Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberCN102640472 A
Publication typeApplication
Application numberCN 200980162918
PCT numberPCT/IB2009/007753
Publication date15 Aug 2012
Filing date14 Dec 2009
Priority date14 Dec 2009
Also published asCN102640472B, EP2514171A1, US9049199, US9462051, US20120290677, US20150215396, WO2011073707A1
Publication number200980162918.3, CN 102640472 A, CN 102640472A, CN 200980162918, CN-A-102640472, CN102640472 A, CN102640472A, CN200980162918, CN200980162918.3, PCT/2009/7753, PCT/IB/2009/007753, PCT/IB/2009/07753, PCT/IB/9/007753, PCT/IB/9/07753, PCT/IB2009/007753, PCT/IB2009/07753, PCT/IB2009007753, PCT/IB200907753, PCT/IB9/007753, PCT/IB9/07753, PCT/IB9007753, PCT/IB907753
InventorsA.约翰逊, A.达莫拉, H.普塔拉特, S.黑尔奎斯特
Applicant瑞典爱立信有限公司
Export CitationBiBTeX, EndNote, RefMan
External Links: SIPO, Espacenet
Dynamic cache selection method and system
CN 102640472 A
Abstract
Node, computer software and method for selecting a resource [20] that is available at multiple caches [14] connected in a communication network [80]. The method includes receiving from a user [18] a request for the resource [20]; identifying one or more caches [14a, 14b, 14c] of the multiple caches [14] that store the resource [20]; determining a total cost associated with a path between the user [18] and each cache [14a, 14b, 14c] of the one or more caches [14] storing the resource [20], the total cost including a static cost [Cf] that reflects a topology of the communication network [80] and a dynamic cost [?] which is indicative of changing conditions of the communication network [80]; and selecting one cache [14a] from the one or more caches [14a, 14b, 14c] that store the resource [20] based on a lowest total cost of the one cache [14a].
Claims(20)  translated from Chinese
1. 一种用于选择在通信网络[80]中连接的多个缓存[14]可用的资源[20]的方法,所述方法包括: 从用户[18]接收对所述资源[20]的请求; 标识所述多个缓存[14]中存储所述资源[20]的一个或多个缓存[14a、14b、14c]; 确定所述用户[18]与存储所述资源[20]的一个或多个缓存[14]中的每个缓存[14a、14b、14c]之间的路径所关联的总成本,其中所述总成本包括反映所述通信网络[80]的拓扑的静态成本[Cf]和指示所述通信网络[80]的变化状况的动态成本[Λ];以及基于一个缓存[14a]的最低总成本从存储所述资源[20]的一个或多个缓存[14a、14b、14c]选择所述一个缓存[14a]。 1. A method for selecting a plurality of buffers in a communications network [80] in connection [14] available resources [20], the method comprising: receiving from a user [18] on the resources [20] identifying the plurality of cache [14] is stored in the resource [20] One or more cache [14a, 14b, 14c];; determining the user [18] and the storage resource [20] a request or a plurality of cache [14] Each buffer [14a, 14b, 14c] between the total cost associated with a path, wherein the total cost includes reflecting said communication network [80] Cost of static topology [Cf Dynamic Cost] and indicates that the communications network [80] change the status [Λ]; and the lowest total cost based on a cache [14a] A from the memory of the resources [20] or more cache [14a, 14b, 14c] to select the one cache [14a].
2.如权利要求I所述的方法,还包括: 将所述通信网络的节点之间的真实路径转换成在任何缓存对之间包括单个路径的虚拟模型,其中所述节点将所述多个缓存彼此连接。 2. The method of claim I, further comprising: converting the real path between nodes of the communication network between the virtual model into any cache including a single path, wherein said plurality of said nodes cache connected to each other.
3.如权利要求2所述的方法,还包括: 对所述虚拟模型中的每个路径赋予指派成本,其中所述指派成本是运营商定义的量且确定所述静态成本。 3. The method of claim 2, further comprising: assigning costs given the virtual model of each path, wherein said cost is the assignment operator-defined amount and determining said static cost.
4.如权利要求3所述的方法,还包括: 接收表征所述真实路径的性能的参数; 生成表征任何缓存对的状态矩阵;以及基于所接收的参数来生成所述动态成本。 4. The method of claim 3, further comprising: receiving said real path characterizing performance parameters; generating a state matrix characterizing any caching; and based on the received parameters to generate the dynamic cost.
5.如权利要求4所述的方法,其中所接收的参数包括抖动、分组丢失、往返时间、IP层拓扑、路径可用容量等中的一个或多个。 5. The method of claim 4, wherein the received parameters including jitter, packet loss, round-trip time, IP layer topology, paths available capacity in one or more.
6.如权利要求4所述的方法,还包括: 通过使用以下中的一个或多个来填充所述状态矩阵的元素:(i)自动测量系统,其中给定某个时间段两个节点以某个概率来在彼此之间执行测量;(ii)作为测量探测的应用数据;以及(iii)所述网络的节点的被动监视。 6. The method of claim 4, further comprising: by using the following elements in one or more of the state of filling of the matrices: (i) automatic measurement system, in which two nodes in a given period of time to a probability measurement is performed between one another; (ii) use as measurement data detected; passive and (iii) the monitoring of the network node.
7.如权利要求6所述的方法,其中所述状态矩阵的元素至少包括所述网络的测量参数、所述测量参数的方差、与进行所述测量的时间关联的时间戳以及指示自此后不能够使用所述测量参数的时间的超时参数。 7. The method of claim 6, wherein said state matrix comprises elements of at least the measured parameters of the network, the measurement variance parameters, and perform the measurement time associated with a time stamp indicating not since then It can measure time parameters using the timeout parameter.
8.如权利要求4所述的方法,还包括: 计算相同路径的动态成本与静态成本之间的差; 对所计算的差应用可信度函数以使所计算的差为时间相关的并确定所计算的差是否在预定范围内;以及在成本函数中组合与所接收的参数关联的动态成本和与所述指派成本关联的静态成本以确定所述总成本。 8. The method of claim 4, further comprising: a difference calculating the cost of the same dynamic and static path between the cost; for poor application reliability function calculated so that the difference between the calculated and determined as the time-dependent the calculated difference is within a predetermined range; and combinations of dynamic cost in the cost function associated with the received parameters and costs associated with the assigned cost to determine the static total cost.
9.如权利要求8所述的方法,还包括: 将所述成本函数计算为(i)所述静态成本与(ii)所计算的差与所述可信度函数的积之间的和。 9. The method of claim 8, further comprising: calculating the cost function of (i) the cost and volume and static (ii) the difference between the calculated and the reliability function between.
10.如权利要求9所述的方法,还包括: 选择所述可信度函数以产生仅在零与一之间的数。 10. The method of claim 9, further comprising: selecting the function to produce only a few reliability between zero and one.
11. 一种用于选择在通信网络[80]中连接的多个缓存[14]可用的资源[20]的节点[14、1600],所述节点[14、1600]包括: 接口[1608],配置成连接到所述通信网络[80]的其它节点[12、14]和/或多个缓存[14];以及处理器[1602],连接到所述接口[1608]并配置成: 从用户[18]接收对所述资源[20]的请求, 标识所述多个缓存[14]中存储所述资源[20]的一个或多个缓存[14、14b、14c],确定所述用户[18]与存储所述资源[20]的一个或多个缓存[14]中的每个缓存[14、14b、14c]之间的路径所关联的成本,其中所述成本包括反映所述通信网络[80]的拓扑的静态成本[Cf]和指示所述通信网络[80]的变化状况的动态成本[△],&& 基于一个缓存[14a]的最低成本从存储所述资源[20]的一个或多个缓存[14、14b、14c]选择所述一个缓存[14a]。 Node [14,1600] 11. A method for selecting a communication network in a plurality of cache [80] is connected to [14] available resources [20], and the node [14,1600] comprising: an interface [1608] configured to other nodes connected to the communication network [80] of [12, 14] and / or a plurality of cache [14]; and a processor [1602], connected to said interface [1608] and is configured to: From User [18] receiving a request for the resource [20], and identifying the plurality of cache [14] is stored in the resource [20] One or more cache [14,14b, 14c], determines the user [18] and the storage resource [20] One or more cache [14] Each buffer [14,14b, 14c] between the associated cost path, wherein said cost reflecting the communication comprises Dynamic Cost Cost static network [80] The topology [Cf] and instruct the communication network [80] change status [△], && cache based on a [14a] from the storage of the lowest cost resource [20] One or more cache [14,14b, 14c] select the one cache [14a].
12.如权利要求11所述的节点,其中所述处理器还配置成: 将所述通信网络的节点之间的真实路径转换成在任何缓存对之间包括单个路径的虚拟模型,其中所述节点将所述多个缓存彼此连接。 12. The node according to claim 11, wherein said processor is further configured to: convert the real path between nodes of the communication network between the virtual model into any cache including a single path, wherein the node connected to each of the plurality of cache.
13.如权利要求12所述的节点,其中所述处理器还配置成: 对所述虚拟模型中的每个路径赋予指派成本,其中所述指派成本是运营商定义的量且确定所述静态成本。 13. The node according to claim 12, wherein said processor is further configured to: assign costs given the virtual model of each path, wherein said cost is the assignment operator-defined amount and determining said static costs.
14.如权利要求13所述的节点,其中所述处理器还配置成: 接收表征所述真实路径的性能的参数; 生成表征任何缓存对的状态矩阵;以及基于所接收的参数来生成所述动态成本。 14. The node according to claim 13, wherein said processor is further configured to: receive characterize the performance of the real path parameters; generating a state matrix characterizing any caching; and based on the received parameters to generate the Dynamic cost.
15.如权利要求14所述的节点,其中所接收的参数包括抖动、分组丢失、往返时间、IP层拓扑、路径可用容量等中的一个或多个。 15. The node according to claim 14, wherein the received parameters including jitter, packet loss, round-trip time, IP layer topology, paths available capacity in one or more.
16.如权利要求14所述的节点,其中所述处理器还配置成: 通过使用以下中的一个或多个来填充所述状态矩阵的元素:(i)自动测量系统,其中给定某个时间段两个节点以某个概率来在彼此之间执行测量;(ii)作为测量探测的应用数据;以及(iii)所述网络的节点的被动监视。 16. The node according to claim 14, wherein said processor is further configured to: by using the element in one or more of the following to populate the matrix state: (i) automatic measurement system, in which a given In a period of two nodes to perform the measurement probability among themselves; (ii) use as measurement data detected; passive and (iii) the monitoring of the network node.
17.如权利要求16所述的节点,其中所述状态矩阵的元素至少包括所述网络的测量参数、所述测量参数的方差、与进行所述测量的时间关联的时间戳以及指示自此后不能够使用所述测量参数的时间的超时参数。 17. The node according to claim 16, wherein said state matrix comprises elements of at least a measure of the network parameters, parameters of the measurement variance, performed the measurements with the associated time stamp indicating the time since then no It can measure time parameters using the timeout parameter.
18.如权利要求14所述的节点,其中所述处理器还配置成: 计算相同路径的动态成本与静态成本之间的差; 对所计算的差应用可信度函数以使所计算的差为时间相关的,以及确定所计算的差是否在预定范围内;以及在成本函数中组合与所接收的参数关联的动态成本和与所述指派成本关联的静态成本以确定所述总成本。 18. The node of claim 14, wherein the processor is further configured to: calculate the difference between the cost of the same dynamic and static path between the cost; the difference between the calculated application of reliability functions to make the difference between the calculated for the time dependent, and determining whether the calculated difference is within a predetermined range; and combinations associated with the received parameters in the cost function and assigned dynamic cost associated with the costs to determine the total cost of the static cost.
19.如权利要求16所述的节点,其中所述处理器还配置成: 将所述成本函数计算为(i)所述静态成本与(ii)所计算的差与所述可信度函数的积之间的和 19. The node according to claim 16, wherein said processor is further configured to: calculate the cost function of (i) the cost of the static and (ii) the difference between the calculated and the reliability function and between product
20. 一种包括计算机可运行指令的计算机可读介质,其中所述指令在被运行时实现用于选择在通信网络[80]中连接的多个缓存[14]可用的资源[20]的方法,所述方法包括:从用户[18]接收对所述资源[20]的请求; 标识所述多个缓存[14]中存储所述资源[20]的一个或多个缓存[14a、14b、14c];确定所述用户[20]与存储所述资源[20]的一个或多个缓存[14]中的每个缓存[14a、14b、14c]之间的路径所关联的成本,其中所述成本包括反映所述通信网络[80]的拓扑的静态成本[Cf]和指示所述通信网络[80]的变化状况的动态成本[Λ];以及基于一个缓存[14a]的最低成本从存储所述资源[20]的一个或多个缓存[14a、14b、14c]选择所述一个缓存[14a]。 20. A computer comprising a computer readable medium operable instructions, wherein said instructions are run to achieve a plurality of buffers for selecting a communication network [80] is connected to [14] available resources [20] The method , the method comprising: receiving a request for the resource [20] from the user [18]; identifying the plurality of cache [14] in the storage resource [20] One or more cache [14a, 14b, 14c]; determining the user [20] and the storage resource [20] One or more cache [14] Each buffer [14a, 14b, cost path] between 14c associated, wherein said dynamic costs include costs reflect the communications network [80] Cost static topology [Cf] and indicates that the communications network [80] change the status [Λ]; and based on a cache [14a] the lowest cost from storage The resources [20] One or more cache [14a, 14b, 14c] select the one cache [14a].
Description  translated from Chinese

