(19)
(11)EP 2 377 328 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
02.05.2018 Bulletin 2018/18

(21)Application number: 10700446.7

(22)Date of filing:  07.01.2010
(51)International Patent Classification (IPC): 
H04Q 11/04(2006.01)
H04L 12/28(2006.01)
H04L 12/24(2006.01)
H04M 3/30(2006.01)
(86)International application number:
PCT/GB2010/000016
(87)International publication number:
WO 2010/082016 (22.07.2010 Gazette  2010/29)

(54)

MANAGEMENT OF TELECOMMUNICATIONS CONNECTIONS

VERWALTUNG VON TELEKOMMUNIKATIONSVERBINDUNGEN

GESTION DE CONNEXIONS DE TÉLÉCOMMUNICATIONS


(84)Designated Contracting States:
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 SE SI SK SM TR

(30)Priority: 15.01.2009 EP 09250095

(43)Date of publication of application:
19.10.2011 Bulletin 2011/42

(73)Proprietor: British Telecommunications public limited company
London EC1A 7AJ (GB)

(72)Inventor:
  • COOPER, Ian, Robert
    Suffolk IP5 2EW (GB)

(74)Representative: British Telecommunications public limited company Intellectual Property Department 
Ground Floor, Faraday Building 1 Knightrider Street
London EC4V 5BT
London EC4V 5BT (GB)


(56)References cited: : 
WO-A-2007/012867
US-A1- 2004 120 390
  
  • KEE BONG SONG ET AL: "Dynamic Spectrum Management for Next-Generation DSL Systems" IEEE COMMUNICATIONS MAGAZINE, October 2002 (2002-10), pages 101-109, XP011092935 ISSN: 0163-6804
  • "Dynamic Line Management for Digital Subscriber Lines" ALCATEL TECHNOLOGY WHITE PAPER, [Online] XP002442256 Retrieved from the Internet: URL:http://www1.alcatel-lucent.com/com/en/ appcontent/apl/18812_DLM_twp_tcm172-228691 635.pdf> [retrieved on 2007-07-12]
  
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] This invention relates to telecommunications systems and in particular to the management of network equipment interfacing between a network and individual customer premises systems. Such equipment is widely dispersed geographically, and has to operate without direct human supervision and in a wide variety of environments and circumstances.

[0002] As shown in Figure 1, in conventional Digital Subscriber Loop (DSL) services, provided from the exchange 39 (or cabinet), each customer premises 2 has a dedicated physical connection 30 to the DSL access multiplexer (DSLAM) 31 in the exchange 39. The connections from the exchange 39 to several different customer premises 2 may pass through a single distribution point 1, but each connection is a complete end-to-end connection.

[0003] A management system 18 can be provided to optimise the service for each customer by maximising the data rate over the physical layer 30 (subject to a predetermined maximum) whilst maintaining the stability of the line. This is achieved for each line using a Dynamic Line Management (DLM) system and a Rate Adaptive Management Box (RAMBo) 41 which automatically selects the optimum rate profile for each line. The chosen profile rate (upstream and downstream) supported by the line is also applied to the BRAS (Broadband Remote Access Server) 42 serving the user connection 32 so that the services provided over the DSL line 30 match the physical capabilities of the line. The BRAS is not located at the exchange but is located deeper in the network. It can handle many thousands of lines and would provide the broadband services for many exchanges).

[0004] The physical layer connectivity is provided by a Digital subscriber line access multiplexer (DSLAM) capped at a predetermined rate limit, e.g. 5Mbit/s, and the BRAS provides the services to the DSLAM so that the services are capped to the same rate limit so that there is rate matching between the physical line and the services that are applied over that line. In order to perform this function it is necessary to gather a wide selection of performance statistics from each line at regular intervals (e.g. every fifteen minutes), store these in a data-warehouse and perform subsequent post-processing of this data in order to choose the correct rate profile for each line. Also, the previous history of the line has to be accounted for in order to provide some hysteresis i.e. to prevent the line profile (rates) being changed by too much, or too frequently, both of which can result in difficulties in maintaining services such as streaming. Typical statistics gathered may include bit-rates, margins, errored seconds, severely errored seconds, and mean time between errors. These statistics are stored by a data collector and fed into the Digital line management system 18, which is responsible for selecting an appropriate DSL profile for each line.

