(19)
(11)EP 2 997 724 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
26.06.2019 Bulletin 2019/26

(21)Application number: 14725072.4

(22)Date of filing:  13.05.2014
(51)International Patent Classification (IPC): 
H04M 11/06(2006.01)
H04M 3/00(2006.01)
H04B 3/32(2006.01)
(86)International application number:
PCT/EP2014/059736
(87)International publication number:
WO 2014/184181 (20.11.2014 Gazette  2014/47)

(54)

DISCONTINUOUS OPERATION IN COMMUNICATION SYSTEMS USING VECTORING

UNTERBROCHENER BETRIEB IN KOMMUNIKATIONSSYSTEMEN MITTELS VEKTORISIERUNG

PROCÉDÉS, DISPOSITIFS ET SYSTÈMES DESTINÉS À SOUTENIR UNE OPÉRATION DISCONTINUE DANS DES SYSTÈMES DE COMMUNICATION METTANT EN OEUVRE UNE VECTORISATION


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

(30)Priority: 13.05.2013 US 201361822478 P

(43)Date of publication of application:
23.03.2016 Bulletin 2016/12

(60)Divisional application:
19167062.9

(73)Proprietor: Lantiq Beteiligungs-GmbH & Co. KG
85579 Neubiberg (DE)

(72)Inventors:
  • STROBEL, Rainer
    80331 Munich (DE)
  • OKSMAN, Vladimir
    Morganville, New Jersey 07751 (US)

(74)Representative: Sticht, Andreas 
Kraus & Weisert Patentanwälte PartGmbB Thomas-Wimmer-Ring 15
80539 München
80539 München (DE)


(56)References cited: : 
EP-A1- 2 466 757
US-A1- 2009 270 038
  
  • BT: "G.fast: Low Power Modes - Continued;TD2013-03-Q4-067", ITU-T DRAFT ; STUDY PERIOD 2013-2016, INTERNATIONAL TELECOMMUNICATION UNION, GENEVA ; CH, vol. 4a/15, 10 May 2013 (2013-05-10), pages 1-8, XP017580484, [retrieved on 2013-05-10]
  
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

TECHNICAL FIELD



[0001] The present application relates to methods, devices and systems relating to discontinuous operation in communication systems using vectoring.

BACKROUND



[0002] Digital Subscriber Line (DSL) technology, like ADSL, ADSL2, VDSL, VDSL2 etc. up to the upcoming G.fast, during all its history, attempted to increase the bit rate in the aim to deliver more broadband services to the customer. Unfortunately, copper loops deployed from a Central Office (CO) to customer premises (CPE) are rather long and do not allow transmission of data with bit rates more than few Mb/s. To increase the bit rates, modern access networks use street cabinets, MDU-cabinets, and similar arrangements, generically also referred to as distribution points (DP) in this application: cabinets are connected to the CO by a high-speed backbone communication line, like multi- gigabit passive optical network (GPON) and installed close to the customer premises. From these cabinets or other DPs, high-speed DSL systems, such as Very-High-Bit-Rate DSL (VDSL), can be deployed. Current VDSL systems (ITU-T Recommendation G.993.2) have range of operation about 1 km, providing bit rates in the range of tens of Mb/s. To increase the bit rate of VDSL systems deployed from the cabinet, recent ITU-T Recommendation G.993.5 defined vectored transmission that allows increasing bit rates up to 100 Mb/s per direction.

[0003] Recent trends in the access communications market show that 100 Mb/s is still not sufficient and bit rates up to 1.0 Gb/s are required. This, based on current techniques, could be only achieved if copper pairs connecting the CPE as short as 50-100m. Operation using so short loops requires installation of many small street/MDU cabinets called Distribution Points (DP) that intends to serve a very small number of customers.

[0004] Vectoring is used in systems operating from a DP, to reduce far-end crosstalk (FEXT), which is helpful for obtaining high bit rates. To perform vectoring, transmissions from all lines deployed from the DP are synchronized (i.e., all downstream transmission are aligned in time and all upstream transmissions are aligned in time). Further, in the downstream direction, vectoring is implemented by precoding the transmit signals (at the DP), so that each signal includes transmission of its own data and FEXT cancellation signals intended to compensate FEXT generated by other lines. In the upstream direction, the received signals from all lines are jointly processed at the DP to remove FEXT components (also referred to as crosstalk cancellation or equalization). Details of the described vectoring technique are defined e.g. in G.993.2 for VDSL2, but may be applicable to other communication protocols.

[0005] Operation using short loops requires installation of many small DPs that intend to serve a small number of customers, such as 8-16, as shown by recent analysis. In some cases, a higher number of served users, like 24 or even 32 may be present. Therefore, it would be helpful if DPs allowed very flexible installation practices: they should be light and easy to install on a pole or house wall, or basement, without air-conditioning. The most challenging issue for these flexible connection plans is providing DPs with power. The solution intended so far is so-called "reverse feeding" when the equipment of the DP is fed by the connected customer.

