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 中国人民解放军理工大学 filed Critical 中国人民解放军理工大学
Publication of WO2015003414A1 publication Critical patent/WO2015003414A1/zh

Links

Classifications

    • 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
    • H04L43/00Arrangements for monitoring or testing data switching networks
    • H04L43/10Active monitoring, e.g. heartbeat, ping or trace-route
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00Arrangements for monitoring or testing data switching networks
    • H04L43/08Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
    • H04L43/0823Errors, e.g. transmission errors
    • H04L43/0829Packet loss
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00Arrangements for monitoring or testing data switching networks
    • H04L43/08Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
    • H04L43/0852Delays
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00Arrangements for monitoring or testing data switching networks
    • H04L43/10Active monitoring, e.g. heartbeat, ping or trace-route
    • H04L43/103Active monitoring, e.g. heartbeat, ping or trace-route with adaptive polling, i.e. dynamically adapting the polling rate
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00Arrangements for monitoring or testing data switching networks
    • H04L43/16Threshold monitoring

Definitions

  • the invention belongs to the technical field of wireless networks, and particularly relates to a simple and fast method for inferring a topology of a wireless network. Background technique
  • the heterogeneous wireless access network is the current state of the wireless environment.
  • the inference of wireless network topology can be different for intelligent terminals.
  • the choice of the wireless access network provides the basis.
  • Network topology is one of the important parameters of wireless networks. It is also the basis for measuring the internal performance parameters of the network. It is also the basic component of network deployment and network management functions.
  • the network topology in wireless networks is dynamically changing, so it is especially important to design a fast and efficient network topology inference method.
  • the topology information derived from the network topology is the basis of wireless network management and network optimization. Through the network topology information, you can establish corresponding routing information, adjust network congestion, help users make decisions, search network bottlenecks, and develop new network protocols and methods.
  • the method of the invention is based on the perception of multiple parameters, and the feedback adjustment in the cognitive module can effectively improve the accuracy of topology inference.
  • the invention can quickly and effectively infer the network logical topology, is simple to implement, has low complexity, and does not require internal node cooperation or deployment inside the network.
  • the object of the present invention is to provide a method for quickly inferring a wireless network topology based on multi-parameter cognitive technology, inferring the connection condition according to the relationship of nodes in the wireless network, and having feedback adjustment, simple implementation, low complexity, and convenient network maintenance.
  • a fast network topology inference method based on multi-parameter cognition includes the following steps: First, detecting packet delay and detection packet loss perception, performing delay difference analysis on the detection packet delay parameter, and calculating the leaf by calculating The similarity between the nodes is used to determine the relationship between the leaf nodes and the leaf node topology is estimated. The leaf node topology is compared with the original topology to determine whether the similarity meets the appropriate threshold requirements. If the requirements are not met, the parameters of the detection packet are feedback adjusted.
  • Completing the cognitive loop if the requirements are met, analyzing the packet loss parameters of the parent node, calculating the Hamming distance between the parent nodes as their similarity, inferring the node relationship according to the similarity, and inferring the network topology.
  • the method of the present invention specifically includes the following steps:
  • the root node continuously multicasts the probe packets to all the leaf nodes; each leaf node records the loss of all the probe packets and calculates the delay of each probe packet;
  • the similarity of the leaf nodes determine whether the relationship of the leaf nodes is a sibling or an isolated node, and then obtain the leaf node topology, and determine the similarity between the leaf node topology and the original topology, that is, determine the father and son, brothers, Whether the relationship between the isolated relationship and the construction in step a is consistent, the similarity of the network topology is equal to the ratio of the number of nodes correctly identified in the node set to the total number of nodes in the node set; if the similarity meets the threshold requirement, proceed to step e;
  • feedback adjusts the parameters of the probe packet, reduces the time interval for sending the probe packet, and returns to step b ;
  • Non-brother nodes Infer the packet loss of the parent node according to the loss of the child node message, and then calculate the Hamming distance between the parent node and the isolated node. If the Hamming of the node satisfies the given relationship, it is regarded as a sibling node, otherwise it is regarded as a sibling node.
  • Non-brother nodes Inferring the relationship between nodes, and then obtaining the network logical topology.
  • the root node multicasts a probe packet to all the leaf nodes, and the probe packets have a certain time interval.
  • step c of the present invention Calculating the similarity of each leaf node in step c of the present invention: For any two leaf nodes, /, first calculate the average value of m delay data, denoted as m , d m , select an appropriate threshold, and then The corresponding m delay data of the leaf nodes and / are compared, if
  • 1, 2, ...,
  • the similarity of the leaf nodes calculated according to the step c is selected, and the appropriate threshold value f is selected, which satisfies the million " ⁇ s is regarded as a sibling node, and if not satisfied, there is no sibling node in the leaf node, which is called
  • the leaf node is divided into two types: the sibling node and the isolated node, and then the leaf node topology is estimated, and the leaf node topology is compared with the original topology. If the network topology similarity meets the appropriate threshold requirement, then Step e is performed. Otherwise, feedback adjusts the parameters of the probe packet, reduces the interval for sending the probe packet, and returns to step b .
  • the threshold value f and the threshold are set according to actual precision requirements.
  • the similarity threshold of the network topology of the present invention ranges from 70 to 99%; the threshold value ranges from d -d
  • the detection of the loss of the detection packet of the parent node is inferred according to the loss of the detection information of the child node, and the loss of the detection message of the parent node is the union of the loss of the message of all the child nodes, and the The message loss is represented by ⁇ , whose child nodes are nodes 1, 2, ... ⁇ , then