[0005] However, as shown in Figure 2, this can become difficult to achieve when fibre-to-the-distribution-point (FttDP) nodes are considered. In such systems the connections 32 between the optical line terminal (OLT) 33 in the exchange and the individual distribution points 1 are provided by optical fibre, each carrying the traffic for all the final drop connections 30 served by that distribution point. This allows the distribution point to serve a large number of customer premises. Instead of a single DSLAM 31 providing the line statistics for thousands of lines at one convenient location, there could be a large number of remote nodes located at the distribution points 1, each provisioning between 8 and 24 lines.

[0006] Because of the transition between optical fibre and electrical "copper" connections at the distribution points, they have more capabilities than a typical copper-to-copper distribution point. Essentially the modem conventionally located in the DSLAM 31 at the exchange 39 is instead located in a mini-DSLAM 34 at the DP 1 (only shown for one DP in Figure 2). Thus the DSLAM 31 and BRAS 42 functions are no longer co-located.

[0007] As well as having some active electronics at the DP, some intelligence can be added. This allows the line characteristics to be measured at the distribution point 1, and such an arrangement is described in International Patent Specification 2007/012867. The DSL modem 34 located at the distribution point has the ability to draw statistics both from itself and the equivalent modem 2 on the other end of the local loop located at the customer premises (i.e. it gathers both upstream and downstream line performance statistics).

[0008] However, in this arrangement each distribution point has to transmit the periodically-gathered statistics back to the remote data collector 43 associated with the central management function 18, so that its associated RAMBo (41) can set the rate for each line. This rate then needs to be communicated to both ends of the connection 30 between the distribution point 1 and customer terminal 30. The dotted lines in Figure 2 illustrate these information flows. This central data collection system adds to the operations, administration, and maintenance overhead that the network has to carry and requires a data warehouse and large central processing capabilities. Moreover all of this statistical data has to be transmitted in-band with user data, thus adding considerably to the overhead burden of the transmission system. Such a system would require a very complex line management system, and such an arrangement would be very inefficient. An additional consideration is that the optical fibre connection 32 is unsuitable for delivering an electrical power supply from the exchange side 31, so power to operate the distribution point, including the optical/electrical transducers, has to be sourced from elsewhere. Related systems are described in published United States Patent application US2004/0120390 (Brown), and by Kee Bong Song et al, ("Dynamic Spectrum Management for Next-Generation DSL Systems": IEEE Magazine October 2002).

[0009] According to the present invention, there is provided a dynamic line management system for processing data relating to the capabilities of each of a plurality of digital subscriber loops each serving a respective termination point of a telecommunications system, the dynamic line management system comprising means for generating a profile of each digital subscriber loop and setting a rate profile to allow control of the transmission of data to the individual termination points, characterised in that the dynamic line management system is associated with a network distribution point operating as a node intermediate between a remote access server and the respective digital subscriber loops, and the network distribution point is arranged to transmit data to the individual termination points under the control of the associated dynamic line management system.

[0010] The present invention takes advantage of the data processing power available at the mini-DSLAM in each distribution point, and the availability at the DP of the data to perform dynamic line management (DLM), to make the DP autonomous in setting its own maximum stable DSL rate, by processing the data relating to line capabilities locally at the distribution points, and implementing any subsequent change of DLM profile locally. This approach still allows decisions on DLM profile choice to be made taking into account demands on neighbouring customer terminals sharing the same DP 1. This configuration allows some of the DSLAM functions to be performed by the individual remote nodes, allowing each remote node to implement a local autonomous system using the physical layer statistics it collects, and to process them locally to provide an optimum DSL profile for each line whilst retaining system stability. In particular, the RAMBo functions 41 are migrated to the distribution points 1.

[0011] The operational environment of each line served by the Distribution Point 1 will be very similar, so there are advantages in implementing an autonomous DLM system at the DP. This would allow each DSL line operating from a remote node located at the DP to be optimised, based upon the local operational environment as seen at that DP.

[0012] In a fibre-to-the-distribution-point system it is impractical to power the distribution points from the exchange side because of the absence of a wired connection. The remote nodes therefore take power from the customer end of the connection. It is therefore important that power consumption is minimised.

[0013] In one embodiment of the invention, the local processing of physical layer DSL line data is achieved using an Artificial Neural Network (ANN). The Multilayer Perceptron (MLP) is the most frequently used Artificial Neural Network, due to its ability to model non-linear systems and establish non-linear decision boundaries in classification or prediction problems. Furthermore, the MLP is a universal function approximator, which makes it a powerful tool in several signal processing fields: pattern recognition, system modelling, control, etc.