[0006] Small internal space of the DP and use of reverse power feeding may imply substantial restrictions on the power consumption of a DP. Therefore, special means to reduce power consumption have been developed and new ones are on development. This application mainly concerns one efficient method of power savings, called Discontinuous operation.

[0007] Document US-A-2009/270038 discloses modifying crosstalk precompensator when a new communication line is added to a vectored group. Document EP-A-2466757 discloses changing crosstalk parameters in a vectored group when communication conditions changes, such as when a new line is introduced in the communication system. Document "G.fast: Low Power Modes - Continued;TD2013-03-Q4-067", ITU-T DRAFT ; STUDY PERIOD 2013-2016, INTERNATIONAL TELECOMMUNICATION UNION, GENEVA; CH, 10 May 2013 discloses discontinuous operation mode for a DSL system.

BRIEF DESCRIPTION OF THE DRAWINGS



[0008] 

Fig. 1 illustrates a communication system.

Fig. 2 is a diagram illustrating discontinuous operation.

Fig. 3 illustrates transmission on a line using two sets of precoder matrices.

Fig. 4 illustrates a precoder of a single distribution point port.

Figs. 5A and 5B illustrate a simulation of signal-to-noise ratios for a 16 line binder with equally distributed lengths. In Fig. 5A, all lines are active, in Fig. 5B, one line is disabled.

Fig. 6 illustrates behavior of the example of Fig. 5 after an update of matrix coefficients.

Fig. 7A and 7B show signal-to-noise ratios for a 16 line binder with equally distributed lengths from 50 meters to 250 meters with nonlinear precoding. In Fig. 7A, all lines are active, in Fig. 7B one line is disabled.

Fig. 8 illustrates a downstream system model with a Tomlinson-Harashima precoder.

Figs. 9A and 9B illustrate signal-to-noise ratio for a 16-line binder with equally distributed length from 50 meters to 250 meters with nonlinear precoding. In Fig. 9A, all lines are active, in Fig. 9B, two lines are disabled. The channel matrix has been updated, and disabled lines were the last ones added to the precoder.

Fig. 10 illustrates a receiver of a single distribution point port.

Fig. 11 is a diagram illustrating precoder settings and updates in a downstream direction.



[0009] The present invention is defined by independent claims 1 and 10 with particular embodiments being represented by the corresponding dependent claims.

[0010] Communication connections discussed in the following may be direct connections or indirect connections, i.e. connections with or without additional intervening elements, as long as the general function of the connection, for example to transmit a certain kind of signal, is preserved. Connections may be wireless connections or wire-based connections unless noted otherwise.

[0011] In some examples, efficient possibilities for updating vectoring coefficients, for example precoding coefficients, in systems using discontinuous operation may be discussed. For example, a precoder matrix may be updated based on actually active lines.

[0012] In other examples, other techniques may be employed. Turning now to the figures, in Fig. 1 a communication system is shown. The system of Fig. 1 comprises a provider equipment 10 communicating with a plurality of CPE units 14-16. While three CPE units 14-16 are shown in Fig. 1, this serves merely as an example, and any number of CPE units may be provided. Provider equipment 10 may be central office equipment, equipment in a distribution point (DP), or any other equipment used on a provider side. In case provider equipment 10 is part of a distribution point, it may for example receive and send data from and to a network via a fiber optic connection 110. In other examples, other kinds of connections may be used.

[0013] In Fig. 1, provider equipment 10 comprises a plurality of transceivers 11-13 to communicate with CPE units 14-16 via respective communication connections 17-19. Communication connections 17-19 may for example be copper lines, e.g. twisted pairs of copper lines. Communication via communication connections 17-19 may be a communication based on a multicarrier modulation like discrete multitone modulation (DMT) and/or orthogonal frequency division multiplexing (OFDM), for example an xDSL communication like ADSL, VDSL, VDSL2, G.Fast etc., i.e. a communication where data is modulated on a plurality of carriers, also referred to as tones. In some examples, the communication system may use vectoring, as indicated by a block 111 (e.g. a crosstalk reduction circuit like a precoder or equalizer) in Fig. 1. Vectoring comprises joint processing of signals to be sent and/or received to reduce crosstalk.

[0014] A communication direction from provider equipment 10 to CPE units 14-16 will also be referred to as downstream direction, and a communication direction from CPE units 14-16 will be also be referred to as upstream direction. Vectoring in the downstream direction is also referred to as crosstalk precompensation, whereas vectoring in the upstream direction is also referred to as crosstalk cancellation or equalization.

[0015] Provider equipment 10 and/or CPE units 14-16 may include further communication circuits (not shown) conventionally employed in communication systems, for example circuitry for modulating, bit loading, Fourier transformation etc.

[0016] In some examples, communication via communication connections 17-19 is a frame-based communication. A plurality of frames may form a superframe. In some examples, the communication uses time division duplexing, as will be explained later.