  • AU 2 U... U ⁇ , Calculate the Hamming distance between the parent node and the isolated node according to the loss of the parent node's probe message, select the two nodes with the smallest Hamming distance, / as the sibling node, and then determine whether other nodes are connected to the node, / is a sibling node, if it meets H, > is considered a brotherhood
  • the invention is based on the sensing of the two parameters of the detection packet delay and the packet loss, and the feedback adjustment of the cognitive module can effectively improve the accuracy of the topology inference;
  • the network tomography based on the present invention estimates the performance parameters of the network by collecting data at the network boundary node; the measurement process does not need to deploy the measurement device inside the network, and does not require the internal nodes of the network to participate in cooperation;
  • the invention combines two parameters of detecting packet delay and packet loss, and the clustering analysis of the delay parameter can effectively alleviate the influence of the difference in performance of the detection packet due to the uncertainty factor, and the Hamming distance of the packet loss parameter.
  • the analysis is simple to implement, easy to analyze, and low in complexity. Combining two parameters can improve the accuracy of the inference results;
  • the algorithm Compared with the Global Positioning System (GPS) with high precision but relatively high cost, the algorithm has the advantages of simple and easy implementation, low algorithm complexity, fast topology inference, and low cost.
  • GPS Global Positioning System
  • Figure 1 is a block diagram of network topology inference based on multi-parameter cognition
  • Figure 2 is a tree topology network model diagram
  • Figure 3 is a tree topology diagram in the embodiment
  • Figure 4 is a graphical representation of the results of clustering based on the degree of dissimilarity between the post-leaf nodes in step d cluster analysis.
  • Figure 5 is a schematic diagram showing the result of node relationship inference e based on the Hamming distance between nodes.
  • FIG. 6 is a schematic diagram showing the result of the step m inferring the node relationship based on the Hamming distance between the nodes.
  • the present invention will be further described below with reference to the accompanying drawings and embodiments. (Please refer to the formula and statement that are inconsistent with the previous ones)
  • the invention is based on multi-parameter cognition, including detecting packet delay and packet loss parameters, and calculating the similarity between leaf nodes according to the difference analysis of the probe packet delay parameters, and determining the relationship of the leaf nodes, and the leaf node topology and the original Topology comparison and judging whether the similarity meets the appropriate threshold requirement. If the requirements are not met, the parameters of the detection packet are feedback adjusted to complete the cognitive loop.
  • the specific operation is shown in the virtual box of Figure 1, to further infer the node according to the packet loss parameter.
  • the relationship is further based on topology inference; the feedback adjustment of the cognitive module can effectively improve the accuracy of topology inference.
  • the topology inference algorithm of the present invention is based on a network tomography technique, and the network topology in the technology is usually described by a logic tree. The more shared links between nodes, the closer the packet delay data and the packet loss situation are. Obtaining the measured values of the leaf nodes can group the nodes with message similarity, and infer the network logical topology.
  • This embodiment adopts a tree topology, as shown in FIG. 3,
  • a fast network topology inference algorithm based on multi-parameter cognition characterized in that the steps include:
  • the root node 0 multicasts the probe packet to the leaf node ⁇ 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 ⁇ , there is a certain time interval between the probe messages;
  • step C According to the similarity of the leaf nodes calculated in step C, and selecting the appropriate threshold value f, satisfying 10,000 " ⁇ s as a sibling node, otherwise regarded as an isolated node, and obtaining a node ⁇ 1, 2, 6, 11, 15 , 18, 19, 20 ⁇ are isolated nodes, ⁇ 3, 4, 5 ⁇ , ⁇ 7, 8 ⁇ , ⁇ 9, 10 ⁇ , ⁇ 12, 13, 14 ⁇ , ⁇ 16, 1 7 ⁇ are brother nodes; Update the leaf node set R to ⁇ 1, 2, 6, 11, 15, 18, 19, 20, 21 (the parent of 3, 4, 5), 22 (the parent of 7, 8), 23 (9, 10) Parent node), 24 (parent nodes of 12, 13, 14), 25 (parent nodes of 16, 17) ⁇ ; If the obtained leaf node topology is less similar to the original topology, adjust the probe packet parameters, minus Repeat the above process for the interval of sending small probe packets until the leaf topology and the original topology similarity meet the threshold requirements;
  • « min is the minimum packet loss rate of the link. After all the siblings are identified, all the siblings are removed from the node set and added to their parent nodes. Calculate and judge that ⁇ 2, 6, 21 ⁇ is a sibling node; update the leaf node set R to ⁇ 1, 11, 15, 18, 19, 20, 22, 23, 24, 25, 26 (2, 6, 21 Parent node) ⁇ ;
  • step 6 calculate the Hamming distance of the two nodes in the updated leaf node set R in step e, and judge that ⁇ 15, 24 ⁇ is a sibling node; and update the leaf node set R to ⁇ 1, 11, 18, 19, 20, 22, 23, 25, 26, 27 (father of 15, 24) ⁇ ;
  • step 6 Repeat step 6 to calculate the Hamming of the two nodes in the updated leaf node set R in step f. Distance, and judge that ⁇ 11, 23 ⁇ is a sibling node; update leaf node set R is ⁇ 1, 18, 19, 20, 22, 25, 26, 27, 28 (the parent node of 11, 23) ⁇ ;
  • step ⁇ to calculate the Hamming distance of the two nodes in the updated leaf node set R in step g, and judge that ⁇ 18, 25 ⁇ is the sibling node; and update the leaf node set R to be ⁇ 1, 19, 20, 22 , 26, 27, 28, 29 (the parent node of 18, 25) ⁇ ;
  • step 6 calculate the Hamming distance of the two nodes in the updated leaf node set R in step h, and judge that ⁇ 19, 27, 29 ⁇ is a sibling node; and update the leaf node set R to ⁇ 1, 20, 22, 26, 28, 30 (the parent of 19, 27, 29) ⁇ ;
  • step 6 calculate the Hamming distance of the two nodes in the updated leaf node set R in step i, and judge that ⁇ 1, 26 ⁇ is a sibling node; and update the leaf node set R to ⁇ 20, 22, 28, 30, 31 (the parent node of 1, 26) ⁇ ;
  • step ⁇ calculating the Hamming distance of the two nodes in the updated leaf node set R in step j, and judging that ⁇ 22, 28 ⁇ is a sibling node; updating the leaf node set R is ⁇ 20, 30, 31, 32 (the parent node of 22, 28)
  • step 6 calculate the Hamming distance of the two nodes in the updated leaf node set R in step k, and judge that ⁇ 20, 30 ⁇ is the sibling node; and update the leaf node set R to be ⁇ 31, 32, 33 ( 20, 30 parent node) ⁇ ;
  • step ⁇ to calculate the Hamming distance of the two nodes in the updated leaf node set R, and judge that ⁇ 31, 32, 33 ⁇ is the sibling node; and update the leaf node set R to be ⁇ 34 ( 31, 32 , The parent node of 33) ⁇ ;