动态缓存选择方法和系统 Method and system for dynamic cache selection

技术领域 Technical Field

[0001] 本发明一般涉及系统、软件和方法,以及更具体来说涉及用于动态地选择具有期望内容的缓存的机制和技术。 [0001] The present invention relates generally to systems, software and methods, and more particularly to the cache for dynamically selecting a desired content of the mechanisms and techniques.

背景技术 Background

[0002] 在过去数年间,社交网络和内容交付网络(content delivery network, Q)N,其在本文中用于涵盖内容交付和内容分发网络二者)的发展是引人注目的。 [0002] In the development of the past few years, social networking and content delivery network (content delivery network, Q) N, which is used to cover the content delivery and content distribution network both in this article) is striking. 越来越多的人优选经由社交网络来联系同辈或朋友或经由CDN来下载内容。 More and more people preferred to contact via social networks or via CDN peers or friends to download content. 内容类型包括万维网对象、可下载对象(媒体文件、软件、文档)、应用、实时媒体流和因特网交付的其它组件(DNS、路由和数据库查询)。 Content types include Web objects, downloadable objects (media files, software, documentation), applications, real-time media streaming and other components of Internet delivery (DNS, routing and database queries).

[0003] 手持装置(大多数为移动电话)的发展进度使得从CDN下载内容的过程成为很多用户的常见任务。 [0003] The handheld devices (mostly mobile phones) development progress makes the process of downloading content from CDN become many common tasks the user. 但是,正如接下来论述的,存在CDN智能地将数据文件分配到缓存子集的容量的限制。 However, as discussed next, there is a CDN intelligently allocate data files to limit the capacity of the cache subset.

[0004] CDN是存储数据的副本的、放置在网络中的各种点以便将用于从遍布网络的客户端访问数据的带宽最大化的计算机的系统。 [0004] CDN is a copy of the data, place the various points in the network to be used throughout the bandwidth of the network from the client to access the data to maximize your computer's system. 客户端访问接近该客户端的数据副本,与所有客户端访问相同中央服务器相反,以便避免接近该服务器的瓶颈。 Client Access close copy of the client's data, with all the clients accessing the same central server contrary, in order to avoid going near the server bottleneck. 但是,客户端访问“接近”该客户端的数据的概念并不非常有效率,因为CDN的当前实现似乎缺少智能机制来将数据文件分配到缓存的子集。 However, Client Access "close" to the concept of data the client is not very efficient, because the current implementation seems to be a lack of CDN intelligent mechanism to assign data files to the cached subset. 现有CDN所使用的ー个原理是只能从客户端直接附接到的边缘节点中的缓存提取数据文件。ー a principle used in the existing CDN edge node can be attached directly from the client cache in extracting data files. 如果客户端需求的数据文件未存储在该边缘节点中,则CDN首先需要将期望的数据文件缓存在那里。 If the client needs the data files are not stored in the edge node, the CDN will first need to desired data file cache there.

[0005] 为了更好地理解传统⑶N网络,图I示出包括多个节点12和14的真实世界网络10的简化示例。 [0005] For a better understanding of traditional ⑶N network, Figure I shows a simplified example comprises a plurality of nodes 12 and 14, 10 of the real-world network. 最下一行的圆表示边缘节点14,而其余的圆表示中间节点12。 The next line circle represents the edge node 14, while the rest of the circle indicates the intermediate node 12. 假定每个最终用户计算机(客户端)可以连接到仅ー个边缘节点14。 Assume that each end-user computer (client) can connect to only one edge node 14 ー. 由图I中的从客户端的计算机或移动装置18到对应边缘节点14的虚线来指示这种连接16。 By a computer or mobile device in Figure I from the client to the corresponding edge node dotted line 18 14 16 to indicate this connection. 还假定仅将内容或数据文件缓存在边缘节点14上。 Also assume that only the content or data files are cached on the edge node 14. 在此特殊示例中,虚线22指示将ー个特定数据文件20存储在三个缓存14a、14b和14c中。 In this particular example, the broken line 22 indicates the ー a specific data file 20 stored in the three cache 14a, 14b and 14c. 注意,虚线22不表示通信链路。 Note that the dotted line does not represent a communication link 22.

[0006] 网络10中的中间节点12将全世界的边缘节点14彼此连接。 [0006] The network 10 to intermediate node 12 connected to each edge node 14 over the world. 节点12和14之间的线是这些节点之间的真实链路。 Lines 12 and 14 are real link between nodes between the nodes. 在实践中,所有节点12和14可以是路由器。 In practice, all the nodes 12 and 14 can be a router. 真实节点12和14之间的每个链路24与由字母“c”指示的通信成本关联。 Costs associated with real link 24 nodes each between 12 and 14 by the letter "c" indicates the communication. 两个边缘节点14之间的通信路径可以取网络10中的不同途径,如图I中所示。 A communication path between two edge nodes 14 may take different routes in the network 10, as shown in Figure I below.

[0007] 因此,当用户18向用户18连接到的边缘节点14d请求数据文件20吋,假定节点14d没有该数据文件,则边缘节点14d需要标识哪个其它节点具有该数据文件,以及哪个节点最接近。 [0007] Thus, when the user 18 is connected to the user 18 to the edge node 14d request data file 20 inches, assuming the absence of the data file node 14d, 14d which is an edge node to other nodes need to be identified with that data file, and which node is closest . 现有CDN网络可以将与链路24关联的静态成本納入考虑。 CDN networks can be incorporated into existing link 24 associated with static cost considerations. 但是,在网络的状况中的任何变化未在现有静态成本中反映,因为此成本不会变化,并且因此,现有CDN网络缺乏此信息并且无法精确地描述实况网络。 However, any change in the condition of the network is not reflected in the cost of the conventional static, because the cost does not change, and thus, the existing CDN network and the lack of this information can not be accurately described in the live network.

