WO2007003787A1

WO2007003787A1 - Method and system for increasing the rate capacity of at least one subscriber line

Info

Publication number: WO2007003787A1
Application number: PCT/FR2006/001559
Authority: WO
Inventors: Mohamed Tlich; Edmond Lassalle; Hubert Mariotte
Original assignee: France Telecom
Priority date: 2005-06-30
Filing date: 2006-06-29
Publication date: 2007-01-11
Also published as: FR2888063A1

Abstract

The invention concerns a method for allocating rate capacity associated with at least one subscriber line connecting a first client (11 to 18, 61 to 68) to a cross-connect device (110, 160), to at least one second client connected to the cross-connect device through a second subscriber line, each subscriber line connecting a client consisting of a first wire connection (LS1 to LS8, LS61 to LS68) connecting the cross-connect device to a secondary cross-connect device (120, 620) and of a second wire connection (LT1 to LT8, LT61 to LT68) connecting the secondary cross-connect device and the client. The method includes the following steps: determining a rate capacity difference between the first wire connection and the second wire connection of a common subscriber line and whether at least one subscriber line has a strictly positive difference and allocating part of the capacity rate of the subscriber line having a strictly positive difference to at least one subscriber line having a strictly negative difference. The invention also concerns the related system.

Description

Method and system for increasing the throughput capacity of at least one subscriber line

The present invention relates to a method and a system for increasing the throughput capacity of at least one subscriber line.

In conventional systems of access to the Internet or other services using linkages, for example of the DSL ₅ each client is connected to a digital multiplexer lines of customers via a distribution device in . DSL stands for Digital Subscriber Line. A digital multiplexer of customer lines is known as the DSLAM acronym for "Digital Subscriber Line Access Multiplexer".

Each customer is connected by a wired link to a distribution device located more or less close to his home. Depending on the length of the wired link connecting a client to the sub-distribution device, the throughput capacity of the wired link varies. The further the client is from the sub-distribution device, the less the throughput capacity of the wire link is important.

The throughput capacity of a wired link is the maximum bit rate per second that can be transmitted over the wired link for a given margin of the signal-to-wireline ratio. A sub-distribution device is a passive element that interconnects for each client for which it is responsible, the wire link connecting it to the client with the wired link that connects it to the customer line multiplexer.

More specifically, a sub-distribution device is connected to a splitter device included in the customer line multiplexer. The splitter device is commonly located in a central office.

Each client is thus connected to the distribution device by a wired connection, conventionally called a subscriber line, consisting of the wire link connecting it to the sub-distribution device and the wire link connecting the sub-distribution device to the distribution device. The longer the subscriber line, the lower the capacity of the subscriber line.

Customers have different throughput requirements. Some want to obtain a high flow, other less flow or zero because some do not have communication equipment type DSL. The capacity of the subscriber line sometimes limits the use of this line by the customer. Some customers can not get high throughput due to the throughput capacity of their subscriber line.

Attempts have been made to try to increase the throughput capacity of the subscriber line of these customers. These techniques are known as dynamic spectrum allocation. Decreasing the power spectral density level of the signal transmitted to one or more clients causes a decrease in noise mainly related to crosstalk on neighboring wire links, thereby increasing the throughput capacity of neighboring wire links.

These techniques are not satisfactory and do not allow an optimal use of the resources of the communication network.

The invention solves the disadvantages of the prior art by providing a method and a system that make it possible to use the resources of the telecommunication network optimally.

To this end, according to a first aspect, the invention proposes a rate capacity allocation method associated with at least a first subscriber line connecting a first client to a splitter device, to at least a second client connected to the device. dispatcher through a second subscriber line. A plurality of clients are connected to the splitter device respectively by a subscriber line, each subscriber line connecting a customer consisting of a first wired link connecting the distributing device to a sub-distribution device and a second wire link connecting the sub-distribution device and the customer. The process is remarkable in that it comprises the steps of:

determining, for at least part of the subscriber lines, a difference in throughput capacity between the first wired link and the second wired link of the same subscriber line,

- if at least one subscriber line has a strictly positive difference, allocating a portion of the subscriber line bit rate capability having a strictly positive difference to at least one subscriber line having a strictly negative difference.

Thus, it is possible to increase the amount of information transmitted to customers. By determining capacity differences between the wired links of each subscriber line, it is possible to know which wired links are underutilized and therefore to use them to increase the amount of information transmitted to other subscriber lines. clients.

According to another aspect of the invention, allocation of a portion of the subscriber line bit rate having a strictly positive difference to at least one subscriber line having a strictly negative difference is effected by allocating a part of the frequency spectrum of the first wired link of the subscriber line having a strictly positive difference, for the transfer of information to the connected customer by a subscriber line having a strictly negative difference.

Thus, the flow capacity allocation is performed in a simple manner. According to another aspect of the invention, a group of donors is formed comprising subscriber lines whose previously determined difference in flow capacity capacity is strictly positive and a group of recipients whose previously determined flow capacity difference is strictly negative. . According to this same aspect of the invention, a matrix of coefficients is formed such that each row of the matrix complies with the constraint that the sum of the values of the coefficients of the row of the matrix is equal to the absolute value of the difference in the matrix. pre-determined throughput capacity of a subscriber line of the donor group and each column of the matrix complies with the constraint that the sum of the values of the coefficients of the column of the matrix is equal to the absolute value of the difference in capacity previously determined flow rate of a subscriber line of the group of recipients. According to this same aspect of the invention, the distance is determined δjj separating the position of each coefficient ay in the diagonal diagonal matrix of the matrix. The distance δy separating the position of a coefficient in the matrix of the diagonal is given by S _{1 j} ≈ \ (j * Nlign) / Ncol -

where i is the row index and j is the column index of the ay coefficient, Nlign is the number of rows and Ncol is the number of columns in the matrix. According to this same aspect of the invention, the values of the coefficients of the matrix are modified to increase the value of the coefficients of the diagonal of the matrix or of the coefficients close to the coefficients of the diagonal of the matrix and to reduce the value of the other coefficients. of the matrix by respecting the preceding constraints. Thus, by increasing the value of the diagonal coefficients of the matrix or the coefficients close to the diagonal coefficients of the matrix and reducing the value of the other coefficients of the matrix, the frequency spectrum of a subscriber line is allocated to a limited number of customers.

According to another aspect of the invention, a list Lsup of index groups m and column n is formed of the coefficients a _m , i of the matrix situated above the diagonal of the matrix and having a non-zero value, where m is the index of the line of the coefficient a _mn and n is the index of the column of the coefficient a _mn and we form a list Linf of groups of indices line k and column 1 of coefficients ay of the matrix located below of the diagonal of the matrix and having a non-zero value, where k is the index of the line of the coefficient aki and 1 is the index of the column of the coefficient ajd ,. According to this aspect of the invention, as long as one of the lists is not empty, for each index group of the list Lsup, for each index group of the list Linf:

the sum of the distances separating the row m and column 1 index group and the row k and column n index group of the diagonal of the matrix and the sum of the distances separating the row m index group are compared and column n and the row index group k and column 1 of the diagonal in the matrix,

and if the sum of the distances separating the index group line m and column 1 and the group of indices line k and column n of the diagonal of the matrix is less than the sum of the distances that separate the group of indices row m and column n and the row index group k and column 1 of the diagonal in the matrix, we determine the minimum value of the coefficients a ^ and aki, we subtract the minimum value previously determined coefficients a _mn and aa, we add the minimum value determined to the coefficients a _m ι and update the list Lsup of the index groups line m and column n of the coefficients a _mn of the matrix located above the diagonal of the matrix and which have a non-zero value and update the list Linf of the index groups line k and column m of the coefficients% of the matrix located below the diagonal of the matrix and which have a non-zero value .