Abstract

一种基于多参数认知的快速网络拓扑推断方法,包括以下步骤:首先对探测报文时延及探测报文丢失感知,对探测报文时延参数进行差值分析,通过计算叶节点间的相似度来判断叶节点的关系并推测得到叶节点拓扑结构,将叶节点拓扑与原拓扑对比判断其相似度是否满足适当的阈值要求,若不满足要求则反馈调整探测报文参数,完成认知循环;若满足要求则分析父节点丢包参数,计算父节点间的Hamming距离作为其相似度,根据相似度推断节点关系,进而推断网络拓扑结构。本发明基于对探测报文时延及丢包两种参数的感知,认知模块的反馈调整能够有效提高拓扑推断的精度;基于多参数感知的网络拓扑推断能够应对网络不稳定带来的数据偏差,推断结果更准确。

Description

说 明 书
一种基于多参数认知的快速网络拓扑推断方法 技术领域
本发明属于无线网络技术领域, 特别涉及一种简单、 快速的无线网络拓扑 结构的推断方法。 背景技术
目前, 随着无线技术的高速发展, 无线接入网络的多样性越来越引起人们 的重视, 无线接入网络异构是目前无线环境的现状, 对无线网络拓扑的推断能 够为智能终端对异构无线接入网络的选择提供依据。
网络拓扑是无线网络的重要参数之一, 也是测量网络内部性能参数的基础 , 更是网络部署和网络管理功能的基本构成部分。 无线网络中网络拓扑是动态 变化的, 所以设计一种快速、 有效的网络拓扑推断方法显得尤为重要。 网络拓 扑推断得到的拓扑信息是无线网络管理和网络优化的基础。 通过网络拓扑信息 , 可以建立相应的路由信息, 可以调整网络拥塞, 帮助用户决策, 搜索网络瓶 颈, 开发新型网络协议及方法等。 本发明方法基于对多参数的感知, 认知模块 中的反馈调整能够有效提高拓扑推断的精度。 本发明能够快速有效的推断网络 逻辑拓扑, 实施简单, 复杂度较低, 且不需要内部节点协作或在网络内部部署
本发明的目的在于基于多参数的认知技术提供一种快速推断无线网络拓扑 的方法, 根据无线网络中节点的关系推断其连接情况, 且具有反馈调整、 实施 简单、 复杂度低、 便于网络维护等特点。
本发明的技术方案是: 一种基于多参数认知的快速网络拓扑推断方法, 它包括以下步骤: 首先对 探测报文时延及探测报文丢失感知, 对探测报文时延参数进行时延差值分析, 通过计算叶节点间的相似度来判断叶节点的关系并推测得到叶节点拓扑结构, 将叶节点拓扑与原拓扑对比判断其相似度是否满足适当的阈值要求, 若不满足 要求则反馈调整探测报文参数, 完成认知循环; 若满足要求则分析父节点丢包 参数, 计算父节点间的 Hamming距离作为其相似度, 根据相似度推断节点关系 , 进而推断网络拓扑结构。
本发明的方法具体包括以下步骤:
a、 在无线网络中构建树状网络拓扑结构, 其中包含根节点、 中间节点和叶 节点, 记录各节点之间的父子、 兄弟、 孤立关系; 兄弟节点在其上层拥有同一 个父节点, 而此兄弟节点称作是父节点的子节点, 孤立节点在本层无兄弟关系 节点;
b、 由根节点连续向所有叶节点组播探测报文; 每个叶节点记录所有探测报 文丢失情况并计算每个探测报文的时延;
c、 对各叶节点探测报文的时延数据进行差值分析, 并计算各叶节点的相似 度;
d、 根据叶节点的相似度, 判断叶节点的关系是兄弟节点或孤立节点, 进而 得到叶节点拓扑结构, 并判断叶节点拓扑与原拓扑的相似度即判断各节点之间 的父子、 兄弟、 孤立关系与步骤 a中的构建的关系是否吻合, 网络拓扑的相似度 等于节点集中被正确识别出的节点数量与节点集中总的节点数量的比值; 如果相似度达到阈值要求, 则进行步骤 e;
如果相似度达不到阈值要求, 则反馈调整探测报文参数, 减小发送探测报 文的时间间隔, 返回步骤 b;
e、 根据子节点报文丢失情况推测父节点的报文丢失情况, 进而计算父节点 及孤立节点间的 Hamming距离, 若节点间的 Hamming满足给定的关系式则视为 兄弟节点, 否则视为非兄弟节点, 以此推断节点间的关系, 进而得到网络逻辑 拓扑结构。
本发明的步骤 a中无线树状网络拓扑结构中, 根节点组播 个探测报文给所 有叶节点, 探测报文间有一定的时间间隔。 本发明的步骤 b中每个叶节点都用一个随机过程 = e m)来表示探测 报文丢失情况, 其中 表示第 个探测报文在节点 处的丢失情况, 如果节点 接收到探测报文, 则 =1, 否则 x =0; 用 表示节点 处第 个探 测报文的时延。
本发明的步骤 c中计算各叶节点的相似度: 针对任意两个叶节点 、 /, 首先 计算 m个时延数据的平均值, 记为 ,m、 d m, 选择合适门限值 , 将两个叶节 点 、 /的相应的 m个时延数据 与 进行比较, 若满足| - |< =1,2,...,