[0014] The MLP is suitable for the field of digital line management as this application can be considered as a combination of pattern recognition (recognising specific patterns of input line statistics) and control system (changing the DLM profile accordingly). Initially it is necessary to 'train' the MLP in order to optimise the physical layer performance of the line from the gathered statistics. The computational burden of implementing an MLP lies in this initial training of the system, that is in calculating the 'weights' of the nodes and links. However, once trained, the MLP requires very little computational power in order to identify the target profile for a loop, given a set of physical layer statistics. This therefore results in low processing and power overheads, which is particularly desirable if power is to be taken from the customer side.

[0015] Existing DLM systems using a Rate Adaptive Management Box (RAMBO) treat all lines individually, without considering local similarities between lines that share the same cable bundle etc. However, at the Distribution Point the operational environment of each line will be very similar, so there are advantages in implementing an autonomous DLM system at the DP. This would allow each DSL line operating from a remote node located at the DP to be optimised, based upon the local operational environment as seen at that DP.

[0016] One advantage of using neural nets for such a system is that all of the processor intensive work can be performed during the training of the neural net. Once trained, the neural net is instantiated in the remote node and each analysis of the input data is a simple single iteration through the neural net, which will be just a few multiplications and additions. Therefore the computational load on the remote node processor resources will be minimal. More processing power would be required if a training algorithm is implemented to allow local adaptation at the remote node, but this training could be performed when there was plenty of spare computational resource available. As these remote nodes are powered from the CPE and that computational power (and electrical power in general) is therefore not a resource to be squandered.

[0017] It would be possible to train a single MLP and then provide an instance of this MLP in all the remote nodes located at the Distribution Points. This would form the basis of the local DLM system in each remote node. As time progresses each MLP can be allowed to slowly mutate into a neural network optimised for the particular statistics generated by the local loops attached to that particular node.

[0018] An embodiment will now be described with reference to Figures 3 and 4 of the drawings.

Figure 1 depicts a conventional digital subscriber loop system, already discussed.

Figure 2 depicts a fibre-to-the-data-point system, , already discussed.

Figure 3 depicts a fibre-to-the-data-point system modified according to the invention,

Figure 4 shows how the neural net digital line management system of the invention is incorporated into the network node functions.

Figure 5 shows the neural net digital line management system in more detail.



[0019] It should be understood that Figures 3 to 5 illustrate the functional elements of the system, which may in practice be embodied in one or more electronic components or in software.

[0020] Figure 3 illustrates the configuration of the system, and by comparison with Figure 2 the modifications made according to he invention can be understood. The principal difference is that the Rate Adaptive Management Box (RAMBo) 41 has been migrated from the management system 18 associated with the optical line termination 33 in the exchange to the distribution point 1. (It will be understood that each distribution point can be similarly equipped). The data collection and processing functions controlling the data rate to be used over the DSL link 30 can therefore be performed at one end of that link. Thus instead of measurement data sent from the DP 1 to the RAMBo 41 in an exchange-based management system 18, and the resulting data rate required returned to the DP 1, the RAMBo 41 located in the DP 1 determines the data rate for itself, and only needs to communicate the required data rate to the BRAS 42 and the individual customer terminals 2, as shown by the dotted lines in Figure 3.

[0021] Figure 4 depicts a node 1 (distribution point) having a wired connection 30 to customer premises equipment 2 and an optical connection 32 to a Digital subscriber line access multiplexer (DSLAM) 31. Each wired customer connection is connected to an xDSL Transmission Unit (Optical) (XTU-O) 16, and the optical connection 32 is connected through an optical network unit (ONU) 15. These are interlinked by a interface unit 17 for handling functionality at levels 2 and 3 of the standard OSI seven-level model, under the control of a dynamic line management system 18. This function includes the multiplexing/demultiplexing of the various customer lines over the optical connection 32.

[0022] Having a local Dynamic line management system 18 in each node reduces the requirement for processing power, memory storage requirements, and communications back to a central DLM controller.

[0023] In the present invention the dynamic line management system is operated under the control of a Multilayer Perceptron 19.

[0024] Figure 5 shows the basic components of the Multi-layer perceptron block 19. They comprise a data collector 50 and a neural net subsystem 51, and operate in conjunction with the dynamic line management system 18 itself.