Thus, the capacity allocation of flow is carried out optimally. A subscriber line that is not used to the maximum of its throughput capacity is allocated to a limited number of customers avoiding multiple allocations.

According to another aspect of the invention, it is performed iteratively, as long as the value of a variable T initialiscc at a predetermined value is greater than a predetermined threshold: the determination of the number of coefficients of zero value of the matrix,

the permutation of two rows or two columns of the matrix to form a new matrix,

modifying the values of the coefficients of the new matrix in order to increase the value of the coefficients of the diagonal of the new matrix or of the coefficients close to the coefficients of the diagonal of the new matrix and to reduce the value of the other coefficients of the new matrix by respecting the constraints.

determining the number of zero value coefficients of the new matrix; selecting the new matrix as matrix if the number of zero value coefficients of the new matrix is greater than or equal to the number of zero value coefficients of the matrix; and decrease in value of the variable,

if the number of zero value coefficients of the new matrix is strictly less than the number of zero value coefficients of the matrix: determining a random value,

calculating the probability that the random value is less than a number depending on the ratio between the difference of the number of null coefficients of the matrices and the value of the variable,

the selection of the new matrix as matrix if the calculated probability is equal to the unit and the value decrease of the variable.

Thus, it is possible to minimize the effect related to the order in which the subscriber lines have been classified in the donor group and the receiver group.

According to another aspect of the invention, it is performed iteratively, as long as the value of a variable is greater than a predetermined threshold: determining the number of zero value coefficients of the matrix,

the selection of two rows or two columns of the matrix,

determining the number of non-zero coefficients of coefficients included in the selected rows or columns, if the number of non-zero coefficients of value determined is greater than one variable:

the permutation of the two rows or the two columns of the matrix to form a new matrix,

modifying the values of the coefficients of the new matrix in order to increase the value of the coefficients of the diagonal of the new matrix or of the coefficients close to the coefficients of the diagonal of the new matrix and to reduce the value of the other coefficients of the new matrix by respecting the constraints,

determining the number of zero value coefficients of the new matrix,

selecting the new matrix as a matrix if the number of zero value coefficients of the new matrix is greater than or equal to the number of zero value coefficients of the matrix and decreasing the value of the variable. Thus, it is possible to minimize the effect related to the order in which the subscriber lines have been classified in the donor group and the receiver group.

According to another aspect of the invention, the allocation of a portion of the rate capacity of the subscriber line having a strictly positive difference to at least one subscriber line having a strictly negative difference is effected by changing the level. power spectral density of the wired link connecting the splitter device to the sub-distribution device of the subscriber line having a strictly positive difference.

Thus, the means for rendering the throughput capacity of the wired links connecting the clients to the sub-distribution device independent of the capacity of the wired links connecting the under-distribution device to the distribution device do not need to redirect information received on a network. wired link from a subscriber line to the wired link of another subscriber line. The sub-distribution device is thus simpler to produce and of lower cost. According to another aspect of the invention, a donor group is formed comprising subscriber lines whose previously determined difference in flow capacity capacity is strictly positive and a group of recipients whose previously determined flow capacity difference is strictly negative. . According to this aspect of the invention, as long as one of the groups is not empty, it is also performed iteratively:

a decrease in the power spectral density level of the wired links of the subscriber lines included in the donor group and which connect the distribution device to the under-distribution device until the previously determined differences in throughput capacity are zero; ,

a determination, for each subscriber line, of the difference between the throughput capacity of the first wired link and the throughput capacity of the second wired link,

- The formation of a group of donors comprising subscriber lines whose previously determined difference in flow capacity is strictly positive and a group of recipients whose previously determined flow capacity difference is strictly negative.

According to another aspect of the invention, the decrease of the power spectral density level of the first wired links of the subscriber lines included in the group of donors until the previously determined differences in throughput capacity are zero, is performed by decreasing progressively or by dichotomy the power spectral density threshold of each first wired link of the subscriber lines included in the group of donors.

According to this aspect of the invention, one eliminates from a binary distribution table associated with each first wired link subscriber lines included in the group of donors, the bits whose portions of energies necessary for their transmission exceed said threshold. .

Correlatively, the invention relates to a flow capacity allocation system of at least a first subscriber line connecting a first customer to a distributor device to at least a second customer connected to the distributor device by a second subscriber line. , a plurality of clients being connected to the splitter device respectively by a subscriber line, each subscriber line connected to a customer consisting of a first wired link connecting the splitter device to a device sub distribution and a second wired link connecting the _'under split device and the client. The system is remarkable in that it comprises:

- means for making the second independent wired connections throughput capacity flow capacity of the first wire connections, ¹ - means for determining, for at least some of the subscriber lines, a flow capacity difference between the first wired link and the second wired link of the same subscriber line,

means for allocating a part of the bit rate capacity of the subscriber line having a strictly positive difference to at least one subscriber line having a strictly negative difference.

According to another aspect of the invention, the means for rendering the throughput capacity of the second wired links independent of the capacity of the first wired links are constituted, for each subscriber line, of an active element. An active element comprises means for determining the throughput capacity of the first wire link and the flow capacity of the second wire link, means for transferring the determined capacitances to a database, means for demodulating the signals received. of the first wired link, means for modulating the demodulated signals and means for transferring modulated signals on the second wired link. According to another aspect of the invention, each active element further comprises means for receiving data representative of frequency spectrum allocation of the first wired link, means for processing said representative frequency spectrum allocation data. the first wired link, means for receiving the signals included in the different parts of the frequency spectrum of the first wired link, means for demodulating the received signals to form demodulated information. According to this aspect of the invention, each active element further comprises means for transferring the demodulated information to the or each active element associated with the client receiving this information, information modulation means received from another active element and means for transferring modulated information to the second wired link.

The invention also relates to the computer program stored on an information carrier, said program comprising instructions for setting the previously described method, when it is loaded and executed by a computer system. /. '

The characteristics of the invention mentioned above, as well as others, will appear more clearly on reading the following description of an exemplary embodiment, said description being given in connection with the attached drawings, among which: Fig. 1 shows the architecture of the wireline rate capacity allocation system of a telecommunications network according to a first embodiment of the present invention; FIG. 2 represents an example of allocation of a part of the frequency spectrum of a wired link of a subscriber line to which a client is connected for the transmission of information intended for another customer; FIG. 3 is a block diagram of a flow capacity manager of the present invention; FIG. 4 represents the algorithm implemented by the flow capacity manager device for allocating wired link frequency spectrum portions of a telecommunications network according to a first embodiment of the present invention; FIG. 5a represents an algorithm implemented by the flow capacity management device, for an optimization of the algorithm of FIG. 4 according to a first embodiment; FIG. 5b represents a. algorithm implemented by the flow capacity management device, for an optimization of the algorithm of FIG. 4 according to a second embodiment; FIG. 6 shows the architecture of the rate capacity allocation system between several wired links of a telecommunication network according to a second embodiment of the present invention; FIG. 7 shows an illustration, according to the second embodiment, of an example of modification of the power spectral density level of the signals transmitted on a wired link of a subscriber line; FIG. 8 represents the algorithm implemented by the flow capacity management device for the allocation of flow capacity by modifying the spectral density of the signals transmitted on the wireline links of the subscriber lines according to a second embodiment of FIG. the present invention. Fig. 1 illustrates the architecture of the rate capacity allocation system of frequency linkage frequency portions of a telecommunications network according to a first embodiment of the present invention. The flow capacity allocation system according to the present invention consists of a flow capacity manager device 100, a distributor device 110 and a sub-distribution device 120 which allocates at least a portion of the spectrum of frequency of a subscriber line to a customer of another subscriber line.

