Method and apparatus for fast and accurate acquisition of crosstalk coefficients

Abstract

In accordance with an embodiment, the method includes inserting a plurality of crosstalk probing signals within the wired multi-carrier communication system for probing the crosstalk from respective ones of the plurality of disturber lines into the victim line, carrying out crosstalk measurements over the victim line, and estimating the crosstalk coefficients from the crosstalk measurements. The method further includes organizing the plurality of disturber lines into subsets of disturber lines, and individually assigning disjoint groups of carriers to the respective subsets of disturber lines. The insertion of the plurality of crosstalk probing signals is confined within the respectively assigned groups of carriers. The subsets of disturber lines and/or the groups of carriers used for a second or subsequent iteration are tailored based on crosstalk characteristics observed for the respective disturber lines during a pervious iteration.

Claims

1. A method for acquiring crosstalk coefficients from a plurality of disturber lines into a victim line in a wired multi-carrier communication system, the method comprising iteratively performing: inserting a plurality of crosstalk probing signals within the wired multi-carrier communication system for probing crosstalk from respective ones of the plurality of disturber lines into the victim line, carrying out crosstalk measurements over the victim line, estimating the crosstalk coefficients from the crosstalk measurements, organizing the plurality of disturber lines into subsets of disturber lines, and individually assigning disjoint groups of carriers to respective subsets of disturber lines, the insertion of the plurality of crosstalk probing signals being confined within the respectively assigned groups of carriers, wherein at least one of the subsets of disturber lines and the assigned groups of carriers used for a second or a subsequent iteration, is adjusted based on the crosstalk coefficients for respective disturber lines during a previous iteration.

2. The method according to claim 1, further comprising: determining an indication of a frequency coherence of respective crosstalk channels based on the crosstalk coefficients.

3. The method according to claim 1, further comprising: determining an indication of amounts of nominal crosstalk incurred from the respective disturber lines based on the crosstalk coefficients.

4. The method according to claim 1, further comprising: determining an indication of amounts of residual crosstalk incurred from the respective disturber lines that remain uncompensated over the victim line based on the crosstalk coefficients.

5. The method according to claim 1, wherein the individually assigning assigns interleaved groups of carriers with respective decimation factors to the subsets of disturber lines, and the respective decimation factors used for a first iteration are greater than a frequency coherence of a crosstalk channel.

6. The method according to claim 1, wherein the subsets of disturber lines include one or more disturber lines.

7. The method according to claim 1, wherein the estimating the crosstalk coefficients from one of the plurality of disturber lines into the victim line, comprises: obtaining respective crosstalk measurements carried out over the victim line at frequency indexes of a carrier group assigned to one of the subsets of disturber lines to which the disturber line belongs, estimating the crosstalk coefficients at the frequency indexes from the respective crosstalk measurements, and determining remaining crosstalk coefficients at the remaining ones of frequency indexes by means of interpolation.

8. The method according to claim 1, wherein the crosstalk measurements include Signal to Noise and Interference Ratio (SNIR) measurements.

9. The method according to claim 8, wherein the plurality of crosstalk probing signals are weighted replicas of a plurality of regular signals transmitted over, or received from, a respective one of the plurality of disturber lines, and the plurality of crosstalk probing signals are combined as a single crosstalk probing signal superimposed over a further regular signal transmitted over, or received from, the victim line.

10. The method according to claim 8, wherein the plurality of crosstalk probing signals are weighted replicas of a regular signal transmitted over, or received from the victim line, and the plurality of crosstalk probing signals are superimposed over respective ones of a plurality of further regular signals transmitted over, or received from, respective ones of the plurality of disturber lines.

11. The method according to claim 1, wherein the crosstalk measurements include slicer error measurements.

12. The method according to claim 11, further comprising: obtaining the plurality of crosstalk probing signals by modulating the carriers of the respective carrier groups with orthogonal probing sequences, and synchronously transmitting the plurality of crosstalk probing signals over respective ones of the plurality of disturber lines during at least one dedicated transmission.

13. A vectoring controller configured to acquire crosstalk coefficients from a plurality of disturber lines into a victim line in a wired multi-carrier communication system, the vectoring controller being configured to iteratively, insert a plurality of crosstalk probing signals for probing the crosstalk from respective ones of the plurality of disturber lines into the victim line, receive crosstalk measurements carried out over the victim line, estimate the crosstalk coefficients from the crosstalk measurements, organize the plurality of disturber lines into subsets of disturber lines, assign disjoint groups of carriers to respective subsets of disturber lines, confine the insertion of the plurality of crosstalk probing signals within the respectively assigned groups of carriers, and adjust at least one of the subsets of disturber lines and the assigned groups of carriers, used for a second or a subsequent iteration, based on the crosstalk coefficients for the respective disturber lines during a previous iteration.