, 将位置 的值置为 1, 否则为 0, 写入对应的时延差值类数据库中, 并统计其中 1的个数, 记为 ¾, 采用公式 1计算任意两个叶节点 、 /的相似度; 前述门限 值 根据实验数据设置。
本发明的步骤 d中, 根据步骤 c计算得到的叶节点相似度, 并选择合适门限 值 f, 满足万《<s的视为兄弟节点, 若不满足则叶节点中无兄弟节点, 称之为 孤立节点, 通过判断将叶节点分为兄弟节点和孤立节点两类, 进而推测得到叶 节点拓扑结构, 并将叶节点拓扑与原拓扑对比, 若网络拓扑的相似度满足适当 的阈值要求, 则执行步骤 e, 否则反馈调整探测报文参数, 减小发送探测报文的 时间间隔, 返回步骤 b; 前述门限值 f和阈值根据实际精度需求进行设置。
本发明的网络拓扑的相似度阈值的范围是 70-99%; 门限值 的范围是 d -d
<5<^^{i = \ ...,m)
2 2
本发明的步骤 e中, 根据子节点探测报文丢失情况推断其父节点的探测报文 丢失情况, 父节点的探测报文丢失情况为所有子节点报文丢失情况的并集, 任 一节点 的报文丢失情况用 ^来表示, 它的子节点为节点 1,2,...^, 则
^ = A U 2 U… U ^, 根据父节点探测报文丢失情况计算父节点及孤立节点 间的 Hamming距离, 选择 Hamming距离最小的两个节点 、 /作为兄弟节点, 再 判断其他节点 是否与节点 、 /是兄弟节点, 若满足 H , > 视为兄弟节
Hd(k ) 点, 否则不是兄弟节点, 这里 = l-«min, 《min链路的最小丢包率, 在识别出所 有兄弟节点后, 从节点集合中去掉所有的兄弟节点, 加入其父节点。 重复步骤 e 直至叶节点集合内剩下一个节点为止。 本发明的有益效果:
本发明基于对探测报文时延及丢包两种参数的感知, 认知模块的反馈调整 能够有效提高拓扑推断的精度;
本发明基于网络断层扫描技术是通过在网络边界节点处进行测量收集数据 , 从而推测网络内部的性能参数; 测量过程不需要在网络内部部署测量设备, 不需要网络内部节点协同参与;
本发明联合了探测报文时延和丢包两个参数, 对时延参数的聚类分析能够 有效缓解因不确定因素导致探测报文性能差异较大的影响, 对丢包参数的 Ham ming距离分析实施简单, 易于分析, 复杂度较低, 联合两个参数能够提高推断 结果的准确度;
本发明与精度很高但是成本也相对较高的全球定位系统 (GPS ) 相比, 该算 法的优点是简单易于实施, 算法复杂度较低, 能够实现快速拓扑推断, 且成本 较低。
,删
图 1为基于多参数认知的网络拓扑推断框图
图 2为树状拓扑结构网络模型图
图 3为本实施例中的树状拓扑结构图
图 4为步骤 d以时延聚类分析后叶节点间的相异度为依据进行聚类的结果示 意图
图 5为步骤 e以节点间的 Hamming距离为依据进行节点关系推断的结果示意 图
图 6为步骤 m以节点间的 Hamming距离为依据进行节点关系推断的结果示意 图 具体实 式 下面结合附图和实施例对本发明作进一步的说明。 (涉及公式和说法与之 前不一致的请修改) 本发明基于多参数认知, 包括探测报文时延及丢包参数, 根据探测报文时 延参数差值分析, 计算叶节点间的相似度来判断叶节点的关系, 将叶节点拓扑 与原拓扑对比并判断其相似度是否满足适当的阈值要求, 若不满足要求则反馈 调整探测报文参数, 完成认知循环, 具体操作如图 1虚框内所示, 为进一步根据 丢包参数推断节点关系进而进行拓扑推断提供依据; 认知模块的反馈调整能够 有效提高拓扑推断的精度。 本发明的拓扑推断算法基于网络断层扫描技术, 而 该技术中的网络拓扑通常用逻辑树来描述, 节点之间的共享链路越多, 其报文 时延数据及丢包情况越相近, 通过获得叶节点的测量值可以实现对具有报文相 似性的节点进行分组, 进而推断出网络逻辑拓扑。
本实施例采用树状拓扑结构, 如图 3所示,
一种基于多参数认知的快速网络拓扑推断算法, 其特征在于步骤包括:
a、 网络中根节点 0组播 个探测报文给叶节点 {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20}, 探测报文间存在一定的 时间间隔;
b、 经过 个连续时间的捕获和分析, 每个叶节点都用一个随机过程 t ^' e )来表示探测报文丢失情况, 其中 ;)表示第 个探测报文在节点 处的丢失情况, 如果节点 接收到探测报文则 ;)=1, 否则 ''〕=0; 用 d[ (i e m)表示节点 k处第 i个探测报文的时延, 如表 1所示。
表 1