The flow capacity manager device 100 and the dispatcher device 110 are included, for example, in a client line multiplexer that allows clients with the number 11 to 18 to access services. A client is, according to the invention, a communication device such as a DSL-type modem or a computer equipped with a DSL modem or a device for receiving television programs, video-on-demand via the network. Internet. Each customer 11 to 18 is connected to the distribution device 110 respectively by a subscriber line. Each subscriber line consists respectively of a wired link LT1 to LT8 connecting a customer 11 to 18 to the sub-distribution device 120 and a wired link LS 1 to LS8 connecting the sub-distribution device 120 to the distribution device 110. Only eight customers 11 to 18 are connected to the sub-distribution device 120 on the

Fig. 1, but it will be understood that the invention also encompasses the cases where a larger number of clients are connected to the sub-distribution device 120.

Similarly, a single sub-distribution device 120 is connected to the distribution device 110 in FIG. 1, but it will be understood that the invention also encompasses the cases where a larger number of sub-distribution devices 120 are connected to the distribution device 110.

According to the invention, the sub-distribution device 120 comprises active elements whose function is to separate the LT1 to LT8 wire links from the LS1 to LS8 wire links to make them independent of one another in terms of flow capacity. .

The sub-distribution device 120 comprises an active element for each subscriber line.

Each active element receives DSL type information from a wire link. For example, the active element connected to the wired link LS1 demodulates this information then the modules to retransmit on the corresponding wire link, in this case the LTl link. The reciprocal is also performed: '

Each active element comprises means for demodulating the signals received from the wire link connecting the distribution device to the sub-distribution device, means for modulating the demodulated signals and means for transferring the modulated signals on the link connecting the sub-distribution device to the customer of the subscriber line.

Each active element is able to carry out the demodulation steps of the signals received from the wired link connecting the distribution device to the device for sub-division, modulation of the demodulated signals and transfer of the modulated signals on the link connecting the device for distribution to the customer. of the subscriber line.

Each active element of each subscriber line further comprises means for determining the flow capacity of the wire link connecting it to the distribution device, and the capacity of the wire link connecting it to the customer of the line d. subscriber, a dialog module with a database 150 that allows the transfer of the calculated flow capacity.

Each active element further comprises means for receiving data from the flow capacity manager device 100 to receive representative data for allocating the frequency spectrum of the link Ic connecting to the distributor device 110. These data are the coefficients determined. by the scheme of FIG. 4 which will be described later.

Each active element further comprises means for receiving data representative of frequency spectrum allocation of the link connecting it to the splitter device, means for processing said representative frequency spectrum allocation data connecting it to the distributing device, means for receiving the signals included in the different parts of the frequency spectrum of the wired link connecting it to the distribution device, means for demodulating the received signals to form demodulated information, means for transferring the demodulated information to the or each active element associated with the client receiving this information, information modulation means received from another active element, means for transferring information modulated to the wire link connecting it to the client.

An active element makes it possible to make the throughput capacities of the wired links LT1 to LT8 connecting the clients 11 to 18 to the independent distribution device 120. the capacity of the wired links LS1 to LS8. An active element increases the throughput capacity of a subscriber line. In fact, if the subscriber line, for example the subscriber line consisting of wired links LS1 and LT1, is of the order of, for example, 5 km, the flow capacity of this line is order of 0.5 Mbps. By isolating the wired links LSI ₅ and LTI throughput capacity of the wire link LSI, long eg 2 km, is of the order of 2Mbit / s and the flow capacity of the wire link ₅ LTI long e.g. of 3 Ion, is of the order of 1 Mbits / s. The bit rate capacity of the subscriber line consisting of wired links LS1 and LT1 is then 1 Mbits / s. In this example, the throughput capacity of the subscriber lines is increased, but the wired link LS1 is underutilized in the sense that its throughput capacity is 2Mbits / s while only 1Mbit / s throughput capacity can be used. to be used.

According to the invention, the capacity management device 100 uses the unused rate capacity of a subscriber line to allocate it to at least one other customer, for example the customer 12, who is in a situation inverse, that is to say in the situation where the capacity of the LS2 wired link is 2 Mbits / s and the capacity of the LT2 wired link is 3 Mbits / s. More specifically, the capacity management device 100 is able to dynamically allocate part of the frequency spectrum of a wired link LS1 to LS8 of a subscriber line of a client for the transfer of information intended for minus another customer than the customer connected to the subscriber line.

The flow capacity manager device 100 is connected to a database 150 which comprises for each wired link LS1 to LS8, LT1 to LT8 the flow capacity thereof. The database 150 is, for example, updated by the operator in charge of the telecommunications network or by the device managing the bit rate capabilities.

100 from the rate capacities transmitted by the distribution device 110 or the sub-distribution device 120.

The flow capacity manager device 100, based on this information, determines an allocation of at least a portion of the frequency spectrum of each wired link LS1 to LS8 to the clients 11 to 18.

The flow capacity manager device 100 transfers the determined allocation to the distributor device 110 and the sub-distribution device 120. The distribution device 110 comprises, for each clear link LS1 to LS8, means for transferring information on at least part of the frequency spectrum of each wired link LS1 to LS8 as a function of the allocation determined by the capacity management device. The information transfer means transfers the information intended for at least two customers on the same subscriber line by allocating to these different parts of the frequency spectrum of the subscriber line in accordance with the information transmitted by the subscriber line. the flow capacity manager device 100.

The active elements of the sub-distribution device 120 are able to receive, on different parts of the frequency spectrum of the LS1 to LS8 wire links, information and to redirect them to the wired link T-T1 to LT8 corresponding to the client. to 18 recipient of this information.

The sub-distribution device 120 receives information from the flow capacity manager device 100, demodulates the received signals in different frequency spectrum parts of the LS1 and LS8 wire links, combines the information for the respective customers, modulates this information and transmits the information. in the wired links LT1 to LT8 respectively.

Fig. 2 represents an example of allocation of a part of the frequency spectrum of a wired link of a subscriber line to which a client is connected for the transmission of information intended for another customer.

Fig. 2 represents the frequency spectrum of two wired links LS1 and LS2.

On the abscissa axes are represented the frequencies noted f, on the ordinate axis are represented the numbers N of the bits transmitted by frequency or by carrier on the wired links LS1 and LS2.

According to the invention, the frequency spectrum of the wired link LS1 is divided into at least two parts denoted P1 and P2.

Only two parts P1 and P2 are shown in FIG. 2, but it will be understood that the invention also encompasses the case where the frequency spectrum is divided into a larger number of parts.