14. A Digital Subscriber Line Access Multiplexer DSLAM comprising a vectoring controller according to claim 13.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The above and other objects and features of the invention will become more apparent and the invention itself will be best understood by referring to the following description of an embodiment taken in conjunction with the accompanying drawings wherein:

(2) FIG. 1 represents an exemplary DSL communication system;

(3) FIG. 2 represents a flow chart of a crosstalk acquisition algorithm as per the present invention;

(4) FIG. 3 represents a DSLAM as per the present invention;

(5) FIG. 4 represents an exemplary assignment of downstream carriers to the respective disturber lines; and

(6) FIG. 5 represents a plot of the mean rate loss as a function of the carrier group decimation factor.

DETAILED DESCRIPTION OF THE INVENTION

(7) There is seen in FIG. 1 a DSL communication system 1 for providing broadband access to subscribers.

(8) The DSL communication system 1 comprises a loop plant 300, a DSLAM 100 and N CPEs 200.sub.1 to 200.sub.N (or CPE.sub.1 to CPE.sub.N). The DSLAM 100 comprises N transceivers 110.sub.1 to 110.sub.N (or TU.sub.1 to TU.sub.N) coupled via N subscriber lines L.sub.1 to L.sub.N to respective ones of the N CPEs 200.sub.1 to 200.sub.N. The subscriber lines L.sub.1 to L.sub.N are for example Unshielded Twisted Pairs (UTP).

(9) The subscriber lines L.sub.1 to L.sub.N are bundled together within a common binder B, and induce crosstalk into each other as they are in close vicinity over whole or part of their length. The subscriber lines L.sub.1 to L.sub.N are assumed to form part of the same vectoring group.

(10) Because downstream and upstream communications are assigned different and non-overlapping frequency bands, a principle commonly referred to as Frequency Division Duplexing (FDD), crosstalk mostly reduces to Far-End Crosstalk (FEXT): some substantial amount of the signal transmitted by a transceiver (the disturber) couples into a neighboring line and impairs reception of the direct signal received over that neighboring line at a remote transceiver (the victim). For instance, the downstream signal transmitted by the transceiver 110.sub.1 over line L.sub.1 couples into line L.sub.N and is detected as noise by the CPE 200.sub.N. Also, the upstream signal transmitted by the CPE 200.sub.N over line L.sub.N couples into line L.sub.1 and is detected as noise by the transceiver 110.sub.1.

(11) FIG. 2 shows a flow chart of a crosstalk acquisition algorithm 1000 as per the present invention.

(12) As aforementioned, there are a number of prior art methods for estimating the crosstalk coefficients from a plurality of disturber lines, say lines L.sub.1 to L.sub.N-1, into a victim line, say line L.sub.N. The number of observations, and hence the time required, is proportional to N1 (i.e., the number of disturber lines): using pilot sequence of length at least N1, one needs error samples error(f) from at least N1 SYNC symbols; using SNIR measurements with one disturber at a time, one needs one shared base plus two perturbed SNIR measurements per disturber line per iteration, i.e. 1+2.Math.(N1) measurements per iteration (or 3.Math.(N1) if using new base SNIR measurement for each disturber line); or using SNIR method with simultaneous perturbations for all disturber lines, one still needs 1+2.Math.(N1) to get a set of linearly-independent equations that can be solved.

(13) The proposed crosstalk acquisition algorithm 1000 comprises the following steps.

(14) In a first step 1001, select a given victim line, and Q disturber lines whose crosstalk towards the victim line needs to be characterized (QN1).

(15) In a second step 1002, split the Q disturber lines into M subsets SL.sub.1 to SL.sub.M (2MQ).

(16) In a third step 1003, divide the tones into disjoint but preferably interleaved carrier groups CG.sub.1 to CG.sub.M assigned to the subsets SL.sub.1 to SL.sub.M respectively.

(17) In a fourth step 1004, insert crosstalk probing signals in parallel for the respective subsets of disturber lines SL.sub.1 to SL.sub.M. The insertion of the crosstalk probing signals is frequency-confined within the respective carrier groups, i.e. within the carrier group CG.sub.1 for disturber lines of the subset SL.sub.1, within the carrier group CG.sub.2 for disturber lines of the subset SL.sub.2, and so on.

(18) In a fifth step 1005, obtain crosstalk measurements from the victim receiver while crosstalk probing signals are being inserted.

(19) In a sixth step 1006, obtain crosstalk estimates from the disturber lines in the subset SL.sub.1 towards the victim line on tones in CG.sub.1 in time proportional to the size of the subset SL.sub.1, from the disturber lines in the subset SL.sub.2 towards the victim line on tones in CG.sub.2 in time proportional to the size of the subset SL.sub.2, and so on; extend estimation using interpolation and/or denoising to the rest of the tones. Optionally, use part or all of these new crosstalk estimates for updating the precoder or postcoder.

(20) In a seventh step 1007, determine crosstalk characteristics for the respective disturber lines, such as the frequency coherence of the crosstalk channels, the most dominant disturber lines for the given victim line, etc.