[0017] Besides others methods, discontinuous operation, which is employed in some examples, is one of the ways to save power, e.g. in DSL lines. With Time Division Duplexing (TDD) transmissions in opposite directions alternate: the unit of transmission time, called TDD frame, is divided into two parts called Downstream Transmission Opportunity (DS_TO) and Upstream Transmission Opportunity (US_TO). The DS_TO is a time period when downstream transmission occurs, and US_TO is the time period when upstream transmission occurs. With discontinuous operation, when no data is available to be transmitted during the TO or a part of the TO, the transmitted simply stops transmission instead of filling up the time gap by idle symbols, as it is done in some conventional systems (frame padding).

[0018] With discontinuous operation the number of symbols transmitted in a particular line during a particular TDD frame may be different, which is illustrated in Fig. 2. Figure 2 also shows markers that indicate when the active transmission will expire. Alternatively, the duration of the transmission in a number of consecutive TDD frames may be set in advance and indicated to the CPE during management communication (which indicates the number of transmitted symbols in a TDD frame). In this case Idle symbols (padding) will be added to frames in which the number of symbols required to transmit available data is less than the communicated value (shown in the right-hand side of Figure 2 with dashed lines). It should be noted that markers and Idle symbols are not supposed to be used at the same time).

[0019] Use of discontinuous operation is beneficial for power savings, especially for services that have dynamically changing number of active users and bit rates. For ultra-high bit rate systems deployed from the DP, further below techniques will be presented to facilitate discontinuous operation with vectoring.

[0020] Discontinuous operation was previously proposed and defined. It provides an ability to transmit symbols only when there is user data available for transmission, while powers the line off when no data for transmission is available. Therefore, with discontinuous operation, the number of symbols transmitted over a particular line of a vectored group during a particular TDD frame is different. In this sense, if vectoring is used, the number of lines in the vectored binder is changing on a symbol-per-symbol basis and pre-coding/post-coding associated with vectoring may need to be updated on a symbol-per-symbol basis, which is equivalent of changing the set of lines of the vectored group every symbol. This very frequent update of the precoder in all lines in conventional approaches would require super-fast processing capabilities or big extra memory to store pre-computed precoder coefficients - both methods results in high power consumption and extra complexity.

[0021] Another conventional approach, illustrated in Fig. 3, uses a second method, and in the aim to reduce the number of sets of coefficients to be stored, it proposes to group lines together, apply padding (to minimize the number of groups) and update precoding matrices small number of times (1-2) during the TDD frame (see Figure 3).

[0022] In Figure 3 there are two groups of lines, and padding symbols are added to establish small number of groups (e.g. 2-3) with same number of transmit symbols in each group. Idle symbols (padding) are transmitted with same power as regular data symbols, which reduces power savings in such an approach. In case the bandwidth of a particular line needs to be increased or substantially reduced, re-grouping is needed. The precoder matrices for each group may be stored and kept during the group assignment. This may limit the flexibility of such a conventional system, while padding itself may substantially reduce potential power savings.

[0023] A specific way of updating the precoder matrices (in the downstream direction) and decoder (postcoder) matrixes (in the upstream direction) are proposed, so that only few coefficients need to be updated and special order in the updating process allows almost no extra power and complexity. This way some examples avoid powerful computations (and high power consumption) and change the matrices on symbol bases without even re-computing matrix coefficients. This essentially provides a full flexibility in picking the number of transmit symbols in every TDD frame. The method according to some embodiments is reliable, guarantees quality of service (QoS), and provides efficient power saving at both DP and CPE.

[0024] Next, downstream vectored operation will be discussed.

[0025] A functional model of downstream vectoring according to some embodiments is presented in Figure 4 (LD stands for Line Driver, AFE for Analog Front End and IFFT for Inverse Fast Fourier Transform). A vectoring precoder is included into the transmitter to compensate FEXT (Far-End Crosstalk): one input of the precoder is the actual transmit signal for the line at tone i, Z(i), others are inputs are from other lines of the vectored group and intended for FEXT cancelling. The precoder applies inputs from other lines via precoding coefficients and special processing to compensate FEXT generated by other lines into line i.

[0026] Two types of precoders are conventionally employed: linear and nonlinear. With linear precoder, FEXT cancellation signals are added to the transmit signal Z(i) after multiplication on selected complex coefficients that represent FEXT coupling in the binder. With a nonlinear precoding, a nonlinear operation, such as MOD, is used to improve vectoring performance against noise.

[0027] If discontinuous operation is used, each line transmits different number of symbols in each TDD frame. That means that from a certain symbol (time instant T2 for Line 2 in Figure 2) the number of actual vectored lines changes - it may change after every symbol. When a line stops transmission (e.g., when Line 2 in Figure 2 stops transmission at time T2), this causes change in the FEXT generated into other lines, and the precoder coefficients (coefficients of the precoding matrix) in all other lines need to be updated to accommodate this change. Thus, precoding matrix may need to be updated every symbol in some cases.

