(19)
(11)EP 3 044 933 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
09.01.2019 Bulletin 2019/02

(21)Application number: 13774618.6

(22)Date of filing:  12.09.2013
(51)Int. Cl.: 
H04L 12/24  (2006.01)
H04L 29/06  (2006.01)
H04L 12/841  (2013.01)
(86)International application number:
PCT/EP2013/068919
(87)International publication number:
WO 2015/036023 (19.03.2015 Gazette  2015/11)

(54)

A METHOD FOR OPERATING AN INFORMATION-CENTRIC NETWORK AND NETWORK

VERFAHREN ZUR INFORMATIONSNETZWERKBENUTZUNG UND NETZWERK

METHODE POUR L' OPERATION D'UN RESEAU D'INFORMATIONS ET RESEAU


(84)Designated Contracting States:
AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

(43)Date of publication of application:
20.07.2016 Bulletin 2016/29

(73)Proprietor: NEC Corporation
Tokyo 108-8001 (JP)

(72)Inventors:
  • SCHNEIDER, Fabian
    69115 Heidelberg (DE)
  • KUTSCHER, Dirk
    69121 Heidelberg (DE)

(74)Representative: Patent- und Rechtsanwälte Ullrich & Naumann 
PartG mbB Schneidmühlstrasse 21
69115 Heidelberg
69115 Heidelberg (DE)


(56)References cited: : 
US-A1- 2012 131 144
US-B1- 8 516 146
US-B1- 7 809 848
  
      
    Note: Within nine months from the publication of the mention of the grant of the European patent, any person may give notice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed in a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).


    Description


    [0001] The present invention relates to a method for operating an information-centric network, ICN, wherein at least one named data object, NDO, is addressable and wherein after an initial request a request aggregation of subsequent requests for the at least one NDO is performed according to a definable rule.

    [0002] Further, the present invention relates to an information-centric network, ICN, comprising at least one addressable named data object, NDO, and means for performing after an initial request a request aggregation of subsequent requests for the at least one NDO according to a definable rule.

    [0003] In ICN, communication is not based on packets that are "sent from" and "destined to" hosts or host interface addresses.

    [0004] In ICN, requestors send requests to the network, asking for named data objects, NDOs, that have been published before and that are available in one or many copies in the network. The network elements, e.g. switches or nodes, that receive a request - unless they have a local copy in their cache - typically have to decide where to forward the request to, for example which interface to use for forwarding the request. Typically a network switch or network node can participate in a routing protocol that helps to distribute this information, or a network switch or network node can employ a Name Resolution Service, NRS, that can map NDOs to locators in underlying networks, for example IP.

    [0005] Once a request has been forwarded and reached the destination, e.g., a NDO copy in a cache, a corresponding response message, i.e., the NDO or a locator, has to be relayed back to the requestor, possibly passing one or more on-path caches that can cache the objects in order to satisfy future requests from their cache. This return path can be determined in different ways: For example network elements can maintain state or they can obtain some information from data structure inside messages, such as a label stack.

    [0006] A survey of information-centric networking is obtainable from Ahlgren, B.; Dannewitz, C.; Imbrenda, C.; Kutscher, D.; Ohlman, B., "A survey of information-centric networking," Communications Magazine, IEEE, vol.50, no.7, pp.26,36, July 2012, doi: 10.1109/MCOM.2012.6231276, URL: http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6231276

    [0007] The basic idea of SDN is to separate network control-plane logic from data-plane forwarding. This means breaking today's feature-rich network elements, e.g. switches, nodes or routers, into dumb forwarding elements and complex network controllers. Software-Defined Networking, SDN, relies on the concept of providing an API, Application Programming Interface/protocol for packet forwarding devices such as switches, nodes and routers, which allows programmability of network elements and entire networks. An example for such a controller to switch protocol is OpenFlow, see OpenFlow Switch Specification, Version 1.0.0 (Wire Protocol 0x01), December 31, 2009. http://www.openflow.org/documents/openflow-spec-v1.0.0.pdf.

    [0008] Information-centric networking approaches, such as CCN, Content Centric Networking, or Netlnf, Network of Information, follow the basic principle of addressing content objects, NDOs, and not hosts or end-points. One often mentioned advantage in ICN is "request aggregation". Request aggregation is the concept to avoid forwarding subsequent requests for the same NDO arriving at the same ICN node or switch, so that the request and the NDO are transmitted only once over the path between that node or switch and the content source. For reference, see also Fig. 1 which is showing a per-node request aggregation. Concretely, Fig. 1 is showing three different points in time, where requests R1, R2 and R3 are performed from different requestors. There is requested a content A which is provided within a source S. The content A shall be transmitted to hosts H1, H2 and H3 according to the indications by arrows within Fig. 1, right part. Nodes N1 to N6 are provided within the shown ICN structure. The left part of Fig. 1 is showing the request R1 for the content A. This request R1 is transmitted from N1 to N6 and the source S as shown within the left and middle parts of Fig. 1. Within this middle part of Fig. 1 a further request R3 is directed to N1 and is aggregated and delayed by node N1. A further request R2 directed to node N2 will be transmitted to N6 and the source S. As obtainable from the right part of Fig. 1 the content A is transmitted two times on two ways from S to N1 and from S to N2 for provision to hosts H1, H2 and H3.