[0008] 因此,需要找到ー种方法将请求的数据本地化到存储该数据文件的副本的“最接近”缓存。 [0008] Thus, the need to find ways ー localization of the requested data to store a copy of the data file is "closest" to the cache. 在此情况中,最接近意味着该缓存节点至客户端连接到的边缘节点具有最小成本的路径。 In this case, it means that the cache closest to the client node connected to the edge node has a path of least cost. 现有CDN中所没有的另一个特征是在给定网络中有限存储量的情况下且为了优化网络资源,能够决定将哪个数据文件存储在哪些边缘节点上的智能机制。 Another feature not available in the existing CDN in the case of a given network and the limited amount of memory in order to optimize network resources, the smart mechanism to decide which data files are stored on which edge node. [0009] 现有CDN的另一个问题在于,虽然有若干个主动端到端探测方法(用于表征网络链路或路径的能力的通用技术),并且新方法正在持续地开发和精化,但是从此类测量获益的应用经常要求比仅仅端到端估计更多的知识。 [0009] Another existing problem is that the CDN, although there are a number of active end detection method (used to characterize the ability of a network link or path common technique), and new methods are continuously developed and refined, but benefit from the application of such measurements often require more than just knowledge-end estimate. 另一个问题是,对于如何维护和汇总来自若干源的测量数据,则没有明确的协议。 Another problem is how to maintain and aggregated measurement data from a number of sources, there is no clear agreement. 例如,CDN实现常常依赖于一些现有方法以推断用于服务内容请求的“最佳”源节点。 For example, CDN achieve some prior methods often rely on to infer for the "best" source node content service requests. 为了CDN能够基于网络测量进行智能内容迁移,需要在所有节点之间保持相干测量统计的系统。 In order to be able to network-based measurements CDN intelligent content migration, the need to maintain coherence between all nodes measuring statistical system.

[0010] 因此,会期望提供避免前述问题和缺点的装置、系统和方法。 [0010] Accordingly, it is desirable to provide avoids the aforementioned problems and disadvantages of devices, systems and methods.

发明内容 DISCLOSURE

[0011] 期望将资产(例如,内容)从对于用户和/或运营商最适合的位置提供到用户。 [0011] desired to assets (eg, content) from the user and / operators are most suitable location or provide to the user. 适合的位置可以是对网络的运营商最便宜的位置。 Suitable location can be on the network operator cheapest location. 以此方式,从对网络的运营商最便宜的位置接收内容释放了网络中的带宽和/或平衡网络的路径。 In this way, from the network operator receives the cheapest location content release the network path bandwidth and / or the balance of the network. 还可以通过优化如流传输媒体的某些应用所要求的路径特征来确定适合的位置。 Also suitable location can be determined by optimizing the flow path characteristics for some applications such as the transmission medium requires.

[0012] 根据一示范实施例,有一种方法用于选择在通信网络中连接的多个缓存可用的资源。 [0012] According to an exemplary embodiment, there is a method for selecting a plurality of buffers in a communications network connecting the available resources. 该方法包括从用户接收对资源的请求;标识多个缓存中存储该资源的一个或多个缓存;确定用户与存储该资源的一个或多个缓存中的每个缓存之间的路径所关联的总成本,该总成本包括反映通信网络的拓扑的静态成本和指示通信网络的变化状况的动态成本;以及基于一个缓存的最低总成本从存储该资源的一个或多个缓存选择该一个缓存。 The method includes receiving a request for resources from the user; identifying a plurality of cache stores one or more caches the resource; determining user and path cache between each cache stores the resources associated with one or more The total cost, the total cost includes the cost of changes in the status of dynamic and static costs reflect indicating a communication network communication network topology; and the lowest total cost based on a cache, select the cache from a cache to store one or more of the resource.

[0013] 根据另一个示范实施例,有一种节点用于选择在通信网络中连接的多个缓存可用的资源。 [0013] In accordance with another exemplary embodiment, there is a node for selecting a plurality of buffers connected in a communications network available resources. 该节点包括配置成连接到通信网络的其它节点和/或多个缓存的接口;以及连接到该接口的处理器。 The node is configured to include other nodes connected to the communication network and / or a plurality of cache interface; and a processor connected to the interface. 该处理器配置成从用户接收对资源的请求;标识多个缓存中存储该资源的一个或多个缓存;确定用户与存储该资源的一个或多个缓存中的每个缓存之间的路径所关联的总成本,该总成本包括反映通信网络的拓扑的静态成本和指示通信网络的变化状况的动态成本;以及基于一个缓存的最低总成本从存储该资源的一个或多个缓存选择该一个缓存。 The processor is configured to receive a request for a resource from a user; identifying a plurality of cache buffer that stores one or more resources; to determine the path of one or more cache user and stores the resources in each cache between the The total cost associated with the change in the total cost includes the cost of dynamic conditions of static and instructs the communication network costs reflect communications network topology; and based on the lowest total cost of a cache buffer to select one from the storage of the one or more resource caches .

[0014] 根据又一个示范实施例,有一种包括计算机可运行指令的计算机可读介质,所述指令在被运行时,实现一种用于选择在通信网络中连接的多个缓存可用的资源的方法。 [0014] According to yet another exemplary embodiment, there is a computer including a computer-readable medium run command, the command being run to achieve a method for selecting a plurality of buffers connected in a communications network resources available Methods. 该方法包括从用户接收对资源的请求;标识多个缓存中存储该资源的一个或多个缓存;确定用户与存储该资源的一个或多个缓存中的每个缓存之间的路径所关联的总成本,该总成本包括反映通信网络的拓扑的静态成本和指示通信网络的变化状况的动态成本;以及基于一个缓存的最低总成本从存储该资源的一个或多个缓存选择该一个缓存。 The method includes receiving a request for resources from the user; identifying a plurality of cache stores one or more caches the resource; determining user and path cache between each cache stores the resources associated with one or more The total cost, the total cost includes the cost of changes in the status of dynamic and static costs reflect indicating a communication network communication network topology; and the lowest total cost based on a cache, select the cache from a cache to store one or more of the resource.

[0015]目的在于克服前文中论述的一些缺点,以及提供能够发现对于用户更适合位置的缓存的节点。 [0015] aimed at overcoming some of the shortcomings discussed previously, as well as provide the ability to find a more suitable location for the user node cache. 一个或多个独立权利要求有利地提供最佳缓存以响应用户的请求。 One or more independent claims advantageously provides optimal cache in response to user requests.

附图说明 Brief Description

[0016] 并入本说明书中并构成其一部分的附图示出一个或多个实施例,并且连同描述来说明这些实施例。 BRIEF [0016] are incorporated in and constitute a part of illustrating one or more embodiments, and together with the description of these embodiments will be described. 在附图中: In the drawings:

图I是传统CDN的示意图; Figure I is a schematic view of a traditional CDN;

图2是根据一示范实施例的状态矩阵的说明; 图3是发送方和接收方之间的路径的示意图; Figure 2 is an exemplary embodiment according to the state matrix description; Figure 3 is a schematic diagram of the path between the sender and the recipient;

图4是彼此连接且具有不同级别的可用带宽的三个不同链路的示意图; Figure 4 is a schematic view of the available bandwidth of the links connected to each other in three different and having different levels;

图5是示出网络的节点之间的主动探测方法的示意图; Figure 5 is a diagram showing active detection method between nodes of the network;

图6是示出由根据一示范实施例的状态矩阵收集到什么信息的示意图; Figure 6 is a schematic view of a state according to the exemplary embodiment of the matrix to collect information about what is shown;

图7是根据一示范实施例的状态矩阵的元素的说明; Figure 7 illustrates an exemplary embodiment according to the state of the matrix element;

图8是对应于真实CDN网络的虚拟网络的示意图; 8 is a diagram corresponding to the virtual network real CDN networks;

图9示出根据一示范实施例的与可用带宽估计关联的动态成本; Figure 9 shows a dynamic cost and available bandwidth associated with the embodiment in accordance with an exemplary estimation;

图10示出根据一示范实施例的与链路容量关联的指派成本; Figure 10 shows a cost associated with the link capacity assignment in accordance with an exemplary embodiment;

图11示出根据一示范实施例的静态成本与动态成本之间的成本差; Figure 11 illustrates an exemplary cost of static and dynamic cost cost difference between the embodiment;

图12示出根据一示范实施例的与成本差关联的可信度函数; Figure 12 shows the difference between the cost and reliability of the associated function in accordance with an exemplary embodiment;

图13示出根据一示范实施例的CDN网络执行的用于向用户提供期望资源的各种步 Figure 13 shows a user to provide a desired resource CDN network according to an exemplary embodiment of the execution of the various steps

骤; Step;

图14示出根据一示范实施例的收集有关CDN网络的数据的决策引擎; Figure 14 illustrates an exemplary decision engine to collect data relating to CDN network according to an embodiment of the;

图15是示出根据一示范实施例的用于选择期望资源的方法的流程图;以及图16是配置成实现图14和/或图15中所示的步骤的节点的示意图。 15 is a flowchart illustrating an exemplary embodiment according to a method of selecting a desired resource; and Figure 16 is a schematic view of a node configured to steps 14 and / or the implementation shown in FIG. 15.

具体实施方式 DETAILED DESCRIPTION

[0017] 示范实施例的下文描述涉及附图。 Described below [0017] The exemplary embodiment relates to the accompanying drawings. 不同附图中的相同引用号标识相同或相似的单元。 Different drawings of the same reference numbers identify the same or similar elements. 下文详细描述并不限制本发明。 Detailed description does not limit the invention. 而是,本发明的范围由所附权利要求书限定。 Rather, the scope of the invention defined by the appended claims. 为了简明,下文实施例是关于CDN网络的术语和结构来论述的。 For simplicity, the examples hereinafter is about terminology and structure to discuss the CDN network. 但是,接下来要论述的实施例不限于这些网络,而是可以应用于其它现有网络。 However, the embodiments to be discussed next are not limited to these networks, but may be applied to other existing networks.