[0028] The complexity and processing associated with the update may depend on the type of the precoding. In the worst case, precoding matrices of all remaining lines need to be obtained and stored at the transmitter prior the session, which requires big amount of memory to keep all the matrices associated with all possible symbol combinations. This doesn't seem to be practical. Thus, the above-desribed conventional approach uses line grouping and padding, as shown in Figure 3, to leverage complexity increase.

[0029] It should be noted that switching of the lines ON and OFF during discontinuous operation should not involve substantial changes in the impedance of the line; if the impedance of the transmitter or receiver changes substantially, it may results in substantial changes of FEXT coupling between all lines in the binder (especially at high frequencies). This may require re-estimating the channel prior to the coefficient update, which is impossible to be done on symbol-by-symbol bases. Therefore, in the following description it is assumed no substantial impedance changes associated with ON/OFF during discontinuous operation, which assumption is at least in good approximation applicable to many practical situations.

[0030] If the impedance stays unchanged, removal of one or more lines from the vectored group may still cause very substantial performance degradation if the matrices of the remaining active lines are not updated. The particular value of the degradation depends also on the type of precoding.

[0031] An example in Figure 5 shows a simulation example for an SNR degradation in lines of a 16-line binder using linear precoding, caused by a removal of just one line - the SNR reduction in some lines is more than 10 dB, which is obviously unacceptable (Fig. 5A shows the SNR with all lines active, Fig. 5B the SNR with one line switched off.)

[0032] To improve the SNR again, an update of matrix coefficients is necessary. After the update of the relevant matrix coefficients, the SNR gets bask, as it is shown in Figure 6.

[0033] For the same 16-line binder using nonlinear precoding (Tomlinson-Harashima vectored precoding, as proposed in some conventional approaches), the change of SNR after one line of the group is disabled is much more dramatic. The SNR degradation is shown in Figures 7A and 7B, Fig. 7A showing SNR before disabling of one line and Fig. 7B showing SNR after the disabling.

[0034] Next, discontinuous operation with linear precoder will be discussed.

[0035] Some examples consist of a method or a device having discontinuous operation and using linear precoder, which results in minimized number of computations for coefficient updates and thus can be performed on symbol bases.

[0036] With linear precoding, a received signal û can be presented as:

where u is the transmit signal prior to precoding, H is a channel matrix, P is the precoder matrix. The frequency equalized (FEQ) matrix G is considered as a part of the channel matrix H to simplify equations.

[0037] If some lines are disabled in the aim to save power, the corresponding rows and columns from the channel matrix H and the precoder matrix P are set to zero.

[0038] Accordingly, we partition the matrix into matrixes of active (a) and of disabled (d) lines. Before deactivation of the line, for perfect crosstalk precompensation the following holds.

I is the unitary matrix with non-zero elements only in the main diagonal. With disabled lines (due to discontinuous operation), equation (2) must still be fulfilled, which requires a different matrix





[0039] Based on the matrix inversion lemma, equation (3) is found for the calculation of the matrix coefficients for active lines:



[0040] With this simple coefficient re-computation, the matrix can be reduced to keep orthogonality after some lines have been disabled. For the case that only one line is disabled and the diagonal elements of P are equal to 1, the equation (4) simplifies to:



[0041] In case that the method is used for discontinuous operation, line deactivation is temporary and therefore a recomputation of the coefficients is disadvantageous. The operation of equation (5) can be, however, incooperated into the precoding operation (i.e., can be implemented by changing the order of multiplication and summation during precoding). This doesn't require additional computation resources.

[0042] For all active lines, we have

x being the transmit signal(s) after precoding. The transmit signal for a single line is given by equation (6):



[0043] With one line deactivated, we get:



[0044] Accordingly, for the remaining active lines, in case when one line is deactivated (discontinued), we have:

which uses the same precoder coefficients as when all lines are active.

[0045] Equation (9) shows that computation of the precoding output has the same number of multiply-accumulated (MAC operations) as when all lines are active since the value Pad·ua needs to be computed only once for all lines and needs to be scaled for each line differently by the corresponding coefficient pid.

[0046] If more than one line is discontinued, equation (9) still holds because this can be seen as sequential deactivation of a number of single lines. Thus, when more than one line is deactivated, the number of MAC operations for the computation on the active lines is still the same.

[0047] An alternative to the sequential calculation is the inversion of the matrix Pdd according to equation (10).



[0048] To avoid recalculation of the coefficients for

the first order approximation of matrix inversion according to equation (11) can be used.



[0049] With more than one line disabled, the assumption of 1 on the diagonal (result of initial PSD normalization) does no longer hold. Thus, this method may results in changes of the transmit PSDs (on the precoder output). Some lines transmit more power than before, while others transmit less. However, these PSD changes are rather small (∼ 2-4 dB), and if lines are enable and disabled for rather short time and randomly, as takes place in discontinuous operation, the average spectrum does not increase.

[0050] Next, discontinous operation with nonlinear precoder will be discussed.

[0051] The model of Tomlinson-Harashima nonlinear precoder is presented in Figure 8 with a forward matrix Pf, a feedback matrix Pb and an equalizer diagonal matrix G. n represents noise.