(21) In a eighth step 1008, decide whether more measurement points are needed for one or more disturber lines out of the Q disturber lines. If so then proceed with the next step, else exit the algorithm.

(22) In a ninth step 1009, and based on the crosstalk observations during step 1007, adjust the number Q of disturber lines selected for further crosstalk characterization, and/or the number M of subsets of disturber lines, and/or the respective composition of the subsets SL.sub.1 to SL.sub.M, and/or the respective composition of the carrier groups CG.sub.1 to CG.sub.M; and reiterate through the steps 1004 to 1008 with the newly determined subsets of disturber lines and/or the newly determined carrier groups until the crosstalk coefficients are adequately estimated.

(23) For example, one or more difficult lines with rapid crosstalk variations across frequency and/or strong crosstalk are grouped in a first subset that is assigned a large closely-spaced carrier group, and similarly easy lines with slow crosstalk variations across frequency and/or weak crosstalk are grouped in a second subset that is assigned a small widely-spaced carrier group.

(24) The subsets of disturber lines and/or the carrier groups can also be tailored based on the residual crosstalk. So, for example, disturbers whose crosstalk has already been adequately suppressed in previous iterations do not need to be included in any subset, and so the size of the subsets can be decreased and/or the number of tones in the carrier group can be increased.

(25) Since crosstalk acquisition is now done in parallel, the time required is proportional to the maximum size of the subsets SL.sub.1 to SL.sub.M, which, if designed properly, should be much less than N1 (typically (N1)/M).

(26) Optionally, the first iteration may use a frequency spacing much larger than the frequency coherence of a typical crosstalk channel. In this case the estimates are not accurate enough to get close to full vectored performance, but this is not the purpose. The purpose is rather to quickly distinguish disturbers that are significant from those which have no significant crosstalk. Then, in one or more subsequent iterations, the significant disturbers are targeted and estimated precisely with smaller frequency spacing. So there is a fast yet rough determination step, followed by a targeted precise estimation step.

(27) Let F1 denote a typical frequency coherence of a typical crosstalk channel expressed in tones (coherence across F1 tones), i.e. how far can one interpolate and still get very precise estimates hence full vectored performance; let F2 denote a secondary frequency coherence, i.e. how far can one subsample and still detect whether a line is a significant disturber or not; let D denote the number of significant disturbers out of the N1 potential disturbers (e.g., those in the same binder as the given victim line). Typically, we have F1<<F2 and D<<N1.

(28) The time required using F1 as default decimation factor is T0=N/F1. In the new approach, the time required is T1=N/F2 for the first iteration, plus T2=D/F1 for the second iteration. Typically T2<<T0 assuming F2>>F1. Typically T1<<T0 assuming D<<N. Hence T1+T2<<T0.

(29) The proposed crosstalk acquisition algorithm 1000 may be used during initial start-up procedures, or for crosstalk tracking, or for rapidly reacting when changes in crosstalk occur for one or more vectored lines.

(30) FIG. 3 provides further details about the DSLAM 100.

(31) The DSLAM 100 comprises the following functional blocks: the N transceivers 110; a Vectoring Processing Unit 120 (or VPU); and a vectoring Control Unit 130 (or VCU) for controlling the operation of the VPU 120.

(32) Each one of the transceivers 110 is coupled to the VPU 120 and to the VCU 130. The VCU 130 is further coupled to the VPU 120.

(33) The VPU 120 and the VCU 130 can be co-located with the transceivers 110 on a single Line Termination (LT) board for board level vectoring, or can form part of a dedicated Printed Board Assembly (PBA) for system level vectoring.

(34) Each one of the transceivers 110 comprises: a Digital Signal Processor (DSP) 111, an Analog Front End (AFE) 112, and a Line Adaptation Unit (LAU) 113.

(35) The N DSPs 111 are coupled to respective ones of the N AFE units 112. The N AFEs 112 are further coupled to respective ones of the N LAUs 113. The N LAUs 113 are further coupled to respective ones of the N lines L.sub.1 to L.sub.N.

(36) Each one of the DSPs 111 is arranged to operate both a downstream and an upstream DSL communication channel.

(37) Each one of the DSPs 111 is for encoding and modulating user and control data into digital data symbols, and for de-modulating and decoding user and control data from digital data symbols.

(38) The following transmit steps are typically performed within the DSPs 111: data encoding, such as data multiplexing, framing, scrambling, error correction encoding and interleaving, signal modulation, comprising the steps of ordering the carriers according to a carrier ordering table, parsing the encoded bit stream according to the bit loadings of the ordered carriers, and mapping each chunk of bits onto an appropriate transmit constellation point (with respective carrier amplitude and phase), possibly with Trellis coding, signal scaling, Inverse Fast Fourier Transform (IFFT) Cyclic Prefix (cP) insertion, and time-windowing.

