Crosstalk identification for spectrum management in broadband telecommunications systems
View Patent ↗Crosstalk between subscriber loops used to transmit different broadband services through the same bundled telephone cable is a significant limitation to providing digital subscriber line services. A method for estimating the crosstalk, identifying the sources of the crosstalk and predicting additional sources of crosstalk are disclosed. The crosstalk sources are identified in the frequency domain by maximizing the correlation with a “basis set” of received crosstalk PSDs which consist of the cascade of a finite set of known transmit PSDs types times a representative set of crosstalk couplings. Multiple crosstalk types are identified with a technique of successive spectral subtraction. Once a type is identified the crosstalk disturber is compared against all other members of the complete set for that type. Additionally, Mutliple Regression (MR) techniques and a Matching Pursuit (MP) algorithm are used to increase the ability of the system and method to identify various crosstalk disturbers. The identification of the crosstalk disturber is then used to perform crosstalk cancellation and spectrum management in a DSL system.
1. A method for identifying the source of crosstalk disturbance in a subscriber loop comprising the steps of:
measuring the power spectral density of the noise present on a subscriber loop;
correlating the power spectral density for said subscriber loop with a predetermined set of power spectral densities for a group of possible crosstalk disturbers; and,
selecting the crosstalk disturber having the most closely correlated power spectral density; and,
refining the selection of the crosstalk disturber using a multiple regression technique.
2. The method of claim 1 further comprising the steps of:
subtracting the power spectral density for the identified crosstalk disturber from the measured power spectral density of said subscriber loop using spectral subtraction to generate a residual power spectral density;
correlating the residual power spectral density with the predetermined set of power spectral densities for the group of possible crosstalk disturbers; and,
selecting the crosstalk disturber having the most closely correlated power spectral density to the residual power spectral density, and
refining the selection of the crosstalk disturber using a multiple regression technique.
3. The method of claim 2 wherein the steps of subtracting, correlating, selecting and refining are iteratively performed until a stopping criterion is reached.
4. The method of claim 3 wherein the stopping criterion is a predetermined correlation threshold.
5. The method of claim 2 wherein negative power spectral densities resulting from said subtracting step are mapped into a non-negative value.
6. The method of claim 2 wherein the disturber type is identified using a reduced crosstalk