[0052] The nonlinear precoder matrices (Pb and Pf) are given by equations (12) to (15) shown below. The coefficients are calculated based on the QR decomposition according to equation (12), where Q is a unitary matrix and R is an upper triangular matrix. The matrix Π is a permutation matrix defining the encoding order of the data streams.









[0053] For discontinuous operation, the encoding order is selected such that the line that is disabled first is encoded first. By doing so, it is guaranteed that if the input signal of the disabled line is set to zero, a corresponding modulo output signal umod is also equal zero and the nonlinear operation does not change.

[0054] So, if lines are discontinued in the order that is reversed to the order they were added to the precoding matrix, the impact on SNR of the remaining lines is much less dramatic than shown in Figure 6 (right-hand side). The technique is possible if the duration of transmission over every line is known prior the start of the TDD frame (i.e., if matrix is computed prior time T0 in Figure 1 with known values of T1, T2, ... TN).

[0055] However, in some cases there may still be an impact on performance. This is because the signal umod at the output of the MOD (e.g. modulo) operation is transformed by the matrix Pf and must be zero to avoid distortions caused by deactivation of the corresponding lines.

[0056] Therefore, some embodiments comprise replacing the signals of the lines which were transmitted prior to deactivation, which, in embodiments, is given by equation (16).



[0057] In case that the precoder matrix Pf shall not be changed, the operation according to equation (16) can be performed to calculate the transmit signal for the remaining active lines. In contrast to the linear precoder case, additional coefficients for the matrix

are required.



[0058] This modification of the precoder matrix results in a change of the transmit PSD (Power Spectral Density), similarly as in the linear precoder case. Similarly as in case of linear precoder, this in embodiments is not expected to be harmful because the discontinued lines reduce the total power transmitted over the binder. After this update of the Pf, in an example scenario, the SNR got improved and recovered back to the values achieved before the discontinuous operation.

[0059] An example is shown in Figs. 9A and 9B, Fig. 9A illustrating a case where all lines are active, and Fig. 9B illustrating a case where two lines were disabled and techniques as disclosed herein for updating coefficients where applied.

[0060] Next, upstream vectored operation will be discussed.

[0061] The functional model of the receiver (e.g. in a DP) is presented in Figure 10. The output of the decoder is a composition of the signal received from the line and the sum of signals received from all other active lines multiplied by matrix coefficients, which corresponds to the aforementioned crosstalk cancellation or equalization.

[0062] If a receiver of a particular line is switched off, the corresponding input of the decoder is zeroed and FEXT components carried by the line are zeroed too. Therefore, similarly to the case of the downstream, the receiver shall update decoder matrix coefficients upon the number of actually transmitting lines changes. This is rather similar to the case of downstream. However, such coefficient recalculations, while greatly improving SNR related to FEXT, may, in some cases, result in an increase of background noise in some lines. At least for most relevant cases, this is expected to be insignificant (in fact, for most relevant cases the background noise will decrease).

[0063] Next, update of decoder matrix coefficients for crosstalk cancellation or equalization will be discussed.

[0064] For the following analysis we assume that when discontinued operation is used, the input of the decoder from the disabled line will be zero (the receiver is switched off, however in the way that the impedance of the line doesn't change (to avoid changes in FEXT coupling).

[0065] The coefficients of the updated matrix in some examples can be found in the way that is similar to the downstream:
Upstream model similar to equation (1):



[0066] Here, the equalizer matrix G has non-zero nondiagonal elements and is used for crosstalk cancellation. Coefficient correction of the decoder similar to equations (4, 18):



[0067] Correction of the receive signal instead of coefficient recalculation according to equation (19):



[0068] This changes the noise environment, because the receive signal y consists of received signal plus noise: H·u+n.

[0069] This noise increase can be compensated by temporary increase of the transmit PSD, or by bit loading change. For both cases additional protocol may be required.

[0070] In some examples, to avoid updating of the decoder matrix coefficients described above, the receiver is kept operating, so that FEXT signals accumulated by the line are added to the decoder. This requires AFE, FFT and precoder to be left ON. Same as in the downstream direction, the dynamic range of the AFE may be reduced if used frequency spectrum is limited to 100 MHz (none or very few pairs has close to negative or negative impulse-to-SNR ratio). This gives additional power saving.

[0071] Next, a fast power allocation update according to some examples will be described. To avoid PSD increase in downstream and noise increase in upstream, a fast update of the transmit subcarrier gains can be used.

[0072] In the downstream direction, the transmit gains scale the input of the precoder such that the output of the precoder does not violate transmit PSD. The scale factor update for each configuration of inactive lines is pre-computed by the DP for the following TDD frames and communicated to the CPEs.

[0073] For the upstream direction, the CPEs receive the update of the subcarrier gains table for the configurations contained in the following TDD frames. For frequencies up to 100 MHz where the direct channel can be assumed to be the strongest path between transmitter and receiver, this scaling will increase performance on average.