[0018] 本说明书通篇对“一个实施例”或“一实施例”的引述意味着结合一实施例描述的特殊特性、结构或特征被包括在本发明的至少一个实施例中。 [0018] Reference throughout this specification to "one embodiment" means that a particular feature or refer to an embodiment described herein, structure, or characteristic "an embodiment" is included in at least one embodiment of the present invention. 因此,本说明书通篇各种位置中出现的短语“在一个实施例中”或“在一实施例中”不一定全部指相同的实施例。 Accordingly, a variety of locations throughout this specification appearing phrases "in one embodiment" or "in an embodiment" are not necessarily all referring to the same embodiment. 另外,可以任何适合的方式在一个或多个实施例中进行组合特殊特性、结构或特征。 In addition, any suitable manner in one or more embodiments are combined special characteristics, structure, or characteristic.

[0019] 根据一示范实施例,一种方法和网络配置成使用捕获网络的一个或多个属性的成本函数。 [0019] According to an exemplary embodiment, a method and network configuration to use to capture network of one or more attributes of the cost function. 该成本函数包括从主动网络测量推导的成本与网络的真实和/或虚拟节点拓扑的组合。 The cost function includes a combination of real and / or virtual node topology derived from the active network measurement and network costs. 该主动网络测量方法(例如,参见2009年的爱思唯尔计算机网络(EI seVier ComputerNetworks,COMNET)53卷15期2617-2645页的Bergfeldt等人的使用过滤和变化检测的实时可用带宽估计,,其全部内容通过引用并入本文)将测量数据存储在稍后定义的状态矩阵中。 The active network measurement method (for example, using real-time filtering and change detection see the available bandwidth estimate for 2009 Elsevier computer network (EI seVier ComputerNetworks, COMNET) 53 volume 15 pages 2617-2645 Bergfeldt et al. ,, the entire contents of which are incorporated herein by reference) measurement data is stored is defined in a later state matrix. 主动网络测量将网络的状态纳入考虑。 Active network measurement into account the state of the network. 基于测量,将动态成本与边缘节点之间的路径关联,而真实和/或虚拟节点拓扑产生静态成本。 Based on the measurement, the path between the dynamic costs associated with the edge node, and the real and / or virtual node topology generate static costs. 可以由运营商修改静态成本,正如接下来论述的。 You can modify the static costs by the operator, as discussed next. 由决策引擎将每个成本组合以形成总成本。 By a decision engine will each cost combined to form the total cost. 因此,在CDN中基于包括动态成本和静态成本的总成本来确定由客户端提取的内容或缓存之间的内容迁移。 Therefore, based on CDN include static and dynamic cost cost determine the contents of the assembly have been extracted by the client or cached content between migration. 该决策引擎可以基于状态矩阵、拓扑和可信度函数来执行成本估计,正如接下来论述的。 The decision engine may be based on the state of the matrix, topology, and reliability functions to perform cost estimates, as discussed next.

[0020] 根据另一个示范实施例,引入状态矩阵,并且状态矩阵表示分布式系统的网络测量数据并将其汇总。 [0020] According to another exemplary embodiment, the introduction of the state of the matrix, and the state matrix network measurement data distributed systems and summary. 此信息能够用于选择要在CDN中的两个通信节点之间(例如客户端与缓存节点之间)使用的适合的路径或ー组路径。 This information can be used to choose the path that you want (for example, between the client and the caching node) between the two communication CDN nodes or ー group path.

[0021] 在论述如何计算CDN中的边缘节点之间的各种路径的总成本并使其为时间相关的之前,本发明示范实施例论述用于生成状态矩阵的各种方式。 [0021] In the discussion of how to calculate the total cost of CDN various paths between the edge node and make it before the time-related, exemplary embodiment of the present invention discussed various ways for generating a state matrix. 状态矩阵可以定义成包括从ー组节点(N1, N2, ...,Nn}之间的主动探测或被动监视获得的性能參数估计。在表示多于ー个性能參数的情况中,可能并行地使用若干状态矩阵,每个性能參数对应于ー个状态矩阵,或单个状态矩阵中每对节点对应于若干元素。性能參数的示例是端到端可用容量、往返时间(RTT)、抖动、IP层拓扑、路径可用容量和丢失(loss)。图2示出状态矩阵的示例,其中元素Xu描述在两个节点Ni和%之间观察到的特定性能參数方面的路径特征。注意,路径Ni至Nj可以具有与反向路径Nj至Ni相比不同的特性。 State matrix can be defined to include groups from ー nodes (N1, N2, ..., performance parameters between active detection} Nn or passive monitoring estimate obtained in the case of personality can represent more than ー parameters, it may be Several state matrix used in parallel, each performance parameter corresponds ー states matrix, or each pair of nodes corresponding to a single state in the number of elements in the matrix. Example-end performance parameters is available capacity, round-trip time (RTT), jitter, IP layer topology path available capacity and loss (loss). Figure 2 shows an example of the state of the matrix, which is described Xu element between two nodes Ni and% observed in the path of the characteristics of the specific performance parameters. Note , the path to the Ni Nj Nj reverse path may have different characteristics compared to Ni.

[0022] 根据ー示范实施例,状态矩阵中的姆个元素Xij包括至少如下參数:性能參数估计、指示何时进行测量的时间戳以及指示何时测量估计到期的超时阈值。 [0022] According to an exemplary embodiment ー, state matrix elements Xij Farm include at least the following parameters: performance parameter estimation, time stamp indicates when the measurement and indicates when the measurement is estimated to expire timeout threshold. 可选地,该状态估计可以包括估计的方差。 Alternatively, the state may include estimates of the variance estimation. 可以使用估计的方差为状态矩阵的用户提供有关其可信度的信息。 You can use the estimated variance of the state matrix to provide users with information about their credibility. 如稍后将论述的,可以使用其它函数来提供此信息。 As will be discussed later, you can use the other functions to provide this information.

[0023] 能够采用各种方式来填充状态矩阵中的元素。 [0023] capable of using a variety of ways to fill the state matrix elements. 下文描述三个示例。 Three examples are described below. 但是,正如本领域技术人员将认识到的,还存在其它可能性。 However, as those skilled in the art will appreciate, there are other possibilities. 第一和第二示例是基于主动探测,并且可以提供如RTT、抖动、延迟、路径可用容量和丢失的性能參数。 The first and second examples are based on active detection, and provide services such as RTT, jitter, delay, path loss of available capacity and performance parameters. 第三示例是基于被动监视,并且可以提供例如链路负载、队列长度和丢弃分组的其它性能參数。 The third example is based on passive monitoring, and can provide a link load for example, discarding the packet queue length and other performance parameters.

[0024] 接下来要论述的示例可以使用本领域中已知的技木。 [0024] Examples to be discussed next may use the techniques known in the art of wood. 为了完整性,现在简要地论述这些技木。 For completeness, we briefly discuss the technology of wood. 主动探测是通过从发送方向接收方发送分组来表征网络链路或路径的ー种通用技木。 Active detection by the receiver to send a packet from a sender to characterize the network link or path ー types of common technology of wood. 图3中示出了此方法。 Figure 3 shows this method. 更确切地来说,发送方30经由网络32向接收方34发送分组。 More specifically, the sender 30 via the network 32 sends a packet to the receiver 34. 发送方30将分组间间隔36随机化。 The sender of the packet interval 30 36 randomized. 接收方34计算新的分组间间隔38,其然后被分析以确定期望的性能參数。 Receiver 34 calculates the new inter-packet interval 38, which is then analyzed to determine the desired performance parameters. 能够用此方法测量的性能參数例如是抖动、分组丢失、往返时间和路径可用带宽。 Performance parameters can be measured by this method, for example, jitter, packet loss, available bandwidth, and the round trip time. 測量这类性能參数的传统工具例如是各种版本的试通程序(ping)。 Traditional tools such measure performance parameters, for example, through the various versions of the test program (ping). 測量可用容量是更为复杂的问题。 Measurement of available capacity is a more complex issue. 一个这种測量技术是实时可用带宽(BART),并且下ー个段落中对此进行描述。 This measurement technique is a real-time available bandwidth (BART), and the next ー paragraph, be described.

[0025] 作为本领域中已知的技木,BART是ー种用于估计分组交换网络路径上实时的路径可用容量和其它容量有关參数的方法。 [0025] As known in the art 'technology, BART is ー kind of real-time on a packet-switched network path paths available capacity and other capacity-related parameters of a method for estimation. 图4中示出性能參数容量、利用和可用容量的概念,其示出作为网络路径的一部分的三个链路40、42和44。 Figure 4 shows the performance parameters capacity, utilization and available capacity concept, showing part of a network path as the three links 40, 42 and 44. 每个链路具有定义能够发送IP业务的最大速率的标称容量。 Each link has a defined IP services can send a maximum rate of nominal capacity. 在任何给定时间点,可以按小于其最大容量来利用链路。 At any given point in time, it can be less than its maximum capacity to take advantage of the link. 这由与每个链路关联的阴影区域来举例说明。 This is indicated by the shaded area associated with each link exemplified. 对每个链路将可用带宽计算为标称链路带宽与估计的链路利用之差。 The available bandwidth for each link is calculated as the difference between the nominal link bandwidth utilization of the link with the estimated.