    [0009] For request aggregation in ICN, a node or switch delays the forwarding of subsequent requests for the same NDO, as he expects the responses to first request to pass by and will then serve both or all requests. Per-node request aggregation is well known for ICN.

    [0010] Even though the known request aggregation mechanism is helpful for reducing the scope of transmitted ICN messages like request and response messages a further enhancement of the effectiveness of use of network resources seems to be useful in view of generally increasing traffic within data networks.

    [0011] Thus, it is an object of the present invention to improve and further develop a method for operating an information-centric network and an according information-centric network for allowing a very effective use of network resources by simple means.

    [0012] In accordance with the invention, the aforementioned object is accomplished by a method comprising the features of claim 1 and by a network comprising the features of claim 15.

    [0013] According to claim 1 the method is characterized in that a software-defined networking, SDN, with a SDN controller is implemented in the ICN and that a network-wide request aggregation is performed by the SDN controller.

    [0014] According to claim 15 the network is characterized in that a software-defined networking, SDN, with a SDN controller is implemented in the ICN and that the SDN controller is configured for performing a network-wide request aggregation. According to the invention it has been recognized that it is possible to allow a very effective use of network resources by further reducing transmitted ICN messages. Such a further reduction of transmitted ICN messages can be provided by an enhancement of the network control by the implementation of software-defined networking, SDN, with a SDN controller. Further according to the invention the SDN controller performs a network-wide request aggregation instead of the known per-node request aggregation within known information-centric networks. The invention is especially beneficial for live streaming of data to multiple receivers or hosts, as many requests for the same NDO arrive at the same time at different nodes or switches of the network. This invention provides a network-wide version of request aggregation made possible through ICN control.

    [0015] On the basis of the invention a very effective use of network resources is possible by simple means.

    [0016] Within a concrete embodiment a centralized SDN controller is used for providing request aggregation mechanisms.

    [0017] For providing a very effective request aggregation the SDN controller can detect duplicate requests.

    [0018] Depending on the definable rule the SDN controller can decide on which requests to delay. Such a decision can be performed centrally. Further depending on the definable rule the SDN controller can decide how to route and/or where to duplicate the at least one NDO. The NDO can be seen as a response to a respective request.

    [0019] With regard to a very reliable operation of the request aggregation mechanism the ICN can operate in a mode wherein a request path and a response path are provided at the same time upon an initial handling of the initial request. However, a similar but simpler approach is possible, wherein the ICN operates in a mode wherein a response path is only provided when a NDO response or a transmission of the NDO or a transmission of the content of the NDO is generated.

    [0020] Regarding a very effective use of the network resources a best content source for the at least one NDO can be determined. In this way a best content source is selected between multiple content sources all providing the same at least one NDO. Such a determination of a best content source can be performed under the knowledge where content of the NDO or the NDO itself is cached. Thus, the determination of a best content source can be cache-aware, i.e. can know where content is cached.

    [0021] Within this description the term "content source" or "source" is used for an ICN switch or node or router actually providing the desired NDO form its storage as well as for an ICN node or switch or router at the egress of the ICN, wherein said ICN node or switch or router knows how to retrieve the content from another ICN node or switch or router. The ICN elements "node" or "switch" or "router" are used herein synonymously.

    [0022] A concrete embodiment of a SDN controller for network-wide request aggregation can comprise a Pending-Request Table, PRT. Such a PRT can also be designated as pending interest table.

    [0023] Regarding a very reliable operation of the request aggregation method entries in the PRT should comprise different information. Thus, the PRT can comprise a name or identification of the requested NDO and/or a counter differentiating between requests for the same name or identification and/or an identifier of an ingress SDN switch/node and/or a selected content source for the NDO and/or a selected return path for the NDO and/or the estimated time of arrival of the requests at the content source.

    [0024] Further, the SDN controller can comprise the topology of the ICN with regard to an optimization of the routing of NDOs, for example.

    [0025] Within a further embodiment the SDN controller can comprise an expected per link one-way delay information. This will provide a very sophisticated routing and scheduling of requests and NDOs.

    [0026] Within a further embodiment the SDN controller can comprise a cost metric for all possible paths and/or content sources. Such a cost metric can serve as input to a spanning tree calculation and can be related to a link delay and/or a link bandwidth and/or a link utilization and/or a content source utilization. A relation to other traffic engineering metrics is also useful and/or possible.

    [0027] Generally, with regard to a reliable and effective request aggregation mechanism the SDN controller can comprise the definable rule.

    [0028] Within a concrete request aggregation method the SDN controller can determine the best available content source SX,i for serving the requested NDO X, wherein i denotes the i' th request seen by the SDN controller and wherein the current time is set to t0. Further, using IX,i, an ingress switch/node of the ICN, and SX,i - the SDN controller can calculate a path for the request RPX,i : IX,iSX,i and a path for returning the NDO PX,i: SX,iIX,i.