According to the example of FIG. 2, the subscriber line constituted by the wired links LS1 and LT1 has a capacity of less than that of the wired link LS1, the wired link LS2 has a throughput capacity less than the capacity of the wired link. LT2. The portion of the frequency spectrum of LS1 is allocated to the client 12 for the transmission of information to it. /;

Fig. 3 represents a block diagram of a flow capacity manager device according to the present invention. The architecture of the flow capacity manager device 100 is identical to the architecture of the flow capacity manager device 600 of FIG. 6. Only the architecture of the flow capacity manager device 100 is described.

The flow capacity manager device 100 comprises a communication bus 301 to which are connected a central unit 300, a non-volatile memory 302, a random access memory 303, a communication interface 307 with the devices 110 and 120 and a database interface. 306.

The non-volatile memory 302 stores the programs implementing the invention such as the algorithms which will be described later with reference to the

Figs. 4, 5a or 5b. The non-volatile memory 302 is, for example, a hard disk. More generally, the programs according to the present invention are stored in storage means. This storage means is readable by a computer or a microprocessor 300. This storage means is integrated or not in the flow capacity manager device 100, and can be removable. When the flow capacity manager device 100 is powered up, the programs are transferred to the random access memory 303 which then contains the executable code of the invention as well as the data necessary for the implementation of the invention.

The flow capacity manager device 100 includes a communication interface 307 with the distributor device 110 and the sub-distribution device 120 according to the first embodiment of the present invention. It should be noted here that, according to the second embodiment, the flow capacity management device 600 includes a communication interface with the splitter device 610.

This communication interface 307 allows the transfer of information representative of the allocation of frequency spectrum portions of several LS1 to LS8 wire links according to the first embodiment.

According to the second embodiment of the present invention, the communication interface of the flow capacity manager device 600 allows the transfer of information representative of the power spectral density levels to be applied on the LS61 to LS68 wired links. The database interface 306 allows the exchange of information between the flow capacity manager device 100 and the database 150.

Fig. 4 represents the algorithm implemented by the flow capacity manager device, for the allocation of frequency band portions of wire links of a telecommunication network according to a first embodiment of the present invention.

In step E400, the processor 300 of the flow rate manager device 100, obtains the flow capacity of the LS1 to LS8, LT1 to LT8 wire links stored in the database 150. For example, the LS1 to LS8 wire links each have a throughput capacity of 4 Mbps. The wired link LT1 has a throughput capacity of 6 Mbits / s, the wired link LT2 has a capacity of 1 Mbits / s, the wired link LT3 has a capacity of 0 Mbits / s, the wired link LT4 has a capacity of 2 Mbits / s, the wired link LT5 has a capacity of 8 Mbits / s, the wired link LT6 has a capacity of 7 Mbits / s, the wired link LT7 has a throughput capacity of 3 Mbps, the LT8 wired link has a throughput capacity of 5 Mbps.

At the following step E401, the processor 300 determines, for each subscriber line, the remaining rate capability denoted ₅ Ci where i is an index representative of the subscriber line. The remaining throughput capacity of a subscriber line i is the difference between the throughput capacity of the wired link LSi of the line i, where i varies from 1 to 8, and the throughput capacity of the wired link LTi of the line i.

According to our example Cl = -2Mbits / s, C2 = + 3Mbits / s, C3 = + 4Mbits / s, C4 = + 2Mbits / s _s C5 = -4Mbits / s, C6 = -3Mbits / s, C7 = + lMbits / s and C8 = -IMbits / s.

In the next step E402, the processor 300 forms groups of donors and recipients. The donor group includes all subscriber lines that have a remaining positive bit capacity Ci. The group of receivers comprises all the subscriber lines which have a capacity of remaining flow Ci strictly negative. In our example, the donor group comprises in the following order LS2, LS3, LS4, LS7. The group of recipients comprises LS1, LS5, LS6 and LS8.

In step E403, the processor 300 forms a matrix M. Each row of the matrix M corresponds to a subscriber line of the group of donors. Each column of the matrix M corresponds to a subscriber line of the group of recipients. The matrix M is, according to our example, a matrix of dimension 4 * 4 which fills the two following constraints: for each line of the matrix M ₃ the sum of the values of the coefficients of the row of the matrix M is equal to the absolute value of the remaining capacity of flow of a subscriber line; for each column of the matrix M, the sum of the values of the coefficients of the column of the matrix M is equal to the absolute value of the remaining capacity of flow of a subscriber line. Lines and columns are ordered according to their respective order in the donor and recipient groups.

The processor 300 determines coefficient values that fulfill these two constraints. For example, the processor 300 determines, according to our example, the matrix M

[750 1250 750 250 ^'

500 1750 1500 250 next: M =

500 750 500 250

250 250 250 250_

The matrix M consists of coefficients a, _j . Each coefficient has _j a value and a position in the matrix M.

Each coefficient value is representative of a rate capacity, expressed in kilobits per second, of a subscriber line of the donor group allocated to a subscriber line of the receiver group. The position of each coefficient of the matrix is defined by the index of the row and column to which the coefficient belongs.

Thus, a coefficient a, _j is located in the matrix M, at the intersection of the i-th row of the matrix M and the jth column of the matrix M. The row and column index group i J thus defines the position of each coefficient in the matrix M.

In the next step E404, the processor 300 forms two lists called Lsup and Linf. The list Lsup comprises the set of row and column index groups of the non-zero coefficients situated in the part of the matrix M strictly greater than the diagonal of the matrix. The list Linf comprises the set of row and column index groups of the non-zero coefficients situated in the part of the matrix M strictly smaller than the diagonal of the matrix M.

For a square matrix, the matrix coefficients M located on the diagonal

Δ of the matrix M are the coefficients whose row index is identical to the column index. When the matrix is rectangular, the matrix coefficients M located on the diagonal Δ of the matrix M are the coefficients whose group ^" of indices satisfies the following equation:

A = {(PU * Nid 1 Ncol), y); l ≤ j ≤ Ncol} where E (x) denotes the integer part of x, i represents the index of the line and j the index of the column, Ncol is the number of columns of the matrix M and Nlign the number of rows of the matrix M.

In our example, Lsup includes the row and column index group 1 and

2, the row and column index group 1 and 3, the row and column index group 1 and 4, the row and column index group 2 and 3, the row and column index group 2 and 4, the group of row and column indices 3 and 4 whose coefficients have the following values: a ₁₂ = 1250, aπ = 750. a ₁₄ = 250, a ₂₃ = 1500, a ₂₄ = 250 and a ₃₄ = 250. Linf includes the row and column index group 2 and 1, the row and column index group 3 and 1, the row and column index group 3 and 2, the row and column index group 4 and ₅ the group of row and column indices 4 and 2 and the group of row and column indices 4 and 3 whose coefficients have the following values: a ₂₁ = 500, a ₃ r =

500, at ₃₂ = 750, ₄₁ = 250, at ₄₂ = 250 and at ₄₃ = 250.

In the next step E405, the processor 300 considers the first group of indices of Lsup, that is to say the group of indices 1 and 2. Subsequently, the line index of a group of indices included in Lsup is called m and the column index of a group of indices included in LM is called n. It is the same for the row and column indices of the coefficients corresponding to these groups of indices.

In the next step E406, the processor 300 sets the variable Modif equal to the null value.

In the next step, E407, the processor 300 considers the first index group Linf, that is to say the index group 2 and 1. Subsequently, the row index of a group of The indices included in Lsup are called k and the column index of a group of indices included in Linf is called 1. The same is true for the row and column indices of the coefficients corresponding to these groups of indices.