[0026] ー个容量有关性能參数是端到端可用容量。 [0026] ー a capacity end of the performance parameters is available capacity. 此參数定义为一组连续链路的最小链路容量。 This parameter is defined as a set of minimum link capacity continuous link. 在图4中,链路40是最小端到端可用容量。 In FIG. 4, the end to end link 40 is the minimum available capacity. BART方法依赖于在网络路径上主动发送探测业务以便确定在哪个探测速率路径显示拥塞的征兆。 BART method relies on a network path sends probe traffic to determine which path detection rate in display signs of congestion. 此速率定义可用容量,即,未被IP业务利用的容量的部分。 This rate is defined portion of the available capacity that is not utilized capacity IP traffic. 如图3中所不,BART发送方30正在按随机化的分组间间隔36向接收方34传送IP分组。 Figure 3 are not, BART sender 30 is press packet interval between randomization 36 IP packets transmitted to the receiver 34. 该间隔受共享网络的其它IP业务的影响。 Sharing networks affected by other IP services that interval. 接收方36对每个入局IP分组标记时间戳,并且计算新的分组间间隔38。 Receiver 36 pairs each incoming IP packets to be time stamped, and calculates the new inter-packet interval 38. 可以通过卡尔曼过滤器(Kalmanfilter)分析在发送方30和接收方34的分组间间隔,这是一种实时地跟踪不可直接观察的特性的统计方法,并且从该分析的输出是可用容量。 By Kalman filter (Kalmanfilter) packet analysis between the sender 30 and receiver 34 at intervals, which is not directly observed statistical tracking characteristics of a real time, and output from the analysis is available capacity.

[0027] 可以由上文提到的第一示例,即自动测量系统使用上文论述的主动探测。 [0027] The first example may be made of the above-mentioned, the automatic measurement system using active detection discussed above. 这种系统的特性是使得给定某个时间段两个节点以某个概率来在彼此之间执行测量。 Characteristics of such a system is such that the two nodes in a given period of time to perform a probability measure between each other. 该测量系统在参与节点上作为后台进程来运行测量。 The measuring system as a background process to run measurements on participating nodes. 如图5中所示,在给定时间点,节点B触发与节点D的测量,而同时节点E触发与节点A的测量。 As shown in FIG. 5, at a given point in time, the node B trigger and measurement node D, while the measurement node E and the node A trigger. 可以触发网络的所有节点执行这类测量,并且最终以每个节点的数据填充状态矩阵。 All network nodes can trigger the execution of such measurements, and eventually fill state of each node in the data matrix.

[0028] 仍基于主动监视的填充状态矩阵的第二示例,使用应用数据作为测量探测。 [0028] The second example is still based on proactive monitoring of the filling state matrix, using the application data as the measurement probe. 例如,在其中缓存彼此往返地迁移数据的CDN系统中两个节点交换数据时,这些节点能够在应用数据分组中捎带探测信息。 For example, CDN caching system in which the movement of data back and forth to each other in the two nodes exchange data, these nodes can be piggybacked probe information in the application data packets. 当这种方法要求在探测业务方面的更多开销时,该方法可以用于估计端到端可用容量。 When this method requires more overhead in terms of the detection business, this method can be used to estimate end-to available capacity. 例如,试通程序和RTT无需再用应用数据。 For example, try to pass the program and no need to re RTT application data.

[0029] 填充状态矩阵的第三示例是基于被动监视,而非主动监视。 Third Example [0029] filling state matrix is based on passive surveillance, rather than proactive monitoring. 此方法使用路由器、缓存和其它节点中的业务的被动观察以便确定负载、丢失和相似参数。 This method uses routers passive observation, caching, and other nodes in the business in order to determine the load loss and similar parameters. 此方法需要对中间网络节点的访问权以及用于检索数据的标准化协议(例如,简单网络管理协议(SNMP))。 This method requires the intermediate network nodes and access a standardized protocol for retrieving data (e.g., Simple Network Management Protocol (SNMP)). 因此,这些网络单元需要驻留在运营商自己的域内或运营商需要对所涉及的网络单元的访问权。 Therefore, these network elements reside on the operator's own domain or operators need access to the network elements involved.

[0030] 所有这三个示例以及关联的方法均可以用于填充状态矩阵。 [0030] All three examples and associated method can be used to fill the state matrix. 根据图6中所示的一示范实施例,状态矩阵60使用(i)自动测量方法62来接收第一组参数,例如RTT、抖动、延迟、丢失、路径可用带宽等(ii)应用数据测量探测方法64以用于确定第二组参数,例如,RTT、抖动、延迟、丢失、路径可用带宽等以及(iii)被动监视方法66以用于确定第三组参数,例如,链路负载、丢弃分组等。 According to an exemplary embodiment shown in FIG. 6 embodiment, the state matrix 60 using (i) 62 to receive automatic measurement method first set of parameters, for example RTT, jitter, delay, loss, available bandwidth, etc. (ii) application of the measurement data detection The method 64 for determining a second set of parameters, e.g., RTT, jitter, delay, loss, and a path available bandwidth (iii) passive monitoring method 66 for determining a third set of parameters, such as link load, dropping packets and so on.

[0031] 根据一示范实施例,可以使用方法62、64和66中的仅一种方法或其组合来填充状态矩阵60。 [0031] According to an exemplary embodiment, a method 62, 64 and 66 in only one or a combination of methods to fill the state matrix 60. 但是,因为来自这些方法的测量可能不是同时进行的,即这些测量可能具有宽时间段内的不同时间戳,所以执行测量的时间和使用测量的时间能够在时间窗口内从数秒改变到数分钟或数小时。 However, since the measurements from these methods may not be performed simultaneously, i.e. it may have a broad measurement period different timestamp, so perform the measurement time and measurement time can be used within the time window change from seconds to minutes or hours. 还可能的情况是对于特殊路径未实行测量。 Also possible is a special path for failure to implement the measure. 因此,根据一示范实施例,引入可靠性函数以提供在状态矩阵中考虑测量的时间相关性的方式。 Thus, according to an exemplary embodiment, it is introduced to provide a measure of the reliability of the function of time-dependent manner in consideration of the state matrix. 换言之,引入可信度函数,其配置成滤出基于时间的相关性。 In other words, the credibility of the introduction of the function, which is configured to filter out the time-based correlation.

[0032] 此函数可以基于估计的方差,并且确保性能参数估计的关联性随着时间降低。 [0032] This function can be estimated based on the variance, and to ensure the relevance of performance parameter estimation decreases over time. 降低的速率取决于感兴趣的性能参数。 Reduced interest rate depends on the performance parameters. 例如,RTT估计在时间上比可用容量的估计更稳定。 For example, RTT estimate is more stable over time than estimated available capacity. 可信度可以由函数f(P,T0, T)表示,其中T是当前时间,Ttl是超时时间,以及p是特定性能参数类型,例如RTT或可用带宽。 The reliability can be represented by the function f (P, T0, T), where T is the current time, Ttl a timeout, and the p-type specific performance parameters, such as available bandwidth or RTT. 函数f的特定定义可以取决于应用而改变,正如接下来将论述的。 Specific definition of the function f may vary depending on the application, as will be discussed next.

[0033] 图7中更详细地示出状态矩阵60的元素Xijt5每个元素可以包括给定参数的估计/测量值70、与值70关联且指示测量或估计该值的时间的时间戳72、与估计值/测量值70关联的超时时间To 74以及估计/测量值70的方差76。 [0033] Figure 7 shows in more detail the state of the matrix element Xijt5 60 to each element may include estimation / measurement values 70, 70 associated with the value measured or estimated and the time stamp indicates the time value of the given parameter 72, and estimated values / measurements associated timeout 70 To 74 and estimation / measurement variance 70 76. 该方差能够是用于指定图12中的可信度函数的附加参数。 The variance can be used to specify the reliability function in Figure 12 additional parameters. 如果方差高,则静态成本线上方或下方的曲线更快速地向此线收敛。 If the variance is high, the cost side or the static line under the curve more rapidly converge to this line. 例如,高方差可以由静态线上方的曲线表示,而低方差能够由静态线下方的曲线来举例说明。 For example, high variance can be represented by a static line side of the curve, while the low variance of the static line below the curve to illustrate.

[0034] 在论述如何将成本函数应用于测量的参数以及该新的智能网络如何能够确定用户所要求内容的最佳缓存之前,论述再一个组成。 [0034] In the discussion on how to measure the parameters of the cost function is applied and how the new intelligent network before the user can determine the optimum cache content requires discusses a further composition. 这涉及上文引入的静态成本。 This involves a static cost of introducing the above. 为了确定网络节点之间的链路的静态成本,可以使用分级拓扑逼近技术。 To determine the cost of static links between network nodes, you can use the hierarchical topology approximation technique. 分级拓扑逼近技术将图I中所示的真实网络转换成图8中所示的在任何边缘节点对之间具有相同路径成本的分级网络。 Classification analysis based on the network topology approximation technique as shown in Figure I is converted into shown in Figure 8 has the same hierarchical network path cost between any of the edge node. 分级网络的生成在本领域中是已知的,并且可以用两个步骤来实现。 Hierarchical network generated are known in the art, and can be implemented in two steps.

[0035] 第一歩骤是创建分级虚拟拓扑,以及第ニ步骤是以将真实网络与虚拟网络之间的路径成本上的差减到最小的方式为虚拟拓扑中的每个链路生成成本。 [0035] The first step is to create a hierarchical ho virtual topology, and the first step is the difference ni cost path between a real network and virtual networks on ways to minimize the virtual topology generation cost for each link. 因为图8的虚拟网络是分级的,所以能够将其建模成类似树形。 Because the virtual network of Figure 8 is hierarchical, it is possible to be modeled to resemble a tree. 另外,虚拟网络中任何边缘节点对之间正好存在ー个路径。 In addition, there just ー paths between the virtual network node for any edge. 这两个特征使得使用较不复杂的算法来计算边缘节点之间的静态成本成为可倉^:。 Both of these features makes the use of less complex algorithms to calculate the static costs between the edge node become available warehouse ^ :.