Figure imgf000007_0001
c、 针对任意两个叶节点 、 /, 首先计算 m个时延数据的平均值, 记为 ,m 、 dm, 选择合适门限值 , 将两个叶节点 、 /的相应的 m个时延数据 与 进行比较, 若满足| - |< ( = 1,2,... ), 将位置 的值置为 1, 否则为 0, 写入 对应的时延差值类数据库中, 如表 2所示 (行为类名, 列为探测报文的序号, 以 图 1为例) , 并统计每一类中 1的个数, 记为 ¾, 采用公式 1计算任意两个叶节 点 、 /的相似度。 P 1 2 3 m-1 m c
1 0 1 1 0
C45 1 0 0 1 0
d、 根据步骤 C中计算得到的叶节点相似度, 并选择合适门限值 f, 满足 万《<s的作为兄弟节点, 否则视为孤立节点, 得到节点 {1, 2, 6, 11, 15, 18 , 19, 20}为孤立节点, {3, 4, 5}、 {7, 8}、 {9, 10}、 {12, 13, 14}、 {16, 1 7}分别为兄弟节点; 更新叶节点集合 R为 {1, 2, 6, 11, 15, 18, 19, 20, 21 (3、 4、 5的父节点) , 22 (7、 8的父节点) , 23 (9、 10的父节点) , 24 (12 、 13、 14的父节点) , 25 (16、 17的父节点) }; 若得到的叶节点拓扑与原拓扑 相似度较小, 则调整探测报文参数, 减小探测报文的发送时间间隔, 重复上述 过程, 直到叶节点拓扑与原拓扑相似度满足阈值要求;
e、 根据子节点探测报文丢失情况推测父节点探测报文丢失情况, 计算步骤 d中更新后的叶节点集合 R中两两节点的 Hamming距离, 选择 Hamming距离最小 的两个节点 、 /作为兄弟节点, 再判断其他节点 '是否与节点 、 /是兄弟节 点, 若满足 ΗΛ Ι > 视为兄弟节点, 否则不是兄弟节点, 这里 = l-«min
Hd(k,k')
«min是链路的最小丢包率, 在识别出所有兄弟节点后, 从节点集合中去掉所有 的兄弟节点, 加入其父节点。 计算并判断得到 {2, 6, 21}为兄弟节点; 更新叶 节点集合 R为 {1, 11, 15, 18, 19, 20, 22, 23, 24, 25, 26 (2、 6、 21的父 节点) };
f、 重复步骤6, 计算步骤 e中更新后的叶节点集合 R中两两节点的 Hamming 距离, 并判断得到 {15, 24}为兄弟节点; 更新叶节点集合 R为 {1, 11, 18, 19, 20, 22, 23, 25, 26, 27 (15, 24的父节点) };
g、 重复步骤6, 计算步骤 f中更新后的叶节点集合 R中两两节点的 Hamming 距离, 并判断得到 {11, 23}为兄弟节点; 更新叶节点集合 R为 {1, 18, 19, 20, 22, 25 , 26, 27, 28 ( 11, 23的父节点) };
h、 重复步骤^ 计算步骤 g中更新后的叶节点集合 R中两两节点的 Hamming 距离, 并判断得到 {18, 25}为兄弟节点; 更新叶节点集合 R为 {1, 19, 20, 22, 26, 27, 28, 29 ( 18, 25的父节点) } ;
i、 重复步骤6, 计算步骤 h中更新后的叶节点集合 R中两两节点的 Hamming 距离, 并判断得到 {19, 27, 29}为兄弟节点; 更新叶节点集合 R为 {1, 20, 22, 26, 28, 30 ( 19, 27, 29的父节点) } ;
j、 重复步骤6, 计算步骤 i中更新后的叶节点集合 R中两两节点的 Hamming距 离, 并判断得到 {1, 26}为兄弟节点; 更新叶节点集合 R为 {20, 22, 28, 30, 31 ( 1, 26的父节点) } ;
k、 重复步骤^ 计算步骤 j中更新后的叶节点集合 R中两两节点的 Hamming 距离, 并判断得到 {22, 28}为兄弟节点; 更新叶节点集合 R为 {20, 30, 31, 32 (22, 28的父节点)
1、 重复步骤6, 计算步骤 k中更新后的叶节点集合 R中两两节点的 Hamming 距离, 并判断得到 {20, 30}为兄弟节点; 更新叶节点集合 R为 {31, 32, 33 (20 , 30的父节点) } ;
m、 重复步骤^ 计算步骤冲更新后的叶节点集合 R中两两节点的 Hamming 距离, 并判断得到 {31, 32, 33}为兄弟节点; 更新叶节点集合 R为 {34 ( 31, 32 , 33的父节点) } ;
本发明未涉及部分均与现有技术相同或可采用现有技术加以实现。