(39) The following receive steps are typically performed within the DSPs 111: time-domain signal equalization, Cyclic Prefix (cP) removal, Fast Fourier Transform (FFT), frequency-domain signal equalization, signal de-modulation and detection, comprising the steps of applying to each and every equalized frequency sample an appropriate constellation grid, the pattern of which depends on the respective carrier bit loading, detecting the expected transmit constellation point and the corresponding transmit bit sequence, possibly with Trellis decoding, and re-ordering all the detected chunks of bits according to the carrier ordering table, data decoding, such as data de-interleaving, RS decoding (byte errors, if any, are corrected at this stage), de-scrambling, frame delineation and de-multiplexing.

(40) Each one of the DSPs 111 is further configured to operate an Embedded Overhead Channel (EOC) that is used to transport control and management messages, such as OLR commands and responses. The EOC data are multiplexed with the user data over the DSL channel.

(41) Each one of the AFES 112 comprises a Digital-to-Analog Converter (DAC) and an Analog-to-Digital Converter (ADC), a transmit filter and a receive filter for confining the signal energy within the appropriate communication frequency bands while rejecting out-of-band interference, a line driver for amplifying the transmit signal and for driving the transmission line, and a Low Noise Amplifier (LNA) for amplifying the receive signal with as little noise as possible.

(42) Each one of the LAUS 113 comprises a hybrid for coupling the transmitter output to the transmission line and the transmission line to the receiver input while achieving low transmitter-receiver coupling ratio (e.g., by means of echo cancellation techniques), further transmit and receive high-pass filters for filtering out any unwanted signals present in the POTS/ISDN frequency bands, impedance-matching circuitry for adapting to the characteristic impedance of the line, and isolation circuitry (typically a transformer).

(43) Each one of the DSPS 111 is further configured to supply transmit frequency-domain samples to the VPU 120 before Inverse Fast Fourier Transform (IFFT) step for joint signal precoding, and to supply receive frequency-domain samples to the VPU 120 after Fast Fourier Transform (FFT) step for joint signal post-processing.

(44) Each one of the DSPS 111 is further configured to receive corrected frequency-domain samples from the VPU 120 for further transmission or detection. Alternatively, the DSPS 111 may receive correction samples to add to the initial frequency-domain samples before further transmission or detection.

(45) The VPU 120 is configured to mitigate the crosstalk induced over the transmission lines L.sub.1 to L.sub.N. This is achieved by multiplying a vector S of transmit frequency-domain samples with a precoding matrix P so as to compensate for an estimate of the coming crosstalk (downstream), or by multiplying a vector R of receive frequency-domain samples with a crosstalk cancellation matrix G so as to cancel an estimate of the incurred crosstalk (upstream).

(46) Let i and j denote line indexes ranging from 1 to N, k a frequency index, and l a Discrete Multi-Tone (DMT) symbol index.

(47) Let S.sub.i,k.sup.l and S*.sub.i,k.sup.l denote the transmit frequency-domain samples transmitted over line L.sub.i during DMT symbol 1 before and after crosstalk pre-compensation by the VPU 121 respectively.

(48) Similarly, let R.sub.i,k.sup.l and R*.sub.i,k.sup.l denote the receive frequency-domain samples received from line L.sub.i during DMT symbol l before and after crosstalk cancellation respectively.

(49) We have:

(50) $\begin{array}{cc}\begin{array}{c}{S}_k^{*l}=\left[\begin{array}{c}{S}_{1,k}^{*l}\\ S_{2,k}^{*l}\\ \mathrm{.Math.}\\ S_{N,k}^{*l}\end{array}\right]\\ =P_k.Math.S_k^l\\ =\left[\begin{array}{cccc}1& P_{1,2,k}& \mathrm{.Math.}& P_{1,N,k}\\ P_{2,1,k}& 1& & \mathrm{.Math.}\\ \mathrm{.Math.}& & & P_{N-1,N,k}\\ P_{N,1,k}& \mathrm{.Math.}& P_{N,N-1,k}& 1\end{array}\right].Math.\left[\begin{array}{c}{S}_{1,k}^{*l}\\ S_{2,k}^{*l}\\ \mathrm{.Math.}\\ S_{N,k}^{*l}\end{array}\right],\end{array}\mathrm{and}& \left(2\right)\\ \begin{array}{c}{R}_k^{*l}=\left[\begin{array}{c}{R}_{1,k}^{*l}\\ R_{2,k}^{*l}\\ \mathrm{.Math.}\\ R_{N,k}^{*l}\end{array}\right]\\ =G_k.Math.R_k^l\\ =\left[\begin{array}{cccc}1& G_{1,2,k}& \mathrm{.Math.}& G_{1,N,k}\\ G_{2,1,k}& 1& & \mathrm{.Math.}\\ \mathrm{.Math.}& & & G_{N-1,N,k}\\ G_{N,1,k}& \mathrm{.Math.}& G_{N,N-1,k}& 1\end{array}\right].Math.\left[\begin{array}{c}{R}_{1,k}^{*l}\\ R_{2,k}^{*l}\\ \mathrm{.Math.}\\ R_{N,k}^{*l}\end{array}\right].\end{array}& \left(3\right)\end{array}$

(51) In the matrix P or G, a row i represents a particular victim line L.sub.i, while a column j represents a particular disturber line L.sub.j. At the intersection, the coupling coefficient that should be applied to the corresponding disturber transmit or receive frequency-domain sample for mitigating over the victim line L.sub.i the crosstalk from the disturber line L.sub.j. Not all the coefficients of the matrix need to be determined, for instance on account of limited vectoring capabilities first assigned to the strongest crosstalkers, or still for instance due to the fact that some lines do not noticeably interact with each other. The undetermined coefficients are preferably set to 0.