In the next step E408, the processor 300 calculates δ1 according to the following formula:

where δ _m> i is the distance separating the position of the line index coefficient m and column 1 from the diagonal of the matrix M and δk, n is the distance separating the position of the index coefficient k and n from the diagonal of the matrix M. The distance separating the position of a coefficient ai _j from the diagonal of the matrix M is obtained from the group of indices line i and column j of the coefficient ay and is given by the following formula:

This distance is calculated for each coefficient ay of the matrix M.

Once this is done, the processor 300 proceeds to step E409 and calculates δ2 according to the following formula: δ 2 = δ _m , _n + γy

In the next step E410, the processor 300 checks whether δ1 is strictly less than 52. In our example, δ1 = 0; 52 = 2. If not, processor 300 proceeds to step E414. If so, processor 300 proceeds to step E411.

At step E411 ₅ the processor 300 determines the rated minimum value Min of the values of the coefficients a _mn and a ^. According to our example Min = 500.

In the next step E412, the processor 300 modifies the values of the following matrix coefficients M: a _mn = a _mn -Min; a _k i = a _kr Miu; a _m pa _m | ₊ Min; a _kn = a _kn + Min. According to our example, the matrix M then becomes:

It should be noted here that the modification of the values of the coefficients preserves the following constraints: for each row of the matrix M ₅ the sum of the values of the coefficients of the row of the matrix M is equal to the absolute value of the remaining capacity of flow the corresponding subscriber line; for each column of the matrix M, the sum of the values of the coefficients of the column of the matrix M is equal to the absolute value of the remaining flow capacity of the corresponding subscriber line.

At this same stage, the processor 300 sets the variable Modif to the value 1. The variable Modif is representative of the modification of the value of a coefficient.

Once this is done, the processor 300 proceeds to the next step E413 which consists in updating the Lsup and Linf lists. The lists Lsup and Linf are updated by removing from these groups of row and column indices of the coefficients whose value has become zero following step E412. The Lsup list is also set day by adding therein groups of indices row and column of coefficients that are above the diagonal of the matrix, the value of ^'the coefficient became strictly positive following step E412. The list Linf is also updated by adding therein the groups of row and column indices of the coefficients situated below the diagonal of the matrix and whose coefficient value has become strictly positive following the step E412.

When a coefficient has a zero value, this implies that no part of the frequency spectrum of the subscriber line corresponding to the row of the matrix comprising the null coefficient is allocated to the subscriber line corresponding to the column of the matrix comprising the null coefficient.

In our example, clusters 2 and 1 are removed from the Linf list.

When this is done, the processor 300 proceeds to step E414 and checks whether there is another group of indices in Linf that has not been processed.

If so, the processor 300 proceeds to step E415 which consists in considering the next set of indices of the list Linf, in this case the index group 3.1 according to our example.

The processor 300 executes the loop consisting of the steps E408 to E415 until all the index groups of the list Linf have been processed.

When all index groups of the Linf list have been processed, the processor 300 proceeds from step E414 to step E416.

At this stage, the processor 300 checks whether there is another group of indices in Lsup that has not been processed.

If so, the processor 300 proceeds to step E417 which consists in considering the next group of indices of the list Lsup. The processor 300 executes the loop consisting of steps E407 to E417 as long as all the index groups of the list Lsup have not been processed.

When all the index groups of the Lsup list have been processed, the processor 300 proceeds from step E416 to step E418.

In step E418, the processor 300 checks whether the variable Modif is equal to 1. If so, the processor 300 returns to step E405 and executes the loop consisting of steps E405 to E418.

If not, the processor 300 proceeds to the next step E419.

According to our example, when all index groups of Lsup have been processed in accordance with the present algorithm, the matrix M is as follows: 2000 1000 0 0

0 3000 1000 0

M =

0 0 2000 0

0 0 0 1000

The steps E405 to E418 serve to diagonalize the matrix M to obtain on or around the diagonal Δ high value coefficients and to have coefficients of zero value in the remainder of the matrix and thus to optimize the number of spectrum parts allocated to other subscriber lines.

In other words, the values of the coefficients of the matrix M are modified so as to increase the value of the coefficients of the diagonal of the matrix or of the coefficients close to the coefficients of the diagonal of the matrix and to reduce the value of the other coefficients. of the matrix.

We mean by near coefficient, a coefficient that has the same line index as a diagonal matrix coefficient and a column index greater or less than one or two units, or even more if the matrix M is large, at the column index of the coefficient of the diagonal of the matrix. The value of a coefficient is representative of a rate capacity of a subscriber line associated with the row of the matrix M allocated to another subscriber line associated with the column of the matrix M.

The present algorithm thus makes it possible to allocate the available throughput capacities to applicants by making the least number of possible cuts. In step E419, the coefficients are transferred to the distributor device 110 and to the sub-distribution device 120.

Fig. 5a represents an algorithm implemented by the flow capacity management device, for an optimization of the algorithm of FIG. 4 according to a first embodiment. When the coefficients of the matrix M have been determined according to the algorithm of FIG. 4, the processor 300 prior to the transfer of the coefficients in step E419 of the algorithm of FIG. 4 performs the present algorithm.

The present algorithm minimizes the effect of the order in which the subscriber lines have been classified in the donor group and the receiver group. The principle of the present algorithm generally aims to promote the reduction of the number of coefficients of zero value in the miatrice M while not systematically prohibiting an increase in the number of coefficients of zero value. In step E500, the processor 300 initializes the present algorithm and sets the counter variable to zero.

In the next step E501, the processor 300 reads the values of the temperature variables T, PasCompleur, of the threshold ε, the matrix M determined according to the algorithm of FIG. 4 and determines the number Z of null coefficients of the matrix M. The variable T is called temperature by analogy with so-called annealing methods.

An example of an annealing method is described in the document by JM Renders, entitled Genetic Algorithms and Neural Networks and published by Hermès, Paris 1995. The temperature variable T is initialized to a predetermined value such as, for example, a thousand value. and decremented by half at each iteration of the present algorithm. In general, T is chosen so as to favor, initially, the selection of matrices even if they include a smaller number of coefficients of zero value than the previously selected matrix.

In step E502, the processor 300 randomly draws the Lincol variable. The value of Lincol is equal to one or zero depending on the random draw.

In step E503, the processor 300 determines whether the value of the variable LinCol is zero.

If so, the processor 300 proceeds to the next step E504. If not, processor 300 proceeds to step E506. In step E504, the processor 300 randomly selects two columns of the matrix M.

In the next step E505, the processor 300 swaps the two selected columns. According to the example used in the description of FIG. 4, columns 3 and 1 are swapped. The new matrix M 'is then equal to:

0 1000 2000 0

1000 3000 0 0

M '=

2000 0 0 0

0 0 0 1000 At the following step E508, the processor 300 performs ^one of a new diagonalization of the matrix M 'as was _previously. .described with reference to steps E404 to E418 of the algorithm of FIG. 4.

According to our example, the processor 300 obtains the new matrix M ':

In step E506, the processor 300 randomly selects two rows of the matrix M and switches them to step E507 to form a matrix M '. This operation performed, the processor 300 goes from step E507 to step

E508 previously described.

The processor 300 moves from the step E508 to the step E509 and determines the number Z 'of coefficients of zero value included in the matrix M'.

In the next step E510, the processor 300 checks whether the number Z 'of coefficients of zero value included in the matrix M' is greater than or equal to the number Z of coefficients of zero value included in the matrix M.