Claims

^ ^ ¾ ¾
1、 一种基于多参数认知的快速网络拓扑推断方法, 其特征在于它包括以下步骤
: 首先对探测报文时延及探测报文丢失感知, 对探测报文时延参数进行差值分 析, 通过计算叶节点间的相似度来判断叶节点的关系并推测得到叶节点拓扑结 构, 将叶节点拓扑与原拓扑对比判断其相似度是否满足适当的阈值要求, 若不 满足要求则反馈调整探测报文参数, 完成认知循环; 若满足要求则分析父节点 丢包参数, 计算父节点间的 Hamming距离作为其相似度, 根据相似度推断节点 关系, 进而推断网络拓扑结构。
2、 如权利要求 1所述的基于多参数认知的快速网络拓扑推断方法, 其特征在于 该方法具体包括以下步骤:
a、 在无线网络中构建树状网络拓扑结构, 其中包含根节点、 中间节点和叶 节点, 记录各节点之间的父子、 兄弟、 孤立关系; 兄弟节点在其上层拥有同一 个父节点, 而此兄弟节点称作是父节点的子节点, 孤立节点在本层无兄弟关系 节点;
b、 由根节点连续向所有叶节点组播探测报文; 每个叶节点记录所有探测报 文丢失情况并计算每个探测报文的时延;
c、 对各叶节点探测报文的时延数据进行差值分析, 并计算各叶节点的相似 度;
d、 根据叶节点的相似度, 判断叶节点的关系是兄弟节点或孤立节点, 进而 得到叶节点拓扑结构, 并判断叶节点拓扑与原拓扑的相似度即判断各节点之间 的父子、 兄弟、 孤立关系与步骤 a中的构建的关系是否吻合, 网络拓扑的相似度 等于节点集中被正确识别出的节点数量与节点集中总的节点数量的比值;
如果相似度达到阈值要求, 则进行步骤 e;
如果相似度达不到阈值要求, 则反馈调整探测报文参数, 减小发送探测报 文的时间间隔, 返回步骤 b;
e、 根据子节点报文丢失情况推测父节点的报文丢失情况, 进而计算父节点 及孤立节点间的 Hamming距离, 若节点间的 Hamming距离满足给定的关系式则 视为兄弟节点, 否则视为非兄弟节点, 以此推断节点间的关系, 进而得到网络 逻辑拓扑结构。
3、 如权利要求 2所述基于多参数认知的快速网络拓扑推断方法, 其特征在于, 所述步骤 a中无线树状网络拓扑结构中, 根节点组播 w个探测报文给所有叶节点 , 探测报文间有一定的时间间隔。
4、 如权利要求 2所述基于多参数认知的快速网络拓扑推断方法, 其特征在于, 所述步骤 b中每个叶节点都用一个随机过程 = e m)来表示探测报文丢失 情况, 其中 表示第 个探测报文在节点 处的丢失情况, 如果节点 接收到 探测报文, 则 = 1, 否则 x = 0 ; 用 表示节点 处第 个探测报文的 时延。
5、 如权利要求 2所述基于多参数认知的快速网络拓扑推断方法, 其特征在于, 所述步骤 c中具体包括:
计算各叶节点的相似度: 针对任意两个叶节点 、 /, 首先计算 m个时延数 据的平均值, 记为 ,m、 dKm , 选择合适门限值 , 将两个叶节点 、 /的相应 的 m个时延数据 与 进行比较, 若满足 | - | < ( = 1, 2, ... ), 将位置 的 值置为 1, 否则为 0, 写入对应的时延差值类数据库中, 并统计其中 1的个数, 记 为¾, 采用公式 1计算任意两个叶节点 、 /的相似度; 前述门限值 根据实验 数据设置;
DkI = ^L(k≠l) ( 1 )
m
6、 如权利要求 2所述基于多参数认知的快速网络拓扑推断方法, 其特征在于, 所述步骤 d中, 根据步骤 c计算得到的叶节点相似度, 并选择合适门限值 f, 满 足万《< s的视为兄弟节点, 若不满足则叶节点中无兄弟节点, 称之为孤立节点 , 通过判断将叶节点分为兄弟节点和孤立节点两类, 进而推测得到叶节点拓扑 结构, 并将叶节点拓扑与原拓扑对比, 若网络拓扑的相似度满足适当的阈值要 求, 则执行步骤 e, 否则反馈调整探测报文参数, 减小发送探测报文的时间间隔 , 返回步骤 b; 前述门限值 f和阈值根据实际精度需求进行设置。
7、 如权利要求 6所述基于多参数认知的快速网络拓扑推断方法, 其特征在于, 网络拓扑的相似度阈值的范围是 70-99%; 门限值 的范围是 (i = \,2,...,m)。
Figure imgf000012_0001
8、 如权利要求 2所述基于多参数认知的快速网络拓扑推断方法, 其特征在于, 所述步骤 e中, 根据子节点探测报文丢失情况推断其父节点的探测报文丢失情况 , 父节点的探测报文丢失情况为所有子节点报文丢失情况的并集, 任一节点 的报文丢失情况用 来表示, 它的子节点为节点 1,2,...^, 则 Xk (m) = Xm) υ^Μ)υ···υ Xm)。 节点 与节点 v之间的 Hamming距离表示为
ΗΑ^) =∑ η>®^ν η>' ㊉为异或运算, w是探测报文的数量。 根据父节点探 测报文丢失情况计算父节点及孤立节点间的 Hamming距离, 选择 Hamming距离 最小的两个节点 、 /作为兄弟节点, 再判断其他节点 ^是否与节点 、 /是兄 弟节点, 若满足 ΗΛ Γ > 视为兄弟节点, 否则不是兄弟节点, 这里 ξ-Ι-α^, «min链路的最小丢包率, 在识别出所有兄弟节点后, 从节点集合中 去掉所有的兄弟节点, 加入其父节点。 重复步骤 e直至叶节点集合内剩下一个节 点为止。
PCT/CN2013/080309 2013-07-12 2013-07-29 一种基于多参数认知的快速网络拓扑推断方法 WO2015003414A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201310293511.2 2013-07-12
CN201310293511.2A CN103326904B (zh) 2013-07-12 2013-07-12 一种基于多参数认知的快速网络拓扑推断方法