[0074] In one example that involves a linear precoder, the transceiver at the DP computes the precoder matrix and normalizes the PSD at an initialization of the vectored group. Further, for each symbol position in each TDD frame, the DP updates the precoder matrix based on actually active lines using any method as described above. Similarly, the DP computes the decoder matrix during the initialization and updates it on symbol-to-symbol bases using any method as described above.

[0075] The DP may adjust transmit PSD normalization if one or more lines are turned to a long term deactivation. Bit loading adjustment in the downstream direction may be a part of this procedure.

[0076] In another example that involves a linear precoder, the DP re-computes precoder coefficients and updates PSD normalization at the start of each TDD frame or a group of same TDD frames (TDD superframe), Figure 10, taking in account all active lines for a particular TDD frame (lines 1-3 and 5 in Frame N of Figure 11) or for a particular group of TDD frames. For the example in Figure 10, the DP performs 3 updates of the precoder matrix based on actually active lines using the method defined in section 5.1.2. The upstream decoder matrix may be updated on symbol-to-symbol bases using any method as described above.

[0077] The PSD normalization may be performed in the range that allows to keep the same bit loading; otherwise bit loading may need to be adjusted for the TDD frame or group of TDD frames.

[0078] In another example that involves a nonlinear precoder, the DP re-computes precoder coefficients at the at the start of each TDD frame or each group of same TDD frames (superframe) in the order lines are discontinuous during the TDD frame. For the scenario in Figure 11, for TDD frame N lines may, e.g., be precoded in order: #5, #2, #1, #3, and for TDD frame (N+1) they will be precoded in order: #6, #5, #2, #1, #3. At each update the DP also modifies appropriately the feed-forward matrix Pf using any method defined above.

[0079] The PSD normalization may be performed in the range that allows to keep the same bit loading; otherwise bit loading may need to be adjusted. This is usually only possible for a group of same TDD frames, so the PSD normalization can be pre-computed and bit loading changes may be exchanged with the CPE.

[0080] In another example, to update the precoder for a group of same TDD frames, padding is added to some transmissions of some frames. In all the mentioned embodiments no action from the receiver is required except when bit loading or subcarrier gains are modified. Method of exchanging new values of bit loading and gains may be trivial and is beyond the scope of this invention.

[0081] Other examples introduce special precoding update techniques (methods and algorithms) and a protocol associated with these updates. These examples may serve as a part of the new G.fast standard.

[0082] In some examples a vectored transmission correction network that employs discontinuous operation is provided, wherein a method or device updates precoder matrices and decoder or postcoder matrices such that a subset of coefficients needs to be updated.


Claims

1. A method, comprising:

reducing crosstalk of a plurality of lines (17-19) in a vectored group by vectoring, and

modifying the crosstalk reduction based on previous crosstalk reduction in case of a change of composition of the vectored group,
characterized in that modifying the crosstalk reduction comprises performing the crosstalk reduction in a case where only a subset of lines from a set of lines are active, based on crosstalk reduction coefficients for a case where all lines of the set of lines are active.


 
2. The method of claim 1, wherein the crosstalk reduction coefficients comprise at least one of precoder coefficients and equalization coefficients.
 
3. The method of claim 1 or 2, wherein, for a case where one line of the set of lines is deactivated, crosstalk is precompensated according to

wherein xai is a signal sent on active line a, P is a precompensation matrix, i indicates an inactive line and ua is a vector of data signals to be sent on the active lines.
 
4. The method of claim 1 or 2, wherein, for a case where one line of the set of lines is deactivated, crosstalk is precompensated according to

wherein for matrix inversion an approximation according to

is used, wherein xa is a vector of signals sent on active lines, P is a precompensation matrix, and ua is a vector of data signals to be sent on the active lines.
 
5. The method of claim 1 or 2, wherein, for a case where one line of the set of lines is deactivated, crosstalk is precompensated according to

wherein H is the channel matrix, wherein the precoding is a nonlinear precoding, wherein xa is a vector of signals sent on active lines, P is a precompensation matrix, and ua is a vector of data signals to be sent on the active lines.
 
6. The method of claim 1 or 2, wherein, for a case where one line of the set of lines is deactivated, crosstalk is cancelled according to

wherein ûa is a corrected received signal, ya is a received and G is an equalization matrix for a case when all lines of the set are active.
 
7. The method of any one of the preceding claims, further comprising performing a power allocation update.
 
8. The method of claim 7, wherein the update is performed based on a scale factor.
 
9. The method of any one of the preceding claims, wherein a change of composition of the vectored group is caused by discontinuous operation.
 
10. A device, comprising:

a transceiver (11-13), and a crosstalk reduction circuit (111) adapted to reduce crosstalk of a plurality of lines (17-19) in a vectored group by vectoring, and

to modify the crosstalk reduction based on previous crosstalk reduction in case of a change of composition of the vectored group,
characterized in that for modifying the crosstalk reduction, the crosstalk reduction circuit (111) is adapted to perform the crosstalk reduction in a case where only a subset of lines from a set of lines are active, based on crosstalk reduction coefficients for a case where all lines of the set of lines are active.