In the affirmative, the processor 300 proceeds to the step E511 in which the matrix M 'is selected as the new matrix. In the next step E512, the variable Z is set to the value of the variable Z '. Once this is done, the processor 300 proceeds to step E518.

If the test of step E510 is negative, processor 300 proceeds to step E513 and determines a random value X between zero and unity.

In the next step E514, the processor 300 calculates a probability p according to the following formula:

In the next step E515, the processor 300 checks whether the probability p is equal to the value 1.

If not, processor 300 proceeds to step E518. If so, processor 300 proceeds to step E516. Matrix M 'is selected as the new matrix. In the next step E517, the variable Z is set to the value of the variable Z '. Once this is done, the processor 300 proceeds to the next step E518. In step E518, the processor 300 increments the counter variable by one.

In the next step E519, the processor 300 checks whether the counter variable is equal to PasComplete.

If not, the processor 300 returns to step E502 and executes the loop consisting of steps E502 to E519.

If so, the processor 300 proceeds to step E520 and decreases the value of T.

In the next step E521, the processor 300 checks whether T is less than ε. If not, processor 300 returns to step E502 and executes the loop consisting of steps E502 through E521.

In affirmative, the processor 300 stops the present algorithm. The coefficients of matrix M 'are then transferred according to step E419 of the algorithm of FIG. 4.

Fig. 5b represents an algorithm implemented by the flow capacity management device, for an optimization of the algorithm of FIG. 4 according to a second variant embodiment.

When the coefficients of the matrix M have been determined according to the algorithm of FIG. 4, the processor 300, prior to the transfer of the coefficients in step E419 of the algorithm of FIG. 4, performs the present algorithm. The present algorithm minimizes the effect of the order in which the subscriber lines have been classified in the donor group and the receiver group.

In step E550, the processor 300 initializes the present algorithm and sets the counter variable to zero. In the next step E551, the processor 300 reads the values of the variables Step,

Step Counter, of the threshold Step, the matrix M determined according to the algorithm of FIG. 4 and determines the number Z of zero coefficients of the matrix M.

The variables Pas, PasComputer, of the threshold Pas are defined according to the size of the matrix. For example and without limitation, the value of Step is ten, the value of Step Counter is equal to one hundred and the value of the threshold Step is equal to two.

In step E552, the processor 300 proceeds to a random draw of the Lincol variable. The value of Lincol is equal to one or zero depending on the random draw. In step E553, the processor 300 determines whether the value of the Lincol Variable is zero.

If so, the processor 300 proceeds to the next step E554. If not, processor 300 proceeds to step E558. In step E554, the processor 300 randomly selects two columns of the matrix M.

In step E555, the processor 300 determines the number D1 of non-zero value coefficients included in the selected columns.

In step E556, the processor 300 checks whether the number D1 of non-zero value coefficients included in the selected columns is greater than or equal to the value of the variable Step.

If not, processor 300 proceeds to step E568. If so, the processor 300 proceeds to step E557 and switches the two selected columns to form the matrix M '. When this is done, processor 300 proceeds to step E562.

In step E558, processor 300 randomly selects two rows of array M.

In step E559, the processor 300 determines the number D2 of non-zero value coefficients included in the selected lines. In step E560, the processor 300 checks whether the number D2 of non-zero value coefficients included in the selected lines is greater than or equal to the value of the variable Step.

If not, processor 300 proceeds to step E568.

If so, the processor 300 proceeds to step E561 and switches the two selected lines to form the matrix M '. When this is done, processor 300 proceeds to step E562.

In step E562, the processor 300 again diagonizes the matrix M ⁵ as previously described with reference to the steps E404 to E418 of the algorithm of FIG. 4. The processor 300 moves from step E562 to step E563 and determines the number

Z 'of coefficients of zero value included in the matrix M \

In the next step E564, the processor 300 checks whether the number Z 'of coefficients of zero value included in the matrix M' is greater than or equal to the number Z of coefficients of zero value included in the matrix M. In the affirmative case, processor 300 proceeds to step E565. Matrix M 'is selected as the new matrix. In the next step E566, the variable Z is set to the value of the variable Z '. When this is done, the processor 300 sets the counter variable to zero in step E567 and returns to step E552 previously described.

If the test of step E564 is negative, processor 300 proceeds to step E568 and increments the counter variable by one.

When this operation is performed, the processor 300 checks in step E569 whether the counter variable is equal to PasComputer. If not, processor 300 returns to step E552.

If so, the processor 300 proceeds to step E570 and decrements the step byte of a unit.

In step E571, the processor 300 checks whether the variable Pas is equal to Threshold Step.

If not, processor 300 returns to step E552. If so, the processor 300 stops the present algorithm. The coefficients of matrix M 'are then transferred according to step E419 of the algorithm of FIG. 4.

Fig. 6 shows the architecture of the rate capacity allocation system between several wired links of a telecommunication network according to a second embodiment of the present invention.

The flow capacity allocation system according to the second embodiment of the present invention consists of a flow capacity management device 600, a distribution device 610 and a sub-distribution device 620. The device Flow Capacity Manager 600 and Dispatcher

610 are included, for example, in a client line multiplexer which allows clients rated 61 to 68 to access services via DSL modems.

Each customer 61 to 68 is connected to the splitter device 610 respectively by a subscriber line. Each subscriber line consists respectively of a wired link LT61 to LT68 connecting a client 61 to 68 to the sub-distribution device

620 and a wired link LS61 to LS68 connecting the sub-distribution device 620 to the splitter device 610. Only eight clients 61 to 68 are connected to the sub-distribution device 620 in FIG. 1, but it should be understood that the invention also encompasses the cases where a larger number of clients are connected to the sub-distribution device 620.

Likewise, a single sub-distribution device 620 is connected to the distribution device 610 in FIG. 6, but it is to be understood that the invention also encompasses the cases where a larger number of sub-distribution devices 620 are connected to the distribution device 610.

According to the second embodiment of the invention, the sub-distribution device 620 comprises active elements whose function is to isolate the wired links LT61 to LT68 from the wired links LS61 to LS68 so as to make the latter independent of each other. others in terms of flow capacity.

The sub-distribution device 620 includes an active element for each subscriber line.

Each active element receives DSL type information from a wire link. For example, the active element connected to the wired link LS61 demodulates this information and then the modules for retransmitting them on the corresponding wired link, in this case the wired link LT61. The reciprocal is also performed. An active element thus makes it possible to increase the throughput capacity of a subscriber line. Each active element is able to evaluate the throughput capacity of the wired links it interconnects and to transfer these throughput capacities to the splitter device 610 which then transfers this information to the flow capacity manager device 600.

The flow capacity management device 600 is connected to a database 650 which comprises, for each wired link LS61 to LS68, LT61 to LT68, the bit rate capacity thereof and the binary distribution table comprising the number of transmittable bits. frequency or carrier and the portion of energy associated with each bit. An example of a binary distribution table comprising the number of bits transmitted by frequency or by carrier and the portion of energy associated with each bit is explained with reference to FIG. 7.

The database 650 is, for example, updated by the operator in charge of the telecommunications network or by the management device of the capacity of flow 600 from the capacity of flow transmitted by the distributor device 610 or by the device of under distribution 620.