[0025] The data collector 50 gathers line data from each local modem 16. A pre-processing unit 52 prepares the data for input to the neural network 51, by changing the format of the data into a form that can be 'read' by the MLP. Such pre-processing may take say a running average of several measurements in order to prevent too sudden a change in input parameters into the MLP which could result in wildly fluctuation DLM profile choice. The neural network 51 assesses the data and identifies the prevailing DSL performance data, to generate an output which is then passed to a post-processor 53 for presenting the data in a form suitable for use by the DLM processor 18, which generates a profile for use by the DSL modem 16.

[0026] The profile selected by the DLM processor 18 impacts the rate at which the DSL system 16 can transmit/receive, so the profile information is also transmitted to the Broadband Remote Access Server (BRAS) 42 in the DSLAM 31. This allows the BRAS to moderate the rate at which it transmits data, to avoid data being provided from the core IP network faster than it can be transmitted over the DSL link 30, and therefore having to be discarded.

[0027] The DLM 18, and neural net that informs it, handles data relating to several lines 30 serving different customer premises equipment 2, so that at times of high contention (the total capacity required by he users exceeding the capabilities of the network equipment), the available capacity can be distributed fairly, for example to ensure that the level of quality of service to each user meets a respective agreed level. These capacity constraints are unlikely to be on the optical connection 32 itself, but in the DSLAM 31 and ONU 15 between which it is connected.

[0028] The inputs to the dynamic line management system 18 may include data on the RF environment, to allow frequencies subject to local interference to be excluded from the transmissions over the wired local connection 30. Such a system is described in the applicant's co-pending International patent application claiming priority from European application 09250100.6, entitled Telecommunications Connections Management, and subsequently published as WO2010/082014.


Claims

1. A dynamic line management system (19) for processing data relating to the capabilities of each of a plurality of digital subscriber loops (30) each serving a respective termination point (2) of a telecommunications system, the dynamic line management system comprising means for generating a profile of each digital subscriber loop and setting a rate profile to allow control of the transmission of data to the individual termination points,
characterised in that the dynamic line management system is associated with a network distribution point (1) operating as a node intermediate between a remote access server (42) and the respective digital subscriber loops (30), and the network distribution point is arranged to transmit data to the individual termination points (2) under the control of the associated dynamic line management system.
 
2. A dynamic line management system according to claim 1, arranged to transmit the rate profile to the remote access server (42).
 
3. A dynamic line management system according to claim 1 or claim 2, comprising an Artificial Neural Network for processing of data relating to the physical layer of each digital subscriber loop (30).
 
4. A dynamic line management system according to claim 3, wherein the artificial neural network is a Multilayer Perceptron.
 
5. A dynamic line management system according to any preceding claim wherein the network distribution point (1) is connected to the remote access server (42) is by means of an optical fibre system (32) and the connections (30) to the individual termination points (2) are made by electrical means.
 
6. A method of controlling the transmission of data to individual network terminations (2) served from a remote access server (42) by respective digital subscriber loops (30) through one or more distribution points (1), wherein data relating to the capabilities of each of the digital subscriber loops is processed by a dynamic line management system (19) to generate a profile of each digital subscriber loop and to set a rate profile,
characterised in that the profiles of the digital subscriber loops (30) associated with the or each distribution point (2) are generated by a dynamic line management system (19) associated with the respective distribution point, and each network distribution point (2) transmits data to the individual termination points under the control of its associated dynamic line management system (19).
 
7. A method according to claim 6, wherein a rate profile is transmitted from the dynamic line management system (19) associated with the, or each, distribution point (2) to the remote access server (42).
 
8. A method according to claim 6 or 7, wherein the dynamic line management system (19) comprises an Artificial Neural Network for processing of data relating to the physical layer of a digital subscriber loop.
 
9. A method according to claim 8, wherein the artificial neural network is a Multilayer Perceptron.
 
10. A method according to claim 8 or claim 9, wherein the neural net is trained prior to installation and then instantiated in the distribution point (2), and wherein each analysis of the input data is a simple single iteration through the neural net.
 
11. A method according to claim 10, wherein after installation each neural network is allowed to mutate to allow optimisation for statistics generated by the individual subscriber connections (2) attached to the respective distribution point.
 
12. A method according to any of claims 7, 8, 9, 10 or 11 in which the connection (32) to the remote access server (42) is an optical fibre connection and the connections (30) to the individual termination points (2) are made by electrical means.
 