(52) The VCU 130 is basically for controlling the operation of the VPU 120, and more specifically for estimating or updating the crosstalk coefficients between vectored lines, and for initializing or updating the precoding matrix P and the crosstalk cancellation matrix G from the so-estimated crosstalk coefficients.

(53) Presently, the VCU 130 makes use of the SNIR method for estimating or updating the crosstalk coefficients. The VCU 130 gathers SNIR measurements carried out over a particular victim line, say line L.sub.N, while weighted replicas of regular signals transmitted over the disturber lines, say lines L.sub.1 to L.sub.N-1, are being superimposed over the regular signal transmitted over the victim line.

(54) For instance, the line L.sub.N may be a legacy line that is being initialized and further brought into service, while the lines L.sub.1 to L.sub.N-1 are active lines in showtime mode. The crosstalk coefficients from and toward the initializing line L.sub.N need to be estimated first before the crosstalk from and toward that line can be mitigated.

(55) The description will now focus on crosstalk acquisition from the disturber lines L.sub.1 to L.sub.N-1 towards the victim line L.sub.N for downstream communication.

(56) In a first step, the VCU 130 splits the whole set of disturber lines L.sub.1 to L.sub.N-1 (Q=N1) into M subsets of disturber lines SL.sub.1 to SL.sub.M comprising one or more disturber lines (2MN1). The subsets SL.sub.1 to SL.sub.M do not necessarily comprise an equal number of disturber lines.

(57) The VCU 130 next assigns disjoint sets CG.sub.1 to CG.sub.M of downstream carriers to the subsets SL.sub.1 to SL.sub.M respectively. The carrier assignment algorithm shall take due account of the measured SNIR granularity that is enforced by the SNIR measurements and reporting primitives.

(58) Typically, SNIR is measured and averaged over multiple contiguous carriers and over successive DMT symbols (typically 256). An SNIR carrier group, i.e. a carrier group for which only a single measured SNIR value is available, may comprise 1, 2, 4 or 8 contiguous carriers, depending on the transmission profile being used and the maximum carrier index being supported.

(59) The VCU 130 first determines the SNIR downstream carrier groups based on the maximum carrier index being supported. Next, the VCU 130 discards any SNIR downstream carrier groups, a carrier of which is not a member of the downstream MEDLEY set, i.e. the overall set of active downstream carriers that are used for downstream communication, as SNIR is not expected to be reported for those groups. Finally, the VCU 130 assigns the remaining SNIR carrier groups to the respective subsets SL.sub.1 to SL.sub.M.

(60) Let denote the overall number of valid SNIR downstream carrier groups, and let SCG.sub.n denote the nth valid SNIR downstream carrier group (1n).

(61) For a first initial crosstalk acquisition, the set CG.sub.m of downstream carriers assigned to a particular subset SL.sub.m (1mM) could be determined as follows:

(62) $\begin{array}{cc}{\mathrm{CG}}_m=\underset{n}{.Math.}{\mathrm{SCG}}_n,\mathrm{with}n=m+.Math.M,0-1,=.Math./M.Math.& \left(4\right)\end{array}$

(63) For instance, if SNIR measurements were averaged over two carriers, carriers would be pair-wise assigned to the respective subsets of disturber lines, say carrier pair {k; k+1} to the first subset, carrier pair {k+2; k+3} to the second subset, and so on.

(64) The remaining SNIR downstream carrier groups SCG.sub..Math.M+1 to SCG.sub., if any, are evenly distributed to selected subsets of disturber lines in a fixed, arbitrary or random manner.

(65) This initial carrier assignment applies a common decimation (or sub-sampling) factor to all the carrier groups, presently the number M of subsets of disturber lines, and achieves an almost-identical inter-carrier spacing pattern, thereby allowing a very first characterization of the crosstalk channels.

(66) Subsequently, and depending on the smoothness and/or strength of the crosstalk estimates, different decimation factors could be applied to the respective subsets of disturber lines.

(67) In a second step, crosstalk probing signals are being superimposed on the regular downstream signal transmitted over the victim line.