The flow capacity management device 600, based on this information, decreases the power spectral density level of the signals transrais on fiilaires bonds LS61 to LS68 subscriber lines ¹ - whose ^"difference between the flow capacity of the wired link connecting the distributor device sub distribution device and the flow capacity of the wired link connecting the device under distribution and the customer is strictly positive, so as to increase the transmission rate capacity of the LS61 to LS68 wired links of the subscriber lines, the difference between the capacity of the wired link connecting the splitter device to the sub-device. The flow rate management device 600 transfers the power spectral density (DSP) levels of the determined signals to the splitter device 610 and the flow capacity of the wired link connecting the sub-distribution device and the client is strictly negative. The distribution device 610 comprises, for each wired link LS61 to LS68 , means for modifying the power spectral density level of the signals transmitted on the wired links LS61 to LS 68 as a function of the information received from the capacity management device 600.

The modification of the power spectral density level of the signals transmitted on the wired links LS61 to 68 is carried out by gradually decreasing the levels of the power spectral densities or by dichotomy. The decrease in power spectral density can also be performed, as described later with reference to FIG. 7, progressively or by dichotomy.

Fig. 7 shows an illustration, according to the second embodiment, of the modification of the power spectral density level of the signals transmitted on a wired link of a subscriber line.

Fig. 7 represents different energy levels associated with the bits transmitted on a carrier, for example the wired link LS61.

On the abscissa axes are represented the frequencies noted f, on the ordinate axis is represented the sum A of each energy level associated with each bit transmitted on a carrier frequency of the wired link LS61.

This energy distribution is divided into two parts Bl and B2. The portion B1 corresponds to a portion of energy associated with bits, greater than a power spectral density threshold S1.

According to this embodiment, when the level of power spectral density is decreased, the bits whose portion are removed from the binary distribution table are eliminated. of energy is in the B1 portion which exceeds the power spectral density threshold S1.

The power spectral density threshold S1 is progressively or dichotomously reduced as long as the difference between the flow capacity of the LS61 link connecting the distribution device to the underloading device and the flow capacity of the connecting LT61 link. the device of under distribution and the customer is not null.

Fig. 8 represents the algorithm implemented by the flow capacity management device, for the allocation of flow capacity by power spectral density level modification of the signals transmitted on the wired links of the subscriber lines according to a second mode. embodiment of the present invention.

In step E800, the processor 300 of the flow capacity manager device 600 obtains the throughput capacities of the wired links LS61 to LS68, LT61 to

LT68 stored in the database 650 and the binary distribution tables comprising the number of bits transmittable by frequency or carrier and the portion of energy associated with each bit.

In the next step E801, the processor 300 determines, for each subscriber line, the remaining rate capacity C6i, where i is an index representative of the subscriber line, i varying from 1 to 8. The capacity of The remaining bit rate of a line 6i is the difference between the bit rate capacity of the LS6i wire link of the subscriber line 6i and the bit rate capacity of the LT6i wire link of the subscriber line 6i.

In the next step E802, the processor 300 forms groups of donors and recipients. The donor group includes all the subscriber lines that have a remaining positive throughput capacity C6i. The receiver group includes all the subscriber lines that have a remaining negative rate capacity C6i.

In the next step E803, the processor 300 decreases the power spectral density level of the signals transmitted on the LS6i wire links included in the donor group until the remaining flow capacitances C6i of the donor groups are zero.

The modification of the power spectral density level of the signals transmitted on the LS6i wired links included in the group of donors is carried out by progressively decreasing the levels of the power spectral densities or by dichotomy as long as the difference between the flow capacity of each wired link LS6i, of the group of donors, connecting the dispatcher device 610 to the sub-distribution device 620 and the capacity of the wire link LT6i connecting the sub-distribution device 620 and the client is not zero.

The reduction of the power spectral density can also be performed, as previously described, progressively or by dichotomy.

Once this is done, the processor 300 proceeds to step E804, which consists in obtaining the new bit rate capacities of the LS61 to LS68, LT61 to

LT68 stored in the database 150. These new flow capacities were transmitted by the active elements of the sub-distribution device 620 following step E804.

Indeed, following the modification of the spectral densities carried out in step E804, the throughput capacities of the various wired links have been modified.

In the next step E805, the processor 300 determines, for each subscriber line, the remaining flow capacity denoted C6i in the same manner as that described in step E80L

In the next step E806, the processor 300 forms groups of donors and recipients in the same manner as described in step E801.

In the next step E807, the processor 300 checks whether the group of recipients is empty. If so, the processor 300 stops the present algorithm. If not, processor 300 proceeds to step E808.

At this stage, the processor 300 checks whether the donor group is empty.

As long as the donor group is not empty, or as long as there is a remaining flow capacity C6i strictly greater than zero, the processor 300 repeats the steps E803 to E808 of the present algorithm. When the donor group is empty, the processor 300 stops the present algorithm.

It should be noted here that, in an alternative embodiment, it is determined whether among the clients connected to the sub-distribution device 120 or 620, there is at least one customer who is not a customer of the operator of the telecommunications network. If yes, this algorithm is only performed for subscriber lines connecting the customers of the telecommunications network operator. The information of the wired links connecting the customers of the operator of the telecommunication network to the sub-distribution device are transmitted in a frequency band different from that used for subscriber lines connecting customers who are not customers of the telecommunication network operator.

Of course, the present invention is not limited to the embodiments described herein, but encompasses, on the contrary, any variant within the scope of those skilled in the art.

Claims

1) Flow capacity allocation method associated with at least a first subscriber line connecting a first client (11 to 18, 61 to 68) to a splitter device (110, 610), to at least a second connected client to the splitter device by a second subscriber line, a plurality of clients being connected to the splitter device respectively by a subscriber line, each subscriber line connecting a customer consisting of a first wired link (LS1 to LS8, LS61 to LS68) connecting the distributor device to a sub-distribution device (120, 620) and a second wire link (LT1 to LTS, LT61 to LT68) connecting the sub-distribution device and the customer, characterized in that the method comprises the steps of: - determining (E401, E801) for at least a portion of the subscriber lines, a difference in throughput capacity between the first wired link and the second wired link of the same subscriber line,

- if at least one subscriber line has a strictly positive difference, allocating (E412, E803) a portion of the subscriber line bit rate having a strictly positive difference to at least one subscriber line having a strictly negative difference.

2) The method according to claim 1, wherein the allocation of a portion of the rate capacity of the subscriber line having a strictly positive difference to at least one subscriber line having a strictly negative difference is performed (E401 ) by allocating a portion of the frequency spectrum of the first wired link of the subscriber line having a strictly positive difference, for the transfer of information to the connected client by a subscriber line having a strictly negative difference.

3) Method according to claim 2, further comprising the steps of:

training (E402) of a group of donors comprising the subscriber lines whose previously determined difference in throughput capacity is strictly positive and of a group of receivers whose difference in flow capacity previously determined is strictly negative,

training (E403) of a matrix of coefficients such that each row of the matrix complies with the constraint that the sum of the values of the coefficients of the row of the matrix is equal to the absolute value of the previously determined flow capacity difference of a subscriber line of the donor group and each column of the matrix respects the constraint that the sum of the values of the coefficients of the column of the matrix is equal to the absolute value of the previously determined flow capacity difference of a subscriber line of the receiver group,

determination (E408, E409) of the distance δjj separating the position of each coefficient ay in the matrix of the diagonal of the matrix, the distance δ, j separating the position of a coefficient in the matrix of the diagonal being given by δ _j

Nligri) / Ncol- i) \ where i is the line index and j is the column index of the coefficient ay, Nlign is the number of rows and Ncol is the number of columns in the matrix.

modifying (E412) the values of the coefficients of the matrix to increase the value of the coefficients of the diagonal of the matrix or coefficients close to the coefficients of the diagonal of the matrix and to reduce the value of the other coefficients of the matrix by respecting the previous constraints.