 
11. The device of claim 10, wherein the crosstalk reduction coefficients comprise at least one of precoder coefficients and equalization coefficients.
 
12. The device of any one of claims 10 or 11, wherein the device is adapted to respond to a change of composition of the vectored group caused by discontinuous operation.
 
13. The device of any one of claims 10-12, wherein the device is comprised in a distribution point.
 
14. The device of any one of claims 10-13, wherein the device is adapted to perform the method of any one of claims 1-9.
 


Ansprüche

1. Verfahren, umfassend:

Verringern eines Übersprechens mehrerer Leitungen (17-19) in einer vektorisierten Gruppe durch Vektorisierung und

Modifizieren der Übersprechungsverringerung, basierend auf voriger Übersprechungsverringerung im Fall einer Änderung einer Zusammensetzung der vektorisierten Gruppe,

dadurch gekennzeichnet, dass Modifizieren der Übersprechungsverringerung Ausführen der Übersprechungsverringerung in einem Fall umfasst, wo nur ein Teilsatz von Leitungen von einem Satz von Leitungen aktiv ist, basierend auf Übersprechungsverringerungskoeffizienten, für einen Fall, wo alle Leitungen des Satzes von Leitungen aktiv sind.


 
2. Verfahren nach Anspruch 1, wobei die Übersprechungsverringerungskoeffizienten mindestens eines umfassen von Präcoder-Koeffizienten und Ausgleichskoeffizienten.
 
3. Verfahren nach Anspruch 1 oder 2, wobei, für einen Fall, wo eine Leitung des Satzes von Leitungen deaktiviert ist, Übersprechen gemäß

vorkompensiert ist, wobei xai ein Signal ist, das auf einer aktiven Leitung a gesendet wird, P eine Vorkompensationsmatrix ist, i eine inaktive Leitung anzeigt und ua ein Vektor von Datensignalen ist, die auf den aktiven Leitungen zu senden sind.
 
4. Verfahren nach Anspruch 1 oder 2, wobei, für einen Fall, wo eine Leitung des Satzes von Leitungen deaktiviert ist, Übersprechen gemäß

vorkompensiert ist, wobei für Matrixumkehr eine Näherung gemäß

verwendet wird, wobei xa ein Vektor von Signalen ist, die auf aktiven Leitungen gesendet werden, P eine Vorkompensationsmatrix ist und ua ein Vektor von Datensignalen ist, die auf den aktiven Leitungen zu senden sind.
 
5. Verfahren nach Anspruch 1 oder 2, wobei, für einen Fall, wo eine Leitung des Satzes von Leitungen deaktiviert ist, Übersprechen gemäß

vorkompensiert ist, wobei H die Kanalmatrix ist, wobei das Vorcodieren ein nichtlineares Vorcodieren ist, wobei xa ein Vektor von Signalen ist, die auf aktiven Leitungen gesendet werden, P eine Vorkompensationsmatrix ist und ua ein Vektor von Datensignalen ist, die auf den aktiven Leitungen zu senden sind.
 
6. Verfahren nach Anspruch 1 oder 2, wobei, für einen Fall, wo eine Leitung des Satzes von Leitungen deaktiviert ist, Übersprechen gemäß

unterdrückt ist, wobei ûa ein korrigiertes empfangenes Signal ist, ya eine empfangene und G eine Ausgleichsmatrix ist, für einen Fall, wenn alle Leitungen des Satzes aktiv sind.
 
7. Verfahren nach einem der vorstehenden Ansprüche, ferner umfassend Ausführen einer Leistungszuteilungsaktualisierung.
 
8. Verfahren nach Anspruch 7, wobei die Aktualisierung basierend auf einem Skalenfaktor ausgeführt wird.
 
9. Verfahren nach einem der vorstehenden Ansprüche, wobei eine Änderung einer Zusammensetzung der vektorisierten Gruppe durch unterbrochenen Betrieb verursacht wird.
 
10. Vorrichtung, umfassend:

einen Sendeempfänger (11-13) und eine Übersprechungsverringerungsschaltung (111), die dazu ausgelegt ist, ein Übersprechen mehrerer Leitungen (17-19) in einer vektorisierten Gruppe durch Vektorisierung zu verringern und

die Übersprechungsverringerung basierend auf voriger Übersprechungsverringerung im Fall einer Änderung einer Zusammensetzung der vektorisierten Gruppe zu modifizieren,

dadurch gekennzeichnet, dass zum Modifizieren der Übersprechungsverringerung die Übersprechungsverringerungsschaltung (111) dazu ausgelegt ist, die Übersprechungsverringerung in einem Fall auszuführen, wo nur ein Teilsatz von Leitungen von einem Satz von Leitungen aktiv ist, basierend auf Übersprechungsverringerungskoeffizienten für einen Fall, wo alle Leitungen des Satzes von Leitungen aktiv sind.