Ansprüche

1. Dynamisches Leitungsmanagement-System (19) zur Verarbeitung von Daten, die sich auf die Leistungsfähigkeit von jeder Digitalen Teilnehmeranschluss-Leitung einer Vielzahl von Digitalen Teilnehmeranschluss-Leitungen (30) beziehen, die jeweils einen entsprechenden Abschlusspunkt (2) eines Telekommunikations-Systems bedienen, wobei das dynamische Leitungsmanagement-System eine Einrichtung zum Erzeugen eines Profils von jeder Digitalen Teilnehmeranschluss-Leitung und zum Bestimmen eines Raten-Profils aufweist, um eine Kontrolle der Übertragung von Daten zu den individuellen Abschlusspunkten zu ermöglichen,
dadurch gekennzeichnet, dass das dynamische Leitungsmanagement-System einem Netzwerk-Verteilerpunkt (1) zugeordnet ist, der als ein Knoten-Zwischenglied zwischen einem Remote-Access-Server (42) und den entsprechenden Digitalen Teilnehmeranschluss-Leitungen (30) arbeitet, und der Netzwerk-Verteilerpunkt dazu eingerichtet ist, Daten an die individuellen Abschlusspunkte (2) unter der Kontrolle des zugeordneten dynamischen Leitungsmanagement-Systems zu übertragen.
 
2. Dynamisches Leitungsmanagement-System nach Anspruch 1, das dazu eingerichtet ist, das Raten-Profil zu dem Remote-Access-Server (42) zu übertragen.
 
3. Dynamisches Leitungsmanagement-System nach Anspruch 1 oder Anspruch 2, das ein künstliches neuronales Netz zur Verarbeitung von Daten aufweist, die sich auf die physikalische Schicht jeder Digitalen Teilnehmeranschluss-Leitung (30) beziehen.
 
4. Dynamisches Leitungsmanagement-System nach Anspruch 3, bei dem das künstliche neuronale Netz ein mehrlagiges Perzeptron ist.
 
5. Dynamisches Leitungsmanagement-System nach einem vorhergehenden Anspruch, bei dem der Netzwerk-Verteilerpunkt (1) durch ein optisches Fasersystem (32) mit dem Remote-Access-Server (42) verbunden ist und die Verbindungen (30) zu den individuellen Abschlusspunkten (2) durch elektrische Mittel erfolgen.
 
6. Verfahren zur Kontrolle der Übertragung von Daten zu individuellen Netzwerk-Abschlusspunkten (2), die von einem Remote-Access-Server (42) durch entsprechend Digitale Teilnehmeranschluss-Leitungen (30) durch einen Verteilerpunkt oder mehrere Verteilerpunkte (1) bedient werden, wobei Daten, die sich auf die Leistungsfähigkeit von jeder der Digitalen Teilnehmeranschluss-Leitungen beziehen, durch ein dynamisches Leitungsmanagement-System (19) verarbeitet werden, um ein Profil jeder Digitalen Teilnehmeranschluss-Leitung zu erzeugen und ein Raten-Profil zu bestimmen,
dadurch gekennzeichnet, dass die Profile der mit dem oder jedem Verteilerpunkt (2) verknüpften Digitalen Teilnehmeranschluss-Leitungen (30) durch ein dynamisches Leitungsmanagement-System (19) erzeugt werden, das dem entsprechenden Verteilerpunkt zugeordnet ist, und jeder Netzwerk-Verteilerpunkt (2) Daten an die individuellen Abschlusspunkte unter der Kontrolle seines zugeordneten dynamischen Leitungsmanagement-Systems (19) überträgt.
 
7. Verfahren nach Anspruch 6, bei dem ein Raten-Profil von dem dynamischen Leitungsmanagement-System (19), das dem oder jedem Verteilerpunkt (2) zugeordnet ist, an den Remote-Access-Server (42) übertragen wird.
 
8. Verfahren nach Anspruch 6 oder 7, bei dem das dynamische Leitungsmanagement-System (19) ein künstliches neuronales Netz zur Verarbeitung von Daten aufweist, die sich auf die physikalische Schicht einer Digitalen Teilnehmeranschluss-Leitung beziehen.
 
9. Verfahren nach Anspruch 8, bei dem das künstliche neuronale Netz ein mehrlagiges Perzeptron ist.
 
10. Verfahren nach Anspruch 8 oder 9, bei dem das neuronale Netz vor der Installation trainiert und dann in dem Verteilerpunkt (2) instantiiert wird, und bei dem jede Analyse der Eingangsdaten eine einfache Einzeliteration durch das neuronale Netz ist.
 