[0036] 在图8中所示的虚拟网络80中,边缘节点14保持与图I的真实网络10中相同,而中间节点82不具有与真实网络10的中间节点12的直接对应性。 [0036] In the virtual network shown in FIG. 8 (80), holding the edge node 14 of FIG. I in the same real network 10, and the intermediate node 82 does not have a real network 10 directly to intermediate node 12 in correspondence. 创建虚拟网络80的特殊方式取决于真实网络、应用和可用技术,如本领域技术人员会认识到的。 A special way to create a virtual network 80 depends on the real network, applications and available technologies, such as skilled in the art will recognize that the. 但是,真实网络由路由器孤岛组成,且ー个孤岛中的路由器较其它孤岛中的路由器更加彼此靠近。 However, the real network of islands composed of routers and router ー a silos closer to each other than the other islands in the router. 这类孤岛可以对应于分级拓扑中的中间节点。 Such silos can be graded intermediate node topology corresponds. 更好地理解虚拟网络的比拟是想象真实社区和街道/道路,以及ー个级别上的虚拟节点可以对应于真实城市,下一更高级别上的虚拟节点可以对应于真实地区,并且再下一更高级别上的虚拟节点可以对应于国家的虚拟模型。 To better understand the analogy is to imagine true virtual network of community and street / road, and on ー virtual node level may correspond to a real city, a virtual node on the next higher level may correspond to the real area and then the next virtual nodes can be on a higher level virtual model corresponding to the country. 生成分级网络的此第一歩骤可以由网络操作员手动实现,或通过參照包括所需信息的数据库的计算程序来实现。 This first generation of hierarchical network ho step can be implemented by a network operator manually, or by referring to the program database includes calculating the required information to achieve.

[0037] 在第二步骤中,可以使用优化程序来以将分级与真实网络之间的边缘到边缘路径成本上的差最小化的方式对虚拟网络中的链路指派成本。 [0037] In a second step, you can use the program to optimize the edge of the network to the poor grading and real way that minimizes costs on the edge of the path cost assigned to the virtual network link. 所使用的特殊优化方法取决于真实网络和与运营商有关的其它考虑,正如本领域中已知的各种技木。 Special optimization method used depends on real network with carriers and other considerations related, as known in the art of various techniques of wood. ー种方式是通过梯度搜索(gradient search)查找误差函数的最小值。ー way is to find the minimum error function by gradient search (gradient search). 该误差函数能够定义为所有路径,即所有可能的边缘节点对之间的所有路径的路径成本上的差的平方和。 The error function can be defined as all paths, namely the squared difference of path costs for all paths of all possible edge between pairs of nodes and on. 一旦定义了分级树,应用哪个特殊的树算法就再次是基于问题的更细分细节进行的挑选和审慎选择的问题。 Once you have defined a hierarchical tree, which is a special tree algorithm once again is to pick and choose the issues more carefully details segments based undertaken.

[0038] 系统的分级树表示使得CDN的运营商/提供商能够指派系统的节点之间的静态成本。 [0038] hierarchical tree representation system enables CDN operators / providers to assign a static cost between nodes of the system. 这些静态成本/权重被⑶N用在决定提取对象/资产的最佳缓存中。 These static cost / weight is used in the optimal cache decision ⑶N extract object / asset's. 这些静态成本是数值,并且可以基于底层链路的一些特征在逻辑上推导。 These costs are static values, and may be based on some of the characteristics of the underlying link logically deduced. 例如,称为开放最短路径优先(OSPF)的技术基于链路的带宽来确定静态成本。 For example, the technique known as Open Shortest Path First (OSPF) link bandwidth is determined based on the static cost. 通过将參考带宽(缺省为100 Mbps)除以接ロ带宽来计算OSPF中的链路的静态成本。 By reference bandwidth (default is 100 Mbps) of bandwidth calculated by dividing the then ro Static OSPF cost of the link. 因此,至目的地的总静态成本是至目的地的路径中的个体链路的静态成本之和。 Thus, the total cost to a destination static static cost of the route to a destination and the individual links.

[0039] 因为静态成本是逻辑值,所以通过有意向地修改链路的静态成本而不考虑链路的底层特征,运营商可以影响使用该链路的方式。 [0039] Because static costs is a logical value, so there is the intention to modify the static through the link regardless of the underlying cost characteristics of the link, the operator can influence the way to use the link. 此能力为运营商运作网络提供更多灵活性。 This capability provides greater flexibility for carriers operating the network. 因此,能够将基于底层物理链路的固定特征进行静态成本的初始指派视为指派的成本。 Therefore, the ability to link the underlying physical characteristics based on a fixed initial assignment of the static costs as a cost assignment. 如果静态成本是基于运营商内部策略,则可以将此静态成本视为基于策略的成本。 If the static internal costs are based on operator policy, you can use this static costs deemed strategy based on cost. 基于策略的静态成本也是指派的静态成本。 Static cost strategies based on static cost is assigned. 但是,静态成本未将网络可能受到的实况修改纳入考虑,并且因此在决定使用哪个缓存来响应用户的请求时不是精确的參数。 However, the cost is not a static network may be modified live into account, and thus to determine which cached response is not accurate when the user requests a parameter.

[0040] 根据ー示范实施例,引入对例如由于业务导致的网络的改变敏感的动态成本。 [0040] According to an exemplary embodiment ー introduced, for example, due to changes in the network traffic caused by dynamic cost sensitive. 可以将动态成本与静态成本组合,从而得到上文提到的新成本函数。 Dynamic cost and can be a combination of static cost, thereby obtaining a new cost function mentioned above. 可以将总成本定义为取决于静态成本和动态成本。 The total cost can be defined as the cost depends on the static and dynamic costs. 可以通过引入上文论述的可信度函数f来使总成本C为时间相关的。 Discussed above by introducing the reliability function f to make the total cost C is time-dependent. 在ー个特殊实施例中,由(i)静态成本以及(ii)动态成本与可信度函数之间的积的和来给出时间相关的总成本C(t)。 In a particular embodiment ー by (i) a static cost and (ii) the cost and reliability of dynamic function between the product and to give time-dependent total cost C (t). 接下来更详细地论述这些量。 Next, these amounts discussed in more detail. [0041] 正如已经论述的,状态矩阵可以存储与关联链路的成本不同的一个或多个测量。 [0041] As already discussed, the matrix may be one or more of the state of measuring the cost of storage associated with different links. 因此,可以使用归一化函数将状态矩阵的元素Xij转换成动态成本。 Therefore, you can use normalized function will state matrix elements Xij is converted into dynamic cost. 换言之,归一化函数将测量的性能参数映射到可以添加到静态成本的动态成本。 In other words, the normalized performance parameters are mapped to the function of the measured dynamic cost can be added to the cost of the static. 例如,图9示出链路的可用容量及其关联的动态成本。 For example, Figure 9 shows a dynamic link available capacity and its associated costs. 通过主动测量方法(例如BART)来测量可用容量。 By actively measuring methods (such as BART) to measure the available capacity. 生成第二列中的动态成本的归一化函数可以具有各种形式,其取决于应用。 Generating a second column dynamic cost of normalization function may have various forms, depending on the application. 图9中所示的动态成本对应于单个测量参数。 As shown in FIG. 9 dynamic cost corresponds to a single measurement parameters. 如果使用多个参数,则可以确定多个动态成本。 If you use multiple parameters, you can define a plurality of dynamic cost. 当链路的性能参数变化时,动态成本变化,并且出于此原因,动态成本更好地反映真实网络的成本。 When the performance parameters of the link, the dynamic changes in cost, and for this reason, dynamic cost better reflect the true cost of the network. 图10在第一列中示出实际容量以及在第二列中示出与此链路容量关联/指派的静态定价。 Figure 10 shows the actual capacity and the second column shows the capacity associated with the link / assignment static pricing in the first column. 可以使用其它归一化函数。 You can use other normalization function. [0042] 基于静态成本和动态成本,可以计算两个成本之差A,如图11中第5列中所示。 [0042] costs based on static and dynamic costs, can calculate the cost difference between the two A, as shown in FIG. 11 in the first five. 更确切地来说,对于相同的链路,将差A计算为从固定网络拓扑特征获得的链路容量与从主动测量获得的实际带宽之间的成本上的差。 More specifically, for a same link, the link capacity is calculated as the difference between A obtained from the fixed network topological features from the cost difference between the actual bandwidth active measurement obtained between. 成本差△可以是负的或正的,如图11中所示。 Cost difference △ can be positive or negative, as shown in FIG. 11. 因此,成本差A取决于静态(指派的成本)与动态成本二者。 Therefore, the cost depends on the difference between A static (assigned costs) and the cost of both dynamic.

[0043] 该测量系统和状态矩阵提供在操作中时网络的实时性能参数。 [0043] The measuring system and the state matrix provided in real-time operation of the network performance parameters. 基于上文论述的归一化函数,成本差△实时地标识节点之间的实际逻辑成本。 Based discussed above normalization functions, real-time cost difference △ logic identifies actual cost between the nodes. 这使得CDN能够基于网络的实际状态提供更好的响应。 This makes it possible to provide better response CDN based on the actual state of the network.