(68) The crosstalk probing signal for estimating the crosstalk coefficients from a given disturber line L.sub.j toward a given victim line L.sub.i is a weighted replica of the regular downstream signal transmitted over the disturber line L.sub.j, and is superimposed over the victim line L.sub.i. The crosstalk probing signal additively or destructively interfere with the actual crosstalk signal from the disturber line L.sub.j depending on the chosen weight values and the crosstalk coupling function.

(69) Also, the VCU 130 confines the insertion of the crosstalk probing signals to the respective carrier groups, meaning that, at a particular frequency index kCG.sub.m, the VCU 130 inserts one or more probing signals for one or more disturber lines that belong to the subset SL.sub.m, and does not insert any probing signal for the disturber lines that do not belong to the subset SL.sub.m.

(70) Thus, we have:

(71) $\begin{array}{cc}{\mathrm{XTS}}_{i,k}^l=\underset{ji}{\overset{}{.Math.}}{}_{j,k}.Math.S_{j,k}^l,& \left(5\right)\end{array}$
wherein XTS.sub.i,k.sup.l denotes the k-th frequency sample of the overall crosstalk probing signal superimposed on the victim line L.sub.i during DMT symbol l, and .sub.j,k denotes a complex weight value used for the disturber line L.sub.j on tone k,
and wherein .sub.j,k=.sub.j (6a) for kCG.sub.m and for a disturber line L.sub.jSL.sub.m, and .sub.j,k=0 (6b) for kCG.sub.m and for a disturber line L.sub.j.Math.SL.sub.m.

(72) Equations (6a) and (6b) ensure that the crosstalk probing signals for the respective disturber lines are frequency-confined within the respective carrier groups. This non-overlapping frequency-interleaved signal insertion allows parallel and fast acquisition of crosstalk coefficients from multiple disturber lines.

(73) Three successive SNIR measurement rounds with three distinct complex weight values are required for estimating both the amplitude and phase of the crosstalk coefficients from a given disturber line. Typically, .sub.j=0 (i.e., no crosstalk probing signal is inserted; this base SNIR measurement can be performed prior to or after the insertion of the crosstalk probing signals), .sub.j= (i.e., a scaled replica of the regular data signal is inserted), and .sub.j=j (i.e., a scaled quadrature replica of the regular data signal is inserted) for the three respective measurement rounds. A suitable value for is chosen such that the impact on the SNR is measurable but not excessive (typically no more than 3 dB SNIR variation at the victim receiver).

(74) If a subset SL.sub.m comprises more than one disturber line, then the corresponding crosstalk probing signals may be sent sequentially one after the other (e.g., as per Appendix III.2 of G.993.5 ITU recommendation), meaning a total of 1+2.Math.size(SL.sub.m) observations are required for acquiring the crosstalk coefficients from this subset of disturber lines, or the crosstalk probing signals may be combined all together with respective weight values {.sub.j}.sub.LjSLm (e.g., as per Appendix III.3 of G.993.5 ITU recommendation). Yet, linearly-independent weight values are required so as to get a set of linearly-independent equations that can be solved, meaning still a total of 1+2.Math.size(SL.sub.m) observations.

(75) The superimposition of the crosstalk probing signals can be easily achieved by adding the weight values .sub.j,k to the respective coefficients P.sub.i,j,k, ji of the precoder matrix P.

(76) Initially, when a new joining line L.sub.i joins an existing vectoring group, the off-diagonal elements of the i-th row of the precoding matrix P are all null as the crosstalk channels toward that new joining line L.sub.i have not been characterized yet. So are the off-diagonal elements of the i-th column as the crosstalk channels from that new joining line L.sub.i have not been characterized yet.

(77) A first SNIR measurement round now takes place while first weight values .sub.j,k.sup.(1)=.sup.(1)/j.sup.(1) are added to the current matrix coefficients values P.sub.i,j,k, ji: P.sub.i,j,k.sup.(1)=.sub.j,k.sup.(1).

(78) Next, the values of the matrix coefficients P.sub.i,j.sup.(1) are determined from the first rough estimates of the crosstalk coefficients, and the VPU 120 starts canceling a part of the crosstalk induced over the victim line L.sub.i. Yet, some residual crosstalk remains owing to the crosstalk estimation errors and SNIR measurement inaccuracies.

(79) A new SNIR measurement round now takes place while new weight values .sub.j,k.sup.(2)=.sup.(2)/j.sup.(2) are added to the current matrix coefficients values P.sub.i,j,k, ji: P.sub.i,j,k.sup.(2)=P.sub.i,j,k.sup.(1)+.sub.j,k.sup.(2).

(80) The value .sup.(2) used for that second measurement round is typically lower than the value .sup.(1) that was used during the first measurement round as part of the crosstalk is already canceled. The process re-iterates till the residual crosstalk power is below a pre-determined relative or absolute threshold.

(81) In a third step, the vCU 130 instructs the corresponding transceiver 110, namely the transceiver 110.sub.N, to fetch SNIR measurement values from the victim CPE, namely the CPE 200.sub.N, while the crosstalk probing signals are being superimposed on the victim line L.sub.N.