    [0029] Then, the SDN controller can determine all other pending request j for the same NDO X, whose request has not yet reached the content source, ETAX,j > t0, and can add them to a list of candidate requests C for aggregation.

    [0030] If the list of candidate requests C is empty, the request will not be aggregated and the SDN controller can configure RPX,i and PX,i and estimate the estimated time of arrival ETAX,i for that request at the content source. If the list of candidate requests C is not empty, the SDN controller can check whether aggregation of requests is possible.

    [0031] Then, the SDN controller can calculate one or more multiple spanning trees, MST, from a set of all possible content sources ({SX,j in C}SX,i) to a set of all ingress switches/nodes ({IX,j in C}IX,i).

    [0032] Then, the SDN controller can evaluate the resulting MSTs regarding their cost. If the lowest cost MST that includes the ingress switch/node IX,i is equivalent to PX,i, then no aggregation will be performed. If there is a MST with lower cost, the SDN controller can instruct the ingress switch/node IX,i to drop the request and update SX,i and all other affected PX,j and can configure a new path in the ICN.

    [0033] The MST can define duplication points in the ICN. Additionally or alternatively, the SDN controller can configure copy rules in the switches/nodes within the ICN.

    [0034] Then, the SDN controller can set the ETA for the current request to the minimum of ETA of the other requests in the selected MST ({ETAX,j in MST}).

    [0035] Within a very simple version of the request aggregation mechanism a feedback can notify the SDN controller when the at least one NDO is sent from its content source. In this case, there is no longer a need for delay annotations in the topology and the ETA estimation.

    [0036] Within a further preferred embodiment, once the at least one NDO has been delivered to the requestor, an entry in the PRT can be removed. This will avoid the storage of data which are no longer necessary for further request aggregation.

    [0037] The present invention provides a mechanism for network-wide request aggregation for ICN leveraging knowledge and control mechanisms of a preferably centralized controller. Main applications for this approach are streaming services or live streaming services, due to the timely synchronized number of requests for the same NDO. The invention can provide a cost-aware optimization of request/response routing and a request/response-timing-aware path-selection. Further, per-request aggregation decisions are possible.

    [0038] The present invention provides a further reduction of transmitted ICN messages, e.g. request and response messages, and a real-time, network-state-aware delivery tree optimization for NDOs.

    [0039] In current ICNs request aggregation is done on a per-node basis. The present invention provides a network-wide approach to realize additional optimization potential.

    [0040] The present invention is useful with ICNs that are deployed with a preferably centralized control approach, that allows to individually steer, drop and copy ICN requests and NDO responses.

    [0041] There are several ways how to design and further develop the teaching of the present invention in an advantageous way. To this end it is to be referred to the patent claims subordinate to patent claim 1 on the one hand and to the following explanation of preferred examples of embodiments of the invention, illustrated by the drawing on the other hand. In connection with the explanation of the preferred embodiments of the invention by the aid of the drawing, generally preferred embodiments and further developments of the teaching will be
    Document US-2012/131144 discloses the closest prior art. It discloses a cloud computing network in which requests for application services coming from multiple subscribers are aggregated by an aggregator node.

    [0042] In the drawings
    Fig. 1
    is showing in a diagram three subsequent situations of a per-node request aggregation within known information-centric networks,
    Fig. 2
    is showing in a diagram three subsequent situations of a network-wide request aggregation according to the invention and
    Fig. 3
    is showing a flow chart of decisions for a network-wide request aggregation at a network controller according to the invention.


    [0043] According to the invention a method for operating an information-centric network, ICN, is provided wherein at least one named data object, NDO, is addressable and wherein after an initial request a request aggregation of subsequent requests for the at least one NDO is performed according to a definable rule. Further, a software-defined networking, SDN, with a SDN controller is implemented in the ICN and a network-wide request aggregation is performed by the SDN controller.

    [0044] The embodiment comprises a network-wide version of request aggregation, made possible through a preferably centralized ICN control by the SDN controller. The detection of duplicate requests will be performed by the SDN controller and the SDN controller can decide centrally on which requests to delay and how to route and where to duplicate the NDOs or responses. Fig. 2 explains the difference to known per-node request aggregation according to Fig. 1. Within Fig. 2 the NDO A needs to be transmitted through the network only once as compared to two times in the per-node request aggregation case - see Fig. 1 - and is duplicated at the node N3 for provision to the nodes N1 and N2 and further transmission to hosts H1, H2 and H3, see the right part of Fig. 2. From the middle part of Fig. 2 is obtainable that subsequent requests R2 and R3 are delayed at nodes N1 and N2 by the SDN controller. Only request R1, see the left part of Fig. 2, is transmitted to the source S for getting the content A. The invention is especially beneficial for streaming or live streaming of data to multiple receivers, as many requests for the same NDO arrive at the same time at different nodes of the network.