[0044] 在理想状况下,无论何时CDN逻辑需要节点之间的更新读取,测量系统就能够即时地提供该更新读取。 [0044] In an ideal situation, whenever between CDN node update logic needs to read the measurement system can provide real-time updates to the read. 但是,在真实世界中,测量花费时间,并且CDN系统可能不等待直到最新结果可用。 However, in the real world, the measurement takes time, and CDN system may not wait until the latest results available. 因此,CDN逻辑需要依赖于上次或最近进行的测量。 Therefore, CDN logic need to rely on recent or last measurement. 目前时间与进行测量的时间之间的时间差可能改变,从数秒到很长时间或完全无时间差。 The current time and the time difference between the measurement time may change from a few seconds to a long or completely without a time difference.

[0045] 因此,根据一示范实施例,引入实现差A的可信度或可靠性的函数,以便提供在CDN中使用时将测量的信息的时间相关性纳入考虑的方式。 [0045] Thus, according to an exemplary embodiment, it is introduced to achieve a function of confidence or reliability difference A in order to provide information on the measured time when used in the correlation CDN incorporated consideration. 上文已论述这种函数f。 The foregoing has discussed this function f. 此函数f提供使得实际逻辑成本为时间相关的方式。 This provides the function f so that the actual cost of the logic of time-dependent manner. 现在更详细地论述此概念。 This concept is now discussed in more detail.

[0046] 此函数使用来自网络拓扑的指派成本(6>)作为基线,如图12中所示。 [0046] This function is used to assign the cost from the network topology (6>) as the baseline, as shown in FIG. 如果没有测量信息可用,则基线用作逻辑固定成本。 If no measurement information is available, the baseline is used as a logical fixed costs. Ti是当前时间(即时时间)以及Ttl (超时时间)是自此后测量值太旧而无法使用的时间。 Ti is the current time (real-time) and the Ttl (time-out) is a measure of time since then too old to use. 图12提供可信度函数f(T,Ttl)的示例。 Figure 12 provides the credibility function f (T, Ttl) examples. 按运营商认为适合的以及正如本领域技术人员会认识到的,可以使用其它函数。 Press operators may think fit and as skilled in the art will recognize, you can use the other functions. 可信度函数的形状可以取决于估计方差或取决于所使用的特定性能参数的固有特性。 Confidence function may depend on the shape of the inherent characteristics of the specific performance parameters or estimation variance depends on the use.

[0047] 可信度函数f随时间推移使差A的值减少。 [0047] the credibility of the function f over time so that the difference between the value of A is reduced. 如果上次测量刚好在当前时间Ti之前进行,则计算的差△能够从0改变至无限大,如图12中所示。 If the last measured just before the current time of Ti, is calculated from the difference △ 0 can be changed to infinity, as shown in FIG. 随着测量时间与目前时间之间的间隙变得越来越大,由于测量的可信度减少,所以△的上和下可能值减小。 With the measurement time and the gap between the time the current becomes larger, due to the reduction in confidence in measurements, so the upper and lower △ possible value decreases. 在时间Ttl之后,A不再适用,并且总成本与指派成本Cf相同。 After the time Ttl, A is no longer applicable, and the total cost and the cost of Cf same assignment. A对总成本的影响如何改变的说明仅是一个示例,并且实际实现取决于网络。 A total cost of the impact of how to change the description is only an example, and the actual implementation depends on the network. 总成本的另一个示例可以包括(i)静态/指派成本乘以(I - f)与(ii)动态成本乘以f之和,其中可信度函数f在0与I之间改变。 Another example of the total cost may include (i) Static / Cost assigned multiplied by (I - f) and (ii) the sum of dynamic cost multiplied by f, where f is the function reliability between 0 and I change. 在此情况中,如果f = 0,德尔塔(delta)不可信,则总成本等于指派成本,而如果f = I,德尔塔充分可信,则总成本等于动态成本。 In this case, if f = 0, delta (delta) can not be trusted, then the total cost is equal to the cost assignment, if f = I, Delta fully trusted, then the total cost is equal to the dynamic cost.

[0048] 因此,真实网络10的两个边缘节点i和j之间的总成本Cij 一般可以基于上文定义的个体量来表达,并且具有如下表达式:其中A是指派成本,△是通过归ー化函数从状态矩阵获得的动态成本,以及f是可信度函数。 [0048] Thus, the total cost Cij real network edge node 10 both i and j are generally based on the individual to express the amount as defined above, and has the following expression: wherein A is assigned costs, △ owned byー Dynamic cost function obtained from the state matrix, and f is the credibility of the function. 运算符(g)指示可以使用的各种数学运算。 Operator (g) indicates the various mathematical calculations can be used. 上文已论述Cij的特定示例。 Cij has been discussed above, a specific example of. 但是,正如本领域技术人员会认识到的,可以使用其它函数。 However, as those skilled in the art will recognize, you can use the other functions.

[0049] 根据ー示范实施例,关于图13论述用于选择资源的过程。 [0049] According ー exemplary embodiment discussed with respect to FIG. 13 for selecting a resource. 在图13中,用户130在步骤150中向⑶N 132请求资产XtXDN 132可以包括重定向引擎134、成本估计器单元136、虚拟拓扑单元138、状态矩阵単元140等。 In Figure 13, the user 130 to request assets XtXDN 132 ⑶N 132 in step 150 may include redirection engine 134, the cost estimation unit 136, a virtual topology unit 138, radiolabeling state matrix element 140 and the like. 可以用软件、硬件或其组合来实现这些单元。 It may be software, hardware or a combination thereof to achieve these units. 在一个应用中,可以在单个节点中的处理器中实现所有这些单元。 In one application, you can achieve all these elements in a single node processor. 在另ー个应用中,这些单元可以分布在CDN的各种节点,各种处理器中。 In another ー one application, these units can be distributed in the various nodes of the CDN, a variety of processors.

[0050] 重定向引擎134在步骤152中确定⑶N 132中是否存在资产X,并且确定是否多个缓存中存在该资产。 [0050] redirection engine 134 determines whether there are assets X ⑶N 132 in step 152, and determines whether the asset exists in more than one cache. 在步骤154中,重定向引擎134查询成本估计器136对于用户130哪个是最佳缓存。 In step 154, the re-estimated cost of the query engine 134 136 130 for the user which is the best cache. 成本估计器136在步骤156中构建至具有资产X的缓存的路径,并将此信息提供到虚拟拓扑单元138。 136 Construction costs are estimated to have assets X path cache in step 156, and provide this information to the virtual topology unit 138. 虚拟拓扑单元138将真实网络转换成分级虚拟网络,正如上文论述的。 Virtual topology unit 138 converts the real network component-level virtual network, as discussed above. 注意,此步骤是可选的,并且CDN网络可以无需虚拟拓扑单元138而执行最佳缓存的评估。 Note that this step is optional, and can CDN network topology without virtual cache unit 138 performs the best evaluation. 在步骤160中将来自虚拟拓扑单元138的信息传递到状态矩阵单元140,状态矩阵单元140还从网络的测量系统接收測量数据。 Transmitting information from the virtual topology unit 138 in step 160 to the state in the matrix unit 140, state matrix measuring unit 140 also receives data from the network measurement system. 基于测量数据和虚拟网络,状态矩阵単元140在步骤162中更新矩阵值,并在步骤164中将此数据提供到成本估计器单元136。 Based on the measured data and virtual networks, state matrix radiolabeling unit 140 in step 162 update the matrix values and provide the cost estimate unit 136 in step 164 in this data. 成本估计器单元136还在步骤166中接收虚拟网络数据,并在步骤168中计算来自用户130连接到的边缘节点和具有资产X的边缘节点的每个路径的总成本。 Cost estimator unit 136 in step 166 also receives a virtual network data, and calculate the total cost of each path from the user edge node 130 is connected to the edge node and having the property X in step 168.

[0051] 基于总成本对具有资产X的边缘节点评级,并在步骤170中将最佳缓存信息(例如,具有最低成本的缓存)提供到重定向引擎134。 [0051] Based on the total cost of the edge node X rating with the asset, and the step 170 will be in the best cache information (e.g., buffer having the lowest cost) to the redirection engine 134. 基于此数据,在步骤172中,重定向引擎134响应用户130的查询向该用户提供最佳缓存的地址。 Based on this data, in step 172, redirection engine 134 in response to a user's query to the user 130 provides the best cache address. 用户130然后从最佳边缘节点提取资产X或由CDN 132命令用户130连接到的边缘节点从最佳边缘节点提取资产X并将该资产X提供到用户130。 User 130 is then extracted from the best assets X edge node or by the CDN 132 instructs the user 130 is connected to the edge node extracted from the best edge node assets and the assets X X 130 is supplied to the user.

[0052] 根据图14中所示的一示范实施例,可以在⑶N 132中实现决策引擎142以便基于 [0052] According to an exemplary embodiment shown in FIG. 14 embodiment, the decision engine may be implemented in ⑶N 132 to 142 based on the

(i)动态成本144和(ii)静态(指派)成本146来决定内容提取和各种缓存之间的资产迁移。 (I) dynamic cost 144 and (ii) static (assignment) costs 146 to determine the content extraction and migration of assets between the various caches.

[0053] 根据图15中所示的一示范实施例,有ー种方法用于选择在通信网络中连接的多个缓存可用的资源。 [0053] FIG. 15 according to an exemplary embodiment shown, there ー methods for selecting a plurality of buffers in a communications network connecting the available resources. 该方法包括从用户接收对资源的请求的步骤1500、标识多个缓存中存储该资源的一个或多个缓存的步骤1502、确定用户与存储该资源的一个或多个缓存中的每个缓存之间的路径所关联的成本的步骤1504(其中该成本包括反映通信网络的拓扑的静态成本和指示通信网络的变化状况的动态成本)、以及基于至ー个缓存的最低成本从存储该资源的一个或多个缓存选择该一个缓存的步骤1506。 The method includes the step of receiving a request for resources from the user's 1500, identifying a plurality of cache memory cache of one or more steps of the resource 1502, to determine the user with one or more of the cache memory in the resource of each cache associated path cost between a step 1504 (which includes the cost of the dynamic changes in the cost of the static condition and instructs the communication cost network topology of a communication network to reflect), and based on a cache to ー storing the lowest cost from a resource or multiple cache selection steps that a cache of 1506.