(82) Thereupon, the transceiver 110.sub.N sends multiple PMD READ commands to read the SNIR values as measured by the CPE 200.sub.N. Typically, a single PMD READ command reads a single SNIR value SNIR.sub.n as averaged over a particular SNIR downstream carrier group SCG.sub.n and over successive DMT symbols. The SNIR values for all the SNIR downstream carrier groups are read sequentially while a single weight value is in force. Next, the weight values are adjusted, and the process re-iterates.

(83) In a fourth step, the VCU 130 estimates the crosstalk coefficient values from the measured SNIR values as set forth in Appendix III of G.993.5 ITU recommendation, the content of which is entirely incorporated therein.

(84) The missing crosstalk coefficients at intermediary frequency indexes are determined by means of interpolation. Denoising techniques may also be involved.

(85) Optionally, the vCU 130 proceeds with the computation or update of the precoding matrix P from the new crosstalk estimates. The VCU 130 can use a first-order matrix inversion to compute the coefficients of the precoding matrix P, or any other suitable method.

(86) The VCU 130 is further configured to quantify the variation across frequency of the crosstalk coefficients (both phase and amplitude), which is indicative of the frequency coherence of the crosstalk channel. The VCU 130 sorts the disturber lines based on the frequency coherence of the respective crosstalk channels. The VCU 130 may further sort the disturber lines based on the magnitude of the respective crosstalk coefficients, which is indicative of the amount of crosstalk impairment induced over the victim line.

(87) Based on the smoothness and/or magnitudes of the crosstalk coefficients, the vCU 130 adjusts the number M of subsets of disturber lines, and/or re-arranges the disturber lines into different subsets, and/or adjusts the decimation factors of the respective carrier groups as well as the dispatching of the downstream SNIR carrier groups to the respective carrier groups. The VCU 130 may also de-select disturber lines that do not (or no longer) noticeably interact with the victim line.

(88) The VCU 130 next re-iterates through the second, third and fourth steps with the newly determined subsets of disturber lines and respective carrier groups until the crosstalk estimates converge.

(89) It can be seen that the algorithm uses parallel rather than successive measurements, and that, after an initial sensing phases, the algorithm tailors the measurements and tone groups such that they focus on the dominant crosstalker(s) and further interpolation and denoising.

(90) If any a priori information is available on the relative crosstalk strength and/or smoothness for some of the disturber lines, then the initial sensing phase can be skipped.

(91) Crosstalk compensation can start at and around the measured tones (possibly with an attenuation to limit the effect of estimation errors). Depending on the smoothness of the channel, crosstalk over a smaller or larger region around the measured tone groups can be compensated. As soon as more measurements become available, a combination of interpolation and denoising is used to improve the crosstalk estimates and to adjust compensation.

(92) For the purpose of interpolation and denoising, weighting may further be used to reduce the impact of measurement errors in the initial measurements.

(93) For mixed vectored and legacy systems, the algorithm can reduce the convergence time of an activating line significantly, as well as the computational complexity of the crosstalk estimation algorithm.

(94) In an alternative embodiment, and still making use of SNIR measurements over the victim line, a crosstalk probing signal is superimposed over each disturber line. The crosstalk probing signal is a weighted replica of the regular signal transmitted over the victim line. The crosstalk probing signals couples into the victim line through the crosstalk channels, and additively or destructively interfere with the direct victim signal depending on the chosen weight values and the crosstalk coupling function. Yet, this method is less preferred as the power of the crosstalk probing signal needs to be substantially increased to achieve the same estimation accuracy, which may severely impair the direct communication over the disturber lines.

(95) In an alternative embodiment, a particular orthogonal pilot sequences assigned to a particular disturber line L.sub.j modulates the carriers of the disjoint carrier groups CG.sub.m assigned to the subset SL.sub.m to which the subscriber line L.sub.j belongs. The remaining carriers that are not member of the carrier group CG.sub.m remain unmodulated, or may carry the SYNC-FLAG information. The resulting crosstalk probing signal is then transmitted over the disturber line L.sub.j during dedicated transmission slots, such as during the SYNC symbols.

(96) Multiple disturber lines may use the same orthogonal pilot sequence, yet this orthogonal pilot sequence will only modulate disjoint groups of carriers over the respective disturber lines.

(97) Slicer errors are measured over the victim line and correlated with the pilot sequence of a given disturber line so as to isolate the crosstalk contribution from that disturber line.

(98) There is seen in FIG. 4 a pictorial view of an exemplary assignment of downstream carriers to the respective disturber lines as per the present invention.

(99) First of all, the communication carriers are split into SNIR carrier groups according to the selected transmission profile. An SNIR carrier group comprise contiguous carriers (=4 in FIG. 4). Let SCG.sub.n* denotes those initial SNIR carrier groups:
SCG.sub.n*={(n);(n)+1; . . . ;((n+1))1}(7)

(100) Next, the SNIR carrier groups are intersected with the downstream MEDLEY set MEDLEY_DS, and any SNIR carrier group that remains unchanged after this intersect operation is kept, while the ones that do not are discarded. The remaining valid SNIR carrier groups are denoted as SCG.sub.n and are renumbered from 1 to .