11. Verfahren nach Anspruch 10, bei dem nach der Installation jedem neuronalen Netz Mutation ermöglicht wird, um eine Optimierung für Statistiken zu ermöglichen, die durch die individuellen Teilnehmer-Verbindungen (2) erzeugt werden, die mit dem entsprechenden Verteilerpunkt verbunden sind.
 
12. Verfahren nach einem der Ansprüche 7, 8, 9, 10 oder 11, bei dem die Verbindung (32) mit dem Remote-Access-Server (42) eine optische Faser-Verbindung ist und die Verbindungen (30) zu den individuellen Abschlusspunkten (2) durch elektrische Mittel erfolgen.
 


Revendications

1. Système de gestion de ligne dynamique (19) pour traiter des données relatives aux capacités de chacune d'une pluralité de boucles d'abonnés numériques (30) desservant chacune un point de terminaison respectif (2) d'un système de télécommunications, le système de gestion de ligne dynamique comprenant un moyen pour générer un profil de chaque boucle d'abonné numérique et pour régler un profil de débit pour permettre la commande de la transmission de données aux points de terminaison individuels,
caractérisé en ce que le système de gestion de ligne dynamique est associé à un point de distribution de réseau (1) fonctionnant comme un noeud intermédiaire entre un serveur d'accès à distance (42) et les boucles d'abonnés numériques respectives (30), et le point de distribution de réseau est conçu pour transmettre des données aux points de terminaison individuels (2) sous la commande du système de gestion de ligne dynamique associé.
 
2. Système de gestion de ligne dynamique selon la revendication 1, conçu pour transmettre le profil de débit au serveur d'accès à distance (42).
 
3. Système de gestion de ligne dynamique selon la revendication 1 ou la revendication 2, comprenant un réseau neuronal artificiel pour traiter des données relatives à la couche physique de chaque boucle d'abonné numérique (30).
 
4. Système de gestion de ligne dynamique selon la revendication 3, dans lequel le réseau neuronal artificiel est un perceptron multicouche.
 
5. Système de gestion de ligne dynamique selon une quelconque revendication précédente, dans lequel le point de distribution de réseau (1) est connecté au serveur d'accès à distance (42) au moyen d'un système à fibre optique (32) et les connexions (30) aux points de terminaison individuels (2) sont réalisées par des moyens électriques.
 
6. Procédé de commande de la transmission de données à des terminaisons de réseau individuelles (2) desservies par un serveur d'accès à distance (42) par des boucles d'abonnés numériques respectives (30) par l'intermédiaire d'un ou de plusieurs points de distribution (1), les données relatives aux capacités de chacune des boucles d'abonnés numériques étant traitées par un système de gestion de ligne dynamique (19) pour générer un profil de chaque boucle d'abonné numérique et pour régler un profil de débit,
caractérisé en ce que les profils des boucles d'abonnés numériques (30) associées au ou à chaque point de distribution (2) sont générés par un système de gestion de ligne dynamique (19) associé au point de distribution respectif, et chaque point de distribution de réseau (2) transmet des données aux points de terminaison individuels sous la commande de son système de gestion de ligne dynamique associé (19).
 
7. Procédé selon la revendication 6, dans lequel un profil de débit est transmis du système de gestion de ligne dynamique (19) associé au, ou à chaque point de distribution (2) vers le serveur d'accès à distance (42).
 
8. Procédé selon la revendication 6 ou 7, dans lequel le système de gestion de ligne dynamique (19) comprend un réseau neuronal artificiel pour traiter des données relatives à la couche physique d'une boucle d'abonné numérique.
 
9. Procédé selon la revendication 8, dans lequel le réseau neuronal artificiel est un perceptron multicouche.
 
10. Procédé selon la revendication 8 ou la revendication 9, dans lequel le réseau neuronal est entraîné avant l'installation et ensuite instancié dans le point de distribution (2), et chaque analyse des données d'entrée étant une seule itération simple par l'intermédiaire du réseau neuronal.
 
11. Procédé selon la revendication 10, dans lequel après l'installation, chaque réseau neuronal est autorisé à muter pour permettre l'optimisation des statistiques générées par les connexions d'abonnés individuelles (2) fixées au point de distribution respectif.
 
12. Procédé selon l'une quelconque des revendications 7, 8, 9, 10 ou 11 dans lequel la connexion (32) au serveur d'accès à distance (42) est une connexion à fibre optique et les connexions (30) aux points de terminaison individuels (2) sont réalisées par des moyens électriques.
 




Drawing




















Cited references

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