Publications (1)

Publication Number Publication Date
WO2015003414A1 true WO2015003414A1 (zh) 2015-01-15

Family

ID=49195447

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2013/080309 WO2015003414A1 (zh) 2013-07-12 2013-07-29 一种基于多参数认知的快速网络拓扑推断方法

Country Status (2)

Country Link
CN (1) CN103326904B (zh)
WO (1) WO2015003414A1 (zh)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111626459A (zh) * 2019-02-27 2020-09-04 武汉国测数据技术有限公司 一种区域充电网络拓扑关系确认方法、装置和系统
CN112986658A (zh) * 2021-01-25 2021-06-18 浙江华云信息科技有限公司 一种低压线路的拓扑识别方法及装置
CN113078630A (zh) * 2020-11-20 2021-07-06 中国科学院沈阳自动化研究所 一种基于实时量测数据的低压配电网拓扑辨识方法
CN113630269A (zh) * 2021-07-29 2021-11-09 中国人民解放军国防科技大学 基于拓扑感知的高性能计算系统运行环境部署加速方法及系统
CN113625217A (zh) * 2021-07-16 2021-11-09 国网浙江省电力有限公司营销服务中心 低压台区电能表运行误差计算方法及系统
CN116467610A (zh) * 2023-03-13 2023-07-21 深圳市壹通道科技有限公司 基于5g消息的数据拓扑分析方法、装置、设备及存储介质

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103716187B (zh) * 2013-12-20 2017-03-29 新浪网技术(中国)有限公司 网络拓扑结构确定方法和系统
CN105306291B (zh) * 2015-09-16 2018-09-04 电子科技大学 一种基于丢包率和时延联合的网络拓扑估计方法
CN108009391A (zh) * 2017-05-29 2018-05-08 兰州交通大学 一种多尺度下点群目标相似度计算方法
CN111181864B (zh) * 2020-02-27 2023-05-09 上海掌门科技有限公司 确定从应用至服务器的链路拥塞状态的方法与设备
CN112906297A (zh) * 2021-02-05 2021-06-04 青岛启超微信息科技有限公司 一种基于电流大数据的多阈值拓扑识别方法
CN113489619B (zh) * 2021-09-06 2021-11-19 中国人民解放军国防科技大学 一种基于时间序列分析的网络拓扑推断方法及装置

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101605064A (zh) * 2009-07-10 2009-12-16 西北工业大学 基于端到端测量的网络拓扑结构构建方法
CN101917310A (zh) * 2010-08-27 2010-12-15 电子科技大学 基于流量分析的网络拓扑推测方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100195553A1 (en) * 2008-03-18 2010-08-05 Myers Theodore J Controlling power in a spread spectrum system
CN101729365B (zh) * 2008-10-27 2012-06-27 华为技术有限公司 路径建立方法、承载网路由器和网络系统