(101) Next, the N1 disturber lines L.sub.1 to L.sub.N-1 are organized into M subsets SL.sub.1 to SL.sub.M (presently M=N1), and the SNIR downstream carrier groups SCG.sub.n are individually and evenly dispatched toward the M subsets SL.sub.1 to SL.sub.M, thereby yielding disjoint and interleaved set of downstream carriers groups CG.sub.1 to CG.sub.M for parallel crosstalk acquisition.

(102) Within each carrier group CG.sub.m, the inter-carrier spacing between contiguous SNIR carrier groups is typically equal to (N2).Math. (with some discontinuities at the frequency gaps of the transmission profile, if any).

(103) A crosstalk probing signal XTS.sub.j is inserted over the victim line L.sub.N for characterizing the crosstalk channel from a given disturber line L.sub.j, jN. The crosstalk probing signal XTS.sub.j only modulates the carriers of the particular carrier group CG.sub.m assigned to the subset SL.sub.m to which the particular disturber line L.sub.j belongs.

(104) The crosstalk probing signals XTS.sub.j, jN are then frequency-combined into one single crosstalk probing signal

(105) ${\mathrm{XTS}}_N=\underset{j=1\mathrm{.Math.}N-1}{\overset{}{.Math.}}{\mathrm{XTS}}_j$
that is superimposed over the regular signal transmitted over the victim line L.sub.N.

(106) The induced SNIR changes are detected and measured at the victim CPE 200.sub.N, and a single average SNIR value SNIR.sub.n is computed for each and every SNIR downstream carrier group SCG.sub.n. The average SNIR values SNIR.sub.n are reported back to the DSLAM 100 for characterization of the crosstalk channels.

(107) FIG. 5 shows a plot of the mean rate loss versus the carrier group sub-sampling factor for a DSL system with eight subscriber lines: the more sub-sampling, the more inter-carrier spacing, the less accurate the interpolation, the more residual crosstalk, and thus the more rate loss compared to the nominal rate figures without any decimation. Yet, even for a very basic interleaving scheme, the rate-loss penalty is relatively low for sufficient smooth crosstalk channels.

(108) Although the above description mostly focuses on acquisition of downstream crosstalk coefficients, it equally applies to the acquisition of upstream crosstalk coefficients.

(109) For instance, the crosstalk probing signals that are used for estimating the upstream crosstalk coefficients are weighted replicas of the receive signal samples from the respective disturber lines, and are superimposed over the receive signal samples from the victim line by appropriately adjusting the coefficients of the crosstalk cancellation matrix G.

(110) The VCU 130 assigns disjoint sets of upstream carriers to the disturber lines. The upstream carriers are chosen from the upstream MEDLEY set. The SNIR is now measured by the transceiver 110.sub.N, and directly made available to the VCU 130 for upstream crosstalk estimation.

(111) It is to be noticed that the term comprising should not be interpreted as being restricted to the means listed thereafter. Thus, the scope of the expression a device comprising means A and B should not be limited to devices consisting only of components A and B. It means that with respect to the present invention, the relevant components of the device are A and B.

(112) It is to be further noticed that the term coupled should not be interpreted as being restricted to direct connections only. Thus, the scope of the expression a device A coupled to a device B should not be limited to devices or systems wherein an output of device A is directly connected to an input of device B, and/or vice versa. It means that there exists a path between an output of A and an input of B, and/or vice versa, which may be a path including other devices or means.

(113) The description and drawings merely illustrate the principles of the invention. It will thus be appreciated that those skilled in the art will be able to devise various arrangements that, although not explicitly described or shown herein, embody the principles of the invention. Furthermore, all examples recited herein are principally intended expressly to be only for pedagogical purposes to aid the reader in understanding the principles of the invention and the concepts contributed by the inventor(s) to furthering the art, and are to be construed as being without limitation to such specifically recited examples and conditions. Moreover, all statements herein reciting principles, aspects, and embodiments of the invention, as well as specific examples thereof, are intended to encompass equivalents thereof.

(114) The functions of the various elements shown in the figures may be provided through the use of dedicated hardware as well as hardware capable of executing software in association with appropriate software. When provided by a processor, the functions may be provided by a single dedicated processor, by a single shared processor, or by a plurality of individual processors, some of which may be shared. Moreover, a processor should not be construed to refer exclusively to hardware capable of executing software, and may implicitly include, without limitation, digital signal processor (DSP) hardware, network processor, application specific integrated circuit (ASIC), field programmable gate array (FPGA), etc. Other hardware, conventional and/or custom, such as read only memory (Rom), random access memory (RAM), and non volatile storage, may also be included.