[0054] 较之如仅使用指派成本的传统方法的情况,一个或多个论述的示范实施例可以为内容迁移和检索提供精确且动态的成本估计。 [0054] As compared with the conventional method using only the assigned cost, one or more exemplary embodiments discussed can provide accurate estimates of the cost for the content and dynamically migrate and retrieval. 再有,公开ー种确定网络的测量信息的可信度的新方法、一种用于通过将基于例如时间的可能改变纳入考虑来表示和汇总分布式系统中的网络测量数据的新方法。 Furthermore, a new method to determine the credibility of the measurement information of the public network ー species, for example, by the time-based change may be taken into account and represent a new method in a distributed system network measurement data summary. 该新方法提供网络状况的动态画面以便在更新的网络状态信息对于整体性能有益的如CDN的数据交换网络中使用。 The new method provides network status of a motion picture to be used in updating the network status information as beneficial for the overall performance of the CDN data exchange network. 一个或多个实施例能够用作用于将动态网络状况纳入考虑来实行运营商策略的输入,并且可以使得如BART的測量工具能够被利用来为改善分布式系统中的性能提供建议。 One or more embodiments can be used as dynamic network conditions into account to carry out the input operator policy, and can be made as BART measurement tools can be utilized to provide recommendations for improving performance of a distributed system.

[0055] 图16中示出根据示范实施例的能够实行操作的代表性结构的示例。 [0055] FIG. 16 illustrates an example of the operation of the representative structure can be implemented in accordance with an exemplary embodiment. 可以使用硬件、固件、软件或其组合来执行本文描述的各种步骤和操作。 You can be implemented using hardware, firmware, software, or a combination thereof as described herein to perform the various steps and operations. 适于执行示范实施例中描述的活动的示范装置1600可以包括服务器1601,服务器1601可以包括图13中所示的単元134、136,138和140中的ー个或多个单元。 Adapted to perform exemplary embodiments described activity demonstration unit 1600 may include a server 1601, server 1601 may include radiolabeling element shown in Figure 13 in the 134, 136 and 140 ー or more units. 这种服务器1601可以包括耦合到随机存取存储器(RAM) 1604和只读存储器(ROM) 1606的中央处理器(CPU) 1602。 Such a server 1601 may include a random access memory coupled to the (RAM) 1604 and a read only memory (ROM) 1606 in the central processor (CPU) 1602. ROM 1606还可以是用于存储程序的其它类型的存储媒体,如可编程ROM (PR0 M)、可擦PROM (EPROM)等。 ROM 1606 may also be other types of storage medium for storing a program, such as programmable ROM (PR0 M), erasable PROM (EPROM) and the like. 处理器1602可以通过输入/输出(I/O)电路系统1608和总线连接(bussing) 1610与其它内部和外部组件通信,以提供控制信号以及诸如此类。 Processor 1602 via the input / output (I / O) circuitry 1608 and bus connections (bussing) 1610 other internal and external components to communicate with in order to provide control signals and the like. 处理器1602按软件和/或固件指令所指示的,实行如本领域中已知的各种各样的功能。 Processor 1602 according to the software and / or firmware instructions indicated, as known in the art to implement various functions.

[0056] 服务器1601还可以包括ー个或多个数据存储装置,包括硬盘驱动器和软盘驱动器1612、⑶-ROM驱动器1614和能够读取和/或存储如DVD等的信息的其它硬件。 [0056] ー Server 1601 may also include one or more data storage devices, including hard drives and floppy disk drives 1612, ⑶-ROM drive 1614 and the ability to read and / or stored, such as DVD and other additional hardware information. 在ー个实施例中,可以将用于实行上文论述的步骤的软件存储和分布在CD-ROM 1616、软磁盘1618或能够便携地存储信息的其它形式的媒体上。 In ー embodiment, can be used to store the software implementation of the steps discussed above and distributed on CD-ROM 1616, diskette 1618 or other form capable of portably storing information medium. 这些存储媒体可以插入到如⑶-ROM驱动器1614、盘驱动器1612等的装置中并被其读取。 These storage media may be inserted into such ⑶-ROM drive 1614, the disk drive apparatus and the like 1612 that reads. 服务器1601可以耦合到显示器1620,显示器1620可以是任何类型的已知显示器或呈示屏幕,如IXD显示器、等离子显示器、阴极射线管(CRT)等。 Server 1601 may be coupled to a display 1620, display 1620 may be any type of known display or rendering the screen, such as IXD display, a plasma display, a cathode ray tube (CRT) and the like. 提供一种用户输入接ロ1622,其包括如鼠标、键盘、麦克风(microphone)、触控板、触摸屏、语音识别系统等的一个或多个用户接ロ机制。 Ro provide a user input interface 1622, which include, for example a mouse, a keyboard, a microphone (microphone), touch pad, touch screen, voice recognition system, one or more user interface mechanisms ro.

[0057] 服务器1601可以经由网络耦合到其它计算装置,如陆上通信线路(Iandline)和/或无线终端和关联的观看器(watcher)应用。 [0057] Server 1601 may be coupled via a network to other computing devices, such as land-based communication line (Iandline) and / or the viewer and the associated wireless terminal (watcher) applications. 该服务器可以是如全球区域网络(GAN)(例如因特网1628)中的更大网络配置的一部分,其允许最终连接到各种陆上通信线路和/或 The server can be a global area network, such as (GAN) (e.g., the Internet 1628) a portion of a larger network configuration, which allows the final connection to the various landline and / or

移动客户端/观看器装置。 Mobile Client / viewer devices.

[0058] 公开的示范实施例提供用于确定资源的节点、方法和计算机程序产品。 [0058] The disclosed exemplary embodiments provide for determining node resources, methods and computer program products. 应该理解本文描述不打算限制本发明。 It should be understood described herein are not intended to limit the invention. 相反,示范实施例打算涵盖在所附权利要求书定义的本发明的精神和范围中包括的备选、修改和等效。 In contrast, the exemplary embodiment is intended to cover in the spirit and scope of the appended claims the invention as defined in the include alternatives, modifications and equivalents. 另外,在示范实施例的详细描述中,阐述了许多特定细节以便提供对要求保护的本发明的全面理解。 In addition, a detailed description of exemplary embodiments, numerous specific details are set forth in order to provide the claimed thorough understanding of the present invention. 但是,本领域技术人员会理解,没有这类特定细节仍可以实施各种实施例。 However, those skilled in the art will appreciate that there is no such specific details can still be implemented with various embodiments.

[0059] 正如本领域技术人员还将领会的,可以在计算机程序产品中或作为方法在电信网络、无线通信装置中实施示范实施例。 [0059] As those skilled in the art will appreciate, the product may be a computer program or as a method in a telecommunication network, the wireless communication apparatus embodiment exemplary embodiment. 相应地,示范实施例可以采用完全硬件实施例的形式或组合硬件和软件方面的实施例的形式。 Accordingly, the exemplary embodiment can be used in the form of an entirely hardware embodiment or a combination of hardware and software aspects of embodiments. 另外,示范实施例可以采取计算机可读存储介质上存储的计算机程序产品的形式,该计算机可读存储介质具有实施在该介质中的计算机可读指令。 In addition, the exemplary embodiments may take the form of a computer-readable storage medium storing a computer program product, the computer readable storage medium having computer-implemented in the medium-readable instructions. 可以利用任何适合的计算机可读介质,包括硬盘、⑶-ROM、数字多用光盘(DVD)、光存储装置或如软盘或磁带的磁存储装置。 Using any suitable computer-readable media, including hard disks, ⑶-ROM, digital versatile disc (DVD), optical storage devices, or magnetic storage device such as a floppy disk or magnetic tape. 计算机可读媒体的其它非限制性示例包括闪存型存储器或其它已知的存储器。 Other non-limiting examples of computer readable media include flash-type memories or other known memories.

[0060] 虽然在实施例中以特殊组合描述了本发明示范实施例的特性和単元,但是每个特性或単元能够在没有实施例的其它特性和単元的情况下单独地使用或以具有或不具有本文公开的其它特性和単元的各种组合来使用。 [0060] Although in the embodiment a special combination described exemplary embodiment of the invention features and radiolabeling yuan, but each feature or radiolabeling element can be used alone or to have or not in the case of other features not embodiments and radiolabeling yuan It has other features and various combinations disclosed herein radiolabeling dollars to use. 本申请中提供的方法或流程图可以在计算机可读存储介质中有形地实施的计算机程序、软件或固件中实现以由专门编程的计算机或处理器来运行。 Methods or flow charts provided in this application may be stored in a computer readable medium tangibly implemented in computer programs, software or firmware to a computer or processor implemented by a specially programmed to run.

Classifications
International ClassificationH04L29/08
Cooperative ClassificationH04L67/101, H04L67/1097, H04L67/1025, H04L67/2852, H04L67/1002, H04L45/121, H04L45/00, H04L45/125, H04L45/124
European ClassificationH04L29/08N9A, H04L29/08N9A1C, H04L45/125, H04L45/121, H04L45/124, H04L45/00
Legal Events
DateCodeEventDescription
15 Aug 2012C06Publication
26 Dec 2012C10Entry into substantive examination
14 Sep 2016C14Grant of patent or utility model