 
11. Vorrichtung nach Anspruch 10, wobei die Übersprechungsverringerungskoeffizienten mindestens eines von Präcoder-Koeffizienten und Ausgleichskoeffizienten umfassen.
 
12. Vorrichtung nach einem der Ansprüche 10 oder 11, wobei die Vorrichtung dazu ausgelegt ist, auf eine Änderung einer Zusammensetzung für die vektorisierte Gruppe zu reagieren, die durch unterbrochenen Betrieb verursacht wird.
 
13. Vorrichtung nach einem der Ansprüche 10-12, wobei die Vorrichtung in einem Verteilungspunkt enthalten ist.
 
14. Vorrichtung nach einem der Ansprüche 10-13, wobei die Vorrichtung dazu ausgelegt ist, das Verfahren nach einem der Ansprüche 1-9 auszuführen.
 


Revendications

1. Procédé, comprenant :

la réduction de la diaphonie d'une pluralité de lignes (17-19) dans un groupe vectorisé par vectorisation, et

la modification de la réduction de diaphonie sur la base de la réduction de diaphonie précédente en cas de changement de composition du groupe vectorisé,

caractérisé en ce que la modification de la réduction de diaphonie comprend la réalisation de la réduction de diaphonie dans un cas où seul un sous-ensemble de lignes d'un ensemble de lignes sont actives, sur la base de coefficients de réduction de diaphonie pour un cas où toutes les lignes de l'ensemble des lignes sont actives.


 
2. Procédé selon la revendication 1, les coefficients de réduction de diaphonie comprenant au moins l'un des coefficients précodeurs et des coefficients d'égalisation.
 
3. Procédé selon la revendication 1 ou 2, pour un cas où une ligne de l'ensemble de lignes étant désactivée, la diaphonie étant précompensée selon

xai étant un signal envoyé sur la ligne active a, p étant une matrice de précompensation, i indiquant une ligne inactive et ua étant un vecteur des signaux de données à envoyer sur les lignes actives.
 
4. Procédé selon la revendication 1 ou 2, pour un cas où une ligne de l'ensemble de lignes est désactivée, la diaphonie étant précompensée selon

pour l'inversion de matrice, une approximation selon

xa étant un vecteur de signaux envoyés sur des lignes actives, P étant une matrice de précompensation, et ua étant un vecteur de signaux de données devant être envoyés sur les lignes actives.
 
5. Procédé selon la revendication 1 ou 2, pour un cas où une ligne de l'ensemble de lignes est désactivée, la diaphonie étant précompensée selon

H étant la matrice de canal, le précodage étant un précodage non linéaire, xa étant un vecteur de signaux envoyés sur des lignes actives, P étant une matrice de précompensation, et ua étant un vecteur de signaux de données devant être envoyés sur les lignes actives.
 
6. Procédé selon la revendication 1 ou 2, pour un cas où une ligne de l'ensemble de lignes est désactivée, la diaphonie étant annulée selon

ûa étant un signal reçu corrigé, ya étant un signal reçu et G étant une matrice d'égalisation pour un cas où toutes les lignes de l'ensemble sont actives.
 
7. Procédé selon l'une quelconque des revendications précédentes, comprenant en outre la réalisation d'une mise à jour de l'allocation de puissance.
 
8. Procédé selon la revendication 7, la mise à jour étant réalisée sur la base d'un facteur d'échelle.
 
9. Procédé selon l'une quelconque des revendications précédentes, un changement de composition du groupe vectorisé étant provoqué par un fonctionnement discontinu.
 
10. Dispositif, comprenant :

un émetteur-récepteur (11-13), et un circuit de réduction de diaphonie (111) adapté pour réduire la diaphonie d'une pluralité de lignes (17-19) dans un groupe vectorisé par vectorisation, et

pour modifier la réduction de diaphonie sur la base de la réduction de diaphonie précédente en cas de changement de composition du groupe vectorisé,

caractérisé en ce que pour modifier la réduction de diaphonie, le circuit de réduction de diaphonie (111) est adapté pour réaliser la réduction de diaphonie dans un cas où seul un sous-ensemble de lignes d'un ensemble de lignes sont actives, sur la base de coefficients de réduction de diaphonie pour un cas où toutes les lignes de l'ensemble de lignes sont actives.


 
11. Dispositif selon la revendication 10, les coefficients de réduction de diaphonie comprenant au moins l'un des coefficients précodeurs et des coefficients d'égalisation.
 
12. Dispositif selon l'une quelconque des revendications 10 ou 11, le dispositif étant adapté pour répondre à un changement de composition du groupe vectorisé provoqué par un fonctionnement discontinu.
 
13. Dispositif selon l'une quelconque des revendications 10-12, le dispositif étant compris dans un point de distribution.
 
14. Dispositif selon l'une quelconque des revendications 10-13, le dispositif étant adapté pour réaliser le procédé selon l'une quelconque des revendications 1-9.
 




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