4) The method of claim 3, further comprising the steps of:

training (E404) of a list Lsup of index groups row m and column n of the coefficients _mn of the matrix situated above the diagonal of the matrix and having a non-zero value, where m is the index of the line of the coefficient a _mn and n is the index of the column of the coefficient a _mn ,

training (E404) of a list Linf of row k and column 1 index groups of coefficients a ^ of the matrix located below the diagonal of the matrix and having a non-zero value, where k is the index of the line of the coefficient a ^ and 1 is the index of the column of the coefficient a _k i, and as long as one of the lists is not empty: for each group of indices of the list Lsup, for each group of d Linf index indices - comparison (E410) of the sum of the distances separating the row m and column 1 index group from the row k and column n index group of the matrix diagonal and the sum of the distances separating the group of indices line m and column n and the group of indices line k and column 1 of the diagonal in the matrix, and if the sum of the distances between the row m and column I index group and the line k and n column index group of the matrix diagonal is less than the sum of the distances separating the row index group m and column n and the row index group k and column 1 of the diagonal in the matrix, - determining (E411) the minimum value of the coefficients a _mn and a _k i,

subtracting (E412) the minimum value previously determined from the coefficients a _mn and ai _d and adding the minimum value determined to the coefficients a _m i and aiaπ,

updating the list Lsup of the groups of indices line m and column n of the coefficients _mn of the matrix situated above the diagonal of the matrix and which have a non-zero value,

updating of the list Linf of the row k and column m index groups of the matrix aw coefficients located below the diagonal of the matrix and which have a non-zero value.

5) Method according to one of claims 3 and 4, further comprising the steps, performed iteratively as the value of a variable T ₅ initialized to a predetermined value, is greater than a predetermined threshold, of:

determination (E501) of the number of zero value coefficients of the matrix; permutation (E505, E507) of two rows or two columns of the matrix to form a new matrix;

modifying (E508) the values of the coefficients of the new matrix in order to increase the value of the coefficients of the diagonal of the new matrix or of the coefficients close to the coefficients of the diagonal of the new matrix and to reduce the value of the other coefficients of the new matrix respecting the constraints.

determination (E509) of the number of coefficients of zero value of the new matrix,

selecting (E511) the new matrix as a matrix if the number of zero value coefficients of the new matrix is greater than or equal to the number of coefficients of zero value of the matrix and decreasing the value of the variable,

if the number of zero value coefficients of the new matrix is strictly less than the number of zero value coefficients of the matrix:

determination (E513) of a random value, calculating (E514) the probability that the random value is less than a number depending on the ratio between the difference of the number of null coefficients of the matrices and the value of the variable,

selecting (E516) the new matrix as a matrix if the calculated probability is equal to the unit and decreasing the value of the variable T.

6) Method according to one of claims 3 and 4, further comprising the steps, performed iteratively as long as the value of a variable is greater than a predetermined threshold, of: - determination (E551) of the number of coefficients of null value of the matrix,

selection (E554, E558) of two rows or two columns of the matrix,

determination (E555, E559) of the number of non-zero coefficients of coefficients included in the selected rows or columns,

- if the number of non-zero coefficients of value determined is greater than one variable:

permutation (E555, E561) of the two rows or the two columns of the matrix to form a new matrix,

modifying (E562) the values of the coefficients of the new matrix in order to increase the value of the coefficients of the diagonal of the new matrix or of the coefficients close to the coefficients of the diagonal of the new matrix and to reduce the value of the other coefficients of the new matrix respecting the constraints,

determination (E563) of the number of zero value coefficients of the new matrix, selection (E565) of the new matrix as matrix if the number of zero value coefficients of the new matrix is greater than or equal to the number of value coefficients. null of the matrix and decrease in value of the variable.

The method of claim 2, wherein allocating a portion of rate capacity of the subscriber line having a strictly positive difference to at least one subscriber line having a strictly negative difference is performed by modifying ( E803) the power spectral density level of the wired link connecting the splitter device to the sub-distribution device of the subscriber line having a strictly positive difference.

8) The method of claim 7, further comprising the steps of: - forming a group of donors comprising subscriber lines whose previously determined difference in flow capacity capacity is strictly positive and a group of recipients whose previously determined flow capacity difference is strictly negative, and as long as one of the groups is not empty, the method also iteratively performs the steps of:

decreasing the power spectral density level of the wired links of the subscriber lines included in the group of donors and which connect the splitter device to the sub-distribution device until the differences in flow capacity previously determined are zero, determination, for each subscriber line, of the difference between the throughput capacity of the first wired link and the throughput capacity of the second wired link,

- Formation of a group of donors including subscriber lines whose previously determined difference in flow capacity is strictly positive and a group of recipients whose previously determined flow capacity difference is strictly negative.

The method according to claim 8, wherein decreasing the power spectral density level of the first wire links of the subscriber lines included in the group of donors until the previously determined differences in throughput capacity are zero, is performed by progressively or dichotomously decreasing the power spectral density threshold of each first wired link of the subscriber lines included in the group of donors and removing from a binary distribution table associated with each first wired link of subscriber lines included in the group of donors, the bits whose portions of energies necessary for their transmission exceed said threshold.

10) Flow capacity allocation system of at least a first subscriber line connecting a first customer (11 to 18, 61 to 68) to a distribution device (110, 610) to at least a second client connected to the splitter device by a second subscriber line, a plurality of clients being connected to the splitter device respectively by a subscriber line, each subscriber line connected to a customer consisting of a subscriber line. first wire link (LS1 to LS8, LS61 to LS68) 'connecting the splitter to a sub-distribution device (120, 620) and a second wire link (LT1 to LT8, LT61 to LT68) connecting the sub-distribution device and the client, characterized in that the system comprises:

means (120, 620) for rendering the flow capacities of the second film links independent of the capacity of the first wired links; determining means (100, 600) for at least part of the subscriber lines, a difference in throughput capacity between the first wired link and the second wired link of the same subscriber line,

means for allocating (100, 600) a portion of the rate capacity of the subscriber line having a strictly positive difference to at least one subscriber line having a strictly negative difference.

11) System according to claim 10 characterized in that the means (120, 620) for making the flow capacity of the second wire links independent of the capacity of the first wired links are constituted, for each subscriber line, of a active element comprising:

means for determining the throughput capacity of the first wired link, and the throughput capacity of the second wired link,

means for transferring the determined capacities to a database,

means for demodulating the signals received from the first wired link; means for modulating the demodulated signals;

- Means for transferring the modulated signals on the second wired link.

12) The system of claim 11, wherein each active element further comprises: - means for receiving data representative of frequency spectrum allocation of the first wired link,

means for processing said representative data for allocating the frequency spectrum of the first wired link, means for receiving the signals included in the different parts of the frequency spectrum of the first wired link,

means for demodulating the received signals to form demodulated information, means for transferring the demodulated information to the or each active element associated with the client 11 to 18 recipient of this information,

means for modulating information received from another active element,

means for transferring modulated information to the second wired link.

13) computer program stored on an information carrier, said program comprising instructions for implementing the method according to any one of claims 1 to 9 when it is loaded and executed by an electronic device.