    [0045] Within an embodiment of the invention the following assumptions can be made:
    • The software-defined ICN is operating in a 'provision request and response path at the same time' upon the initial handling of the request. When the response path is only provisioned when the NDO response is generated a similar but simpler approach is possible at the SDN controller.
    • The suggested embodiment includes a way to determine the best source for the requested content. This function can or cannot be cache-aware, i.e. know where content is cached.
    • For the rest of our description we use the term "content source" or "source" for both an ICN node actually providing the desired NDO from its storage as well as for an ICN node at the egress of the software-defined ICN network, which knows how to retrieve the content from another ICN.


    [0046] Within the present document and description of the embodiment the term "node" is used synonymously for the elements "node" or "switch" or "router". The term "node" is used for all network elements which can route and/or transmit NDOs or data.

    [0047] This embodiment of a software-defined SDN controller for network-wide request aggregation comprises:
    1. 1. A pending-request table PRT, often called pending interest table in ICN. Entries in the PRT need the following information, see also PRT fields in Fig. 3:
      1. a. The name of the requested NDO (X)
      2. b. A counter differentiating between requests for the same name (n)
      3. c. The identifier of the ingress SDN switch (I)
      4. d. The selected content source, e.g. cache, for the NDO (S)
      5. e. The selected return path for the NDO (P)
      6. f. The requests estimated time of arrival at the content source (ETA)
    2. 2. The topology of the SDN controlled ICN, along with a cost metric for all the possible paths/content sources. The topology needs to be annotated with expected per link one-way delay information. The cost metric serves as input to a multiple spanning tree calculation and can be related to one or a combination of the following information:
      1. a. Link delay
      2. b. Link bandwidth
      3. c. Link utilization
      4. d. ICN content source utilization
      5. e. Other traffic engineering metrics are applicable as well
    3. 3. A procedure to determine if and how to aggregate requests. Fig. 3 gives the high-level overview. The procedure starts in the top left corner ("Start") whenever the SDN controller detects a new request at the ingress of its network. We assume the request is for an NDO with the name X. Furthermore we assume this request is the /th request seen by the SDN controller of this ICN and we set the current time to t0. Finally we denote the ingress node by IX,i.
      1. a. First the controller determines the best available content source SX,i for serving the requested NDO X.
      2. b. Next using IX,i and SX,i the controller calculates a path for the request RPX,i: IX,iSX,i and a path for returning the NDO PX,i: SX,iIX,i
      3. c. Then in Section A the controller determines all other pending requests j for the same NDO with name X, whose request has not yet reached the content source (ETAX,j > t0) and adds them to a list of candidate requests C for aggregation.
      4. d. If this list of candidates is empty (not /C/> 0) the request is not aggregated and the controller configures RPX,i and PX,i. This is e.g. the case when no other requests for the same NDO are present in the ICN. In addition the controller estimates the estimated time of arrival ETAX,i for that request at the content source. The next step is step f.
      5. e. If however the list of candidates is not empty (/C/> 0), the controller enters in Section B and tries to see if there is a way to aggregate requests. For that purpose the controller calculates multiple spanning trees, MSTs, from the set of all possible sources ({SX,j in C} U SX,i) to the set of all ingress nodes ({IX,j in C} U IX,i). Then the controller evaluates the resulting MSTs regarding their cost.
        1. i. If the lowest cost MST that includes the ingress node IX,i is equivalent to PX,i then there is no need for request aggregation and the controller continues with the actions described in step d.
        2. ii. However if there is a better spanning tree with lower cost the controller instructs the ingress IX,i switch to drop the request and update SX,i. It will also update PX,i and all other affected PX,j and it configures the new path in the network. The spanning tree defines the duplication points in the network, and the controller will configure copy rules in the switches accordingly. As a final step before continuing with step f the controller sets the ETA for the current request to the minimum of the ETA of the other requests in the selected MST ({ETAX,j in MST}).
      6. f. Enter request in PRT and end.


    [0048] Within the above description i and j are non-negative integers.

    [0049] A simplified version of this control mechanism could be to configure the network in a way so that explicit feedback notifies the SDN controller when NDOs are sent out. That way there is no longer a need for delay annotations in the topology and the ETA estimation - done in the two boxes before the "Enter to PRT" box in Fig. 3. In Section A the ETA checking would be swapped out for a simple check if the NDO response has already been sent out. This requires storing this binary information in the PRT, replacing the ETA information. However this simplified version requires communicating the sending of the NDO to the SDN controller and a processing of this trigger, updating the PRT. This will either cause overhead at the ICN object source node or the SDN controller or switch next to it.

    [0050] In the embodiment described above we limit our optimizations to requests that have not yet reached their content source, because we need to be able to configure the return path. With tighter control over the information-centric network it is also possible to keep optimizing even if the NDO is already on its way to the requestor(s). However there is a trade-off between tight control and state-tracking overhead at the SDN controller and in the control network.

    [0051] An even tighter control would be required for a system that exploits knowledge about on-path caching, and uses predictions to direct requests to nodes not yet holding the requested NDO.

    [0052] Once NDOs have been delivered to the requestor the entry in the PRT should be removed. However the PRT eviction strategy is not very relevant for this embodiment, as we preselect the candidate PRT entries based on the ETA of the request at the content source.