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101605064A (zh) * 2009-07-10 2009-12-16 西北工业大学 基于端到端测量的网络拓扑结构构建方法
CN101917310A (zh) * 2010-08-27 2010-12-15 电子科技大学 基于流量分析的网络拓扑推测方法

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
LI, YONGJUN ET AL.: "Research on Network Topology Inference Algorithm Based on End-to-End Packet Loss", JOURNAL ON COMMUNICATIONS, vol. 28, no. 10, October 2007 (2007-10-01), pages 85 - 91 *
ZHAO, HONGHUA ET AL.: "Multiple Parameters Network Topology Inference Based on Tomograpgy", JOURNAL OF BEIJING UNIVERSITY OF POSTS AND TELECOMMUNICATIONS, vol. 31, no. 4, August 2008 (2008-08-01), pages 24 - 28 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111626459A (zh) * 2019-02-27 2020-09-04 武汉国测数据技术有限公司 一种区域充电网络拓扑关系确认方法、装置和系统
CN111626459B (zh) * 2019-02-27 2024-04-30 深圳电蚂蚁数据技术有限公司 一种区域充电网络拓扑关系确认方法、装置和系统
CN113078630A (zh) * 2020-11-20 2021-07-06 中国科学院沈阳自动化研究所 一种基于实时量测数据的低压配电网拓扑辨识方法
CN113078630B (zh) * 2020-11-20 2023-12-15 中国科学院沈阳自动化研究所 一种基于实时量测数据的低压配电网拓扑辨识方法
CN112986658A (zh) * 2021-01-25 2021-06-18 浙江华云信息科技有限公司 一种低压线路的拓扑识别方法及装置
CN112986658B (zh) * 2021-01-25 2023-07-18 浙江华云信息科技有限公司 一种低压线路的拓扑识别方法及装置
CN113625217A (zh) * 2021-07-16 2021-11-09 国网浙江省电力有限公司营销服务中心 低压台区电能表运行误差计算方法及系统
CN113625217B (zh) * 2021-07-16 2023-09-26 国网浙江省电力有限公司营销服务中心 低压台区电能表运行误差计算方法及系统
CN113630269A (zh) * 2021-07-29 2021-11-09 中国人民解放军国防科技大学 基于拓扑感知的高性能计算系统运行环境部署加速方法及系统
CN113630269B (zh) * 2021-07-29 2023-11-10 中国人民解放军国防科技大学 基于拓扑感知的高性能计算系统运行环境部署加速方法及系统
CN116467610A (zh) * 2023-03-13 2023-07-21 深圳市壹通道科技有限公司 基于5g消息的数据拓扑分析方法、装置、设备及存储介质
CN116467610B (zh) * 2023-03-13 2023-10-10 深圳市壹通道科技有限公司 基于5g消息的数据拓扑分析方法、装置、设备及存储介质

Also Published As

Publication number Publication date
CN103326904A (zh) 2013-09-25
CN103326904B (zh) 2016-12-28

Similar Documents

Publication Publication Date Title
WO2015003414A1 (zh) 一种基于多参数认知的快速网络拓扑推断方法
CN109005055B (zh) 基于多尺度拓扑空间的复杂网络信息节点重要度评价方法
US10333854B2 (en) Method and apparatus for detecting type of network data flow
WO2012167644A1 (zh) 网络状态上报方法、检测方法、装置和系统
CN107276793B (zh) 基于概率跳转随机游走的节点重要性度量方法
CN109902203A (zh) 基于边的随机游走的网络表示学习方法和装置
CN108696453B (zh) 轻量级的电力通信网sdn业务流感知方法及系统
CN106789333B (zh) 一种基于时间分层的复杂网络传播源点定位方法
US20060282546A1 (en) System and method for inferring causal paths in a distributed computing environment
Gao et al. iPath: Path inference in wireless sensor networks
WO2018142703A1 (ja) 異常要因推定装置、異常要因推定方法及びプログラム
WO2020020098A1 (zh) 网络流测量的方法、网络测量设备以及控制面设备
CN104348665A (zh) 基于聚类判断准则的节点对融合网络拓扑估计方法
CN113537788A (zh) 一种基于病毒传播理论的城市交通拥堵识别方法
CN104270283B (zh) 一种基于高阶累积量的网络拓扑估计方法
US20160125005A1 (en) Apparatus and Method for Profiling Activities and Transitions
CN111342991B (zh) 基于跨社交网络的信息传播方法
CN108400907B (zh) 一种不确定网络环境下的链路丢包率推理方法
CN112597699B (zh) 一种融入客观赋权法的社交网络谣言源识别方法
Wu et al. On-demand Intelligent Routing Algorithms for the Deterministic Networks
Hagos et al. Classification of delay-based TCP algorithms from passive traffic measurements
Fei et al. Improving maximum-likelihood-based topology inference by sequentially inserting leaf nodes
Lin et al. Queuing network topology inference using passive measurements
CN115118610B (zh) 一种基于网路层析成像的动态拓扑估计方法
Xu et al. An adaptive spatial outlier detection algorithm with no parameter for WSN

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: 13889082

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 28/06/2016)

122 Ep: pct application non-entry in european phase

Ref document number: 13889082

Country of ref document: EP

Kind code of ref document: A1