    [0053] The invention can co-exist with cache optimization and other traffic engineering mechanisms. The cache optimization should be interfaced with the "Determine best content source" function in Fig. 3. Traffic engineering should probably use the same cost metric as the multiple spanning tree calculation.

    [0054] Many modifications and other embodiments of the invention set forth herein will come to mind the one skilled in the art to which the invention pertains having the benefit of the teachings presented in the foregoing description and the associated drawings. Therefore, it is to be understood that the invention is not to be limited to the specific embodiments disclosed, the scope of the invention being defined by the appended claims. Although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.


    Claims

    1. A method for operating an information-centric network, ICN, wherein at least one named data object, NDO, is addressable, wherein after an initial request a request aggregation of subsequent requests for the at least one NDO is performed according to a definable rule,
    wherein a software-defined networking, SDN, with a SDN controller is implemented in the ICN, wherein a network-wide request aggregation is performed by the SDN controller and
    wherein the SDN controller decides how to route and/or where to duplicate the at least one NDO.
     
    2. A method according to claim 1, wherein the SDN controller detects duplicate requests and/or
    wherein the SDN controller decides on which requests to delay.
     
    3. A method according to claim 1 or 2, wherein the ICN operates in a mode wherein a request path and a response path are provided at the same time upon an initial handling of the initial request, or
    wherein the ICN operates in a mode wherein a response path is only provided when a NDO response is generated.
     
    4. A method according to one of claims 1 to 3, wherein a best content source for the at least one NDO will be determined,
    wherein preferably the determination is performed under the knowledge where content of the NDO or the NDO is cached.
     
    5. A method according to one of claims 1 to 4, wherein the SDN controller comprises a Pending-Request Table, PRT,
    wherein preferably the PRT comprises a name (X) or identification of the requested NDO and/or a counter (n) differentiating between requests for the same name (X) or identification and/or an identifier (I) of an ingress SDN switch/node and/or a selected content source (S) for the NDO and/or a selected return path (P) for the NDO and/or the estimated time of arrival (ETA) of the requests at the content source.
     
    6. A method according to one of claims 1 to 5, wherein the SDN controller comprises the topology of the ICN and/or
    wherein the SDN controller comprises an expected per link one-way delay information.
     
    7. A method according to one of claims 1 to 6, wherein the SDN controller comprises a cost metric for all possible paths and/or content sources (S),
    wherein preferably the cost metric serves as input to a spanning tree calculation and/or
    wherein preferably the cost metric is related to a link delay and/or a link bandwidth and/or a link utilization and/or a content source (S) utilization.
     
    8. A method according to one of claims 1 to 7, wherein the SDN controller comprises the definable rule.
     
    9. A method according to one of claims 1 to 8, wherein the SDN controller determines the best available content source SX,i for serving the requested NDO X, wherein i denotes the i'th request seen by the SDN controller and wherein the current time is set to t0,
    wherein preferably - using IX,i, an ingress switch/node of the ICN, and SX,i - the SDN controller calculates a path for the request RPX,i: IX,iSX,i and a path for returning the NDO PX,i: SX,iIX,i,
    wherein further preferably the SDN controller determines all other pending requests for the same NDO X, whose request has not yet reached the content source, ETAX,j> t0, and adds them to a list of candidate requests C for aggregation,
    wherein still further preferably, if the list of candidate requests C is empty, the request will not be aggregated and the SDN controller configures RPX,i and PX,i and estimates the estimated time of arrival ETAX,i for that request at the content source or
    wherein still further preferably, if the list of candidate requests C is not empty, the SDN controller checks whether aggregation of requests is possible.
     
    10. A method according to claim 9, wherein the SDN controller calculates one or more multiple spanning trees, MST, from a set of all possible content sources ({SX,i in C} U SX,i) to a set of all ingress switches/nodes ({IX,j in C} U IX,i),
    wherein preferably the SDN controller evaluates the resulting MSTs regarding their cost,
    wherein further preferably, if the lowest cost MST that includes the ingress switch/node IX,i is equivalent to PX,i, then no aggregation will be performed or
    wherein further preferably, if there is a MST with lower cost, the SDN controller instructs the ingress switch/node IX,i to drop the request and update SX,i and all other affected PX,j and configures a new path in the ICN.
     
    11. A method according to claim 10, wherein the MST defines duplication points in the ICN and/or
    wherein the SDN controller configures copy rules in the switches/nodes within the ICN.
     
    12. A method according to claim 11, wherein the SDN controller sets the ETA for the current request to the minimum of ETA of the other requests in the selected MST ({ETAX,j in MST}).
     
    13. A method according to one of claims 1 to 12, wherein a feedback notifies the SDN controller when the at least one NDO is sent from its content source.
     
    14. A method according to one of claims 5 to 13, wherein an entry in the PRT is removed, once the at least one NDO has been delivered to the requestor.
     
    15. An information-centric network, ICN, adapted to carry out the method according to any one of claims 1 to 14, comprising at least one addressable named data object, NDO, and means for performing after an initial request a request aggregation of subsequent requests for the at least one NDO according to a definable rule, the ICN further comprising a software-defined network, SDN, said SDN including an SDN controller, wherein the SDN controller is configured for performing a network-wide request aggregation and
    wherein the SDN controller is adapted to decide how to route and/or where to duplicate the at least one NDO.
     


    Ansprüche

    1. Ein Verfahren zum Betreiben eines informationszentrierten Netzwerks, ICN, wobei zumindest ein benanntes Datenobjekt, NDO, adressierbar ist, wobei nach einer anfänglichen Anforderung eine Anforderungsaggregation nachfolgender Anforderungen für das zumindest eine NDO gemäß einer definierbaren Regel durchgeführt wird,
    wobei ein softwaredefinierter Netzwerkbetrieb, SDN, mit einem SDN-Controller in dem ICN implementiert wird, wobei eine netzwerkweite Anforderungsaggregation durch den SDN-Controller durchgeführt wird und
    wobei der SDN-Controller entscheidet, wie das zumindest eine NDO zu routen und/oder wo das zumindest eine NDO zu duplizieren ist.
     
    2. Ein Verfahren nach Anspruch 1, wobei der SDN-Controller doppelte Anforderungen detektiert und/oder
    wobei der SDN-Controller entscheidet, auf welche Anfragen hin ein Verzögern durchzuführen ist.
     
    3. Ein Verfahren nach Anspruch 1 oder 2, wobei das ICN in einem Modus arbeitet, wobei ein Anforderungspfad und ein Antwortpfad auf ein anfängliches Bearbeiten der anfänglichen Anforderung hin zur selben Zeit bereitgestellt werden, oder wobei das ICN in einem Modus arbeitet, wobei ein Antwortpfad nur bereitgestellt wird, wenn eine NDO-Antwort erzeugt wird.
     
    4. Ein Verfahren nach einem der Ansprüche 1 bis 3, wobei eine Quelle mit einem besten Inhalt für das zumindest eine NDO bestimmt wird,
    wobei vorzugsweise die Bestimmung mit dem Wissen durchgeführt wird, wo Inhalt des NDO oder das NDO gecached wird.
     
    5. Ein Verfahren nach einem der Ansprüche 1 bis 4, wobei der SDN-Controller eine Tabelle ausstehender Anforderungen, Pending-Request Table, PRT, aufweist,
    wobei vorzugsweise die PRT einen Namen (X) oder eine Identifizierung des angeforderten NDO und/oder einen Zähler (n), der zwischen Anforderungen für denselben Namen (X) oder dieselbe Identifizierung differenziert, und/oder einen Identifikator (I) eines Zugangs-SDN-Switches/Knotens und/oder eine ausgewählte Inhaltsquelle (S) für das NDO und/oder einen ausgewählten Rückpfad (P) für das NDO und/oder die abgeschätzte Ankunftszeit (ETA) der Anforderungen an der Inhaltsquelle aufweist.
     
    6. Ein Verfahren nach einem der Ansprüche 1 bis 5, wobei der SDN-Controller die Topologie des ICN umfasst und/oder
    wobei der SDN-Controller eine erwartete Einwegsverzögerungsinformation pro Link aufweist.
     
    7. Ein Verfahren nach einem der Ansprüche 1 bis 6, wobei der SDN-Controller eine Kostenmetrik für alle möglichen Pfade und/oder Inhaltsquellen (S) aufweist,
    wobei vorzugsweise die Kostenmetrik als Eingabe für eine Spannbaumberechnung dient und/oder
    wobei vorzugsweise die Kostenmetrik mit einer Linkverzögerung und/oder einer Linkbandbreite und/oder einer Linknutzung und/oder einer Nutzung einer Inhaltsquelle (S) in Beziehung steht.
     
    8. Ein Verfahren nach einem der Ansprüche 1 bis 7, wobei der SDN-Controller die definierbare Regel aufweist.
     
    9. Ein Verfahren nach einem der Ansprüche 1 bis 8, wobei der SDN-Controller die beste verfügbare Inhaltsquelle SX,i zum Bedienen des angeforderten NDO X bestimmt, wobei i die aus Sicht des SDN-Controllers i-te Anforderung bezeichnet und wobei die aktuelle Zeit auf t0 gesetzt wird,
    wobei vorzugsweise - unter Verwendung von IX,i, eines Zugangs-Switches/Knotens des ICN, und SX,i - der SDN-Controller einen Pfad für die Anforderung RPX,i: IX,iSX,i und einen Pfad für das Rückführen des NDO PX,i: SX,iIX,i berechnet,
    wobei weiter vorzugsweise der SDN-Controller alle weiteren ausstehenden Anforderungen für dasselbe NDO X bestimmt, dessen Anforderung die Inhaltsquelle noch nicht erreicht hat, ETAX,j > t0, und diese zu einer Liste von Kandidatenanforderungen C für eine Aggregation hinzufügt,
    wobei noch weiter vorzugsweise, falls die Liste von Kandidatenanforderungen C leer ist, die Anforderung nicht aggregiert wird und der SDN-Controller RPX,i und PX,i konfiguriert und die abgeschätzte Ankunftszeit ETAX,i für diese Anforderung an der Inhaltsquelle abschätzt oder
    wobei noch weiter vorzugsweise, falls die Liste von Kandidatenanforderungen C nicht leer ist, der SDN-Controller prüft, ob eine Aggregation von Anforderungen möglich ist.
     
    10. Ein Verfahren nach Anspruch 9, wobei der SDN-Controller einen oder mehrere multiple Spannbäume, MST, aus einem Set von allen möglichen Inhaltsquellen ({SX,j in C} U SX,i) zu einem Set von allen Zugangs-Switches/Knoten ({IX,j in C} U IX,i) berechnet,
    wobei vorzugsweise der SDN-Controller die resultierenden MSTs hinsichtlich ihrer Kosten evaluiert,
    wobei weiter vorzugsweise dann keine Aggregation durchgeführt wird, falls der MST mit den niedrigsten Kosten, der den Zugangs-Switch/Knoten IX,i einschließt, zu PX,i äquivalent ist, oder
    wobei weiter vorzugsweise, falls es einen MST mit geringeren Kosten gibt, der SDN-Controller den Zugangs-Switch/Knoten IX,i instruiert, die Anforderung fallen zu lassen und SX,i und alle weiteren betroffenen PX,j zu aktualisieren, und einen neuen Pfad in dem ICN konfiguriert.
     
    11. Ein Verfahren nach Anspruch 10, wobei der MST Duplizierungspunkte in dem ICN definiert und/oder
    wobei der SDN-Controller Kopierregeln in den Switches/Knoten innerhalb des ICN konfiguriert.
     
    12. Ein Verfahren nach Anspruch 11, wobei der SDN-Controller die ETA für die aktuelle Anforderung auf das ETA-Minimum der anderen Anforderungen in dem ausgewählten MST ({ETAX,j in MST}) setzt.
     
    13. Ein Verfahren nach einem der Ansprüche 1 bis 12, wobei eine Rückkopplung den SDN-Controller benachrichtigt, wenn das zumindest eine NDO von seiner Inhaltsquelle gesendet wird.
     
    14. Ein Verfahren nach einem der Ansprüche 5 bis 13, wobei ein Eintrag in der PRT entfernt wird, sobald das zumindest eine NDO zu dem Anfordernden übermittelt worden ist.
     
    15. Ein informationszentriertes Netzwerk, ICN, das zum Ausführen des Verfahrens nach einem der Ansprüche 1 bis 14 angepasst ist, umfassend zumindest ein adressierbares benanntes Datenobjekt, NDO, und Mittel zum Durchführen einer Anforderungsaggregation von nachfolgenden Anforderungen für das zumindest eine NDO gemäß einer definierbaren Regel und zwar nach einer anfänglichen Anforderung,
    wobei das ICN weiter ein softwaredefiniertes Netzwerk, SDN, aufweist, wobei das SDN einen SDN-Controller aufweist,
    wobei der SDN-Controller zum Durchführen einer netzwerkweiten Anforderungsaggregation konfiguriert ist und
    wobei der SDN-Controller angepasst ist, um zu entscheiden, wie das zumindest eine NDO zu routen und/oder wo das zumindest eine NDO zu duplizieren ist.
     


    Revendications

    1. Procédé pour exploiter un réseau centré sur l'information, ICN, dans lequel au moins un objet de données nommé, NDO, est adressable, dans lequel après une requête initiale, une agrégation de requêtes de requêtes subséquentes pour l'au moins un NDO est effectuée selon une règle définissable,
    dans lequel un réseautage défini par logiciel, SDN, avec un contrôleur SDN est mis en oeuvre dans l'ICN, dans lequel une agrégation de requêtes à travers le réseau est effectuée par le contrôleur SDN et
    dans lequel le contrôleur SDN décide comment router et/ou où dupliquer l'au moins un NDO.
     
    2. Procédé selon la revendication 1, dans lequel le contrôleur SDN détecte des requêtes en double et/ou
    dans lequel le contrôleur SDN décide quelles requêtes retarder.
     
    3. Procédé selon la revendication 1 ou 2, dans lequel l'ICN fonctionne dans un mode dans lequel un chemin de requête et un chemin de réponse sont fournis en même temps lors d'une prise en charge initiale de la requête initiale, ou
    dans lequel l'ICN fonctionne dans un mode dans lequel un chemin de réponse n'est fourni que lorsqu'une réponse NDO est générée.
     
    4. Procédé selon l'une des revendications 1 à 3, dans lequel une meilleure source de contenu pour l'au moins un NDO sera déterminée,
    dans lequel de préférence la détermination est effectuée en sachant où le contenu du NDO ou le NDO est mis en cache.
     
    5. Procédé selon l'une des revendications 1 à 4, dans lequel le contrôleur SDN comprend une Table de Requêtes en Attente, PRT,
    dans lequel de préférence la PRT comprend un nom (X) ou une identification du NDO demandé et/ou un compteur (n) différenciant des requêtes pour le même nom (X) ou identification et/ou un identifiant (I) d'un commutateur/noeud SDN d'entrée et/ou une source de contenu sélectionnée (S) pour le NDO et/ou un chemin de retour sélectionné (P) pour le NDO et/ou le temps d'arrivée estimé (ETA) des requêtes au niveau de la source de contenu.
     
    6. Procédé selon l'une des revendications 1 à 5, dans lequel le contrôleur SDN comprend la topologie de l'ICN et/ou
    dans lequel le contrôleur SDN comprend une information de retard unidirectionnel par liaison attendu.
     
    7. Procédé selon l'une des revendications 1 à 6, dans lequel le contrôleur SDN comprend une métrique de coût pour tous les chemins et/ou sources de contenu (S) possibles,
    dans lequel de préférence la métrique de coût sert d'entrée à un calcul d'arbre recouvrant et/ou
    dans lequel de préférence la métrique de coût est liée à un retard de liaison et/ou une largeur de bande de liaison et/ou une utilisation de liaison et/ou une utilisation de source de contenu (S).
     
    8. Procédé selon l'une des revendications 1 à 7, dans lequel le contrôleur SDN comprend la règle définissable.
     
    9. Procédé selon l'une des revendications 1 à 8, dans lequel le contrôleur SDN détermine la meilleure source de contenu disponible SX,i pour desservir le NDO X demandé, où i désigne la i'ième requête vue par le contrôleur SDN et où le temps actuel est établi à t0,
    dans lequel de préférence - en utilisant IX,i, un commutateur/noeud d'entrée de l'ICN, et SX,i - le contrôleur SDN calcule un chemin pour la requête RPX,i : IX,iSX,i et un chemin pour retourner le NDO PX,i : SX,iIX,i,
    dans lequel en outre de préférence le contrôleur SDN détermine toutes les autres requêtes en attente pour le même NDO X, dont la requête n'a pas encore atteint la source de contenu, ETAX,j > t0, et les ajoute à une liste de requêtes candidates C pour l'agrégation,
    dans lequel encore en outre de préférence, si la liste des requêtes candidates C est vide, la requête ne sera pas agrégée et le contrôleur SDN configure RPX,i et PX,i et estime le temps d'arrivée estimé ETAX,i pour cette requête au niveau de la source de contenu ou
    dans lequel encore en outre de préférence, si la liste de requêtes candidates C n'est pas vide, le contrôleur SDN vérifie si l'agrégation de requêtes est possible.
     
    10. Procédé selon la revendication 9, dans lequel le contrôleur SDN calcule un ou plusieurs arbres recouvrants multiples, MST, à partir d'un ensemble de toutes les sources de contenu possibles ({SX,j dans C} U SX,i) sur un ensemble de tous les commutateurs/noeuds d'entrée ({IX,j dans C} U IX,i),
    dans lequel de préférence le contrôleur SDN évalue les MST résultants concernant leur coût,
    dans lequel en outre de préférence, si le MST au coût le plus bas qui inclut le commutateur/noeud d'entrée IX,i est équivalent à PX,i, alors aucune agrégation ne sera effectuée ou
    dans lequel en outre de préférence, s'il y a un MST avec un coût inférieur, le contrôleur SDN donne pour instruction au commutateur/noeud d'entrée IX,i d'abandonner la requête et de mettre à jour SX,i et tous les autres PX,j affectés et configure un nouveau chemin dans l'ICN.
     
    11. Procédé selon la revendication 10, dans lequel le MST définit des points de duplication dans l'ICN et/ou
    dans lequel le contrôleur SDN configure des règles de copie dans les commutateurs/noeuds au sein de l'ICN.
     
    12. Procédé selon la revendication 11, dans lequel le contrôleur SDN établit l'ETA pour la requête actuelle au minimum d'ETA des autres requêtes dans le MST sélectionné ({ETAX,j dans MST}).
     
    13. Procédé selon l'une des revendications 1 à 12, dans lequel une rétroaction notifie au contrôleur SDN quand l'au moins un NDO est envoyé depuis sa source de contenu.
     
    14. Procédé selon l'une des revendications 5 à 13, dans lequel une entrée dans la PRT est supprimée, une fois que l'au moins un NDO a été remis au demandeur.
     
    15. Réseau centré sur l'information, ICN, adapté pour effectuer le procédé selon l'une quelconque des revendications 1 à 14, comprenant au moins un objet de données nommé, NDO, adressable et des moyens pour effectuer après une requête initiale, une agrégation de requêtes de requêtes subséquentes pour l'au moins un NDO selon une règle définissable,
    l'ICN comprenant en outre un réseau défini par logiciel, SDN, ledit SDN incluant un contrôleur SDN,
    dans lequel le contrôleur SDN est configuré pour effectuer une agrégation de requêtes à travers le réseau et
    dans lequel le contrôleur SDN est adapté pour décider comment router et/ou où dupliquer l'au moins un NDO.
     




    Drawing












    REFERENCES CITED IN THE DESCRIPTION



    This list of references cited by the applicant is for the reader's convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard.

    Patent documents cited in the description




    Non-patent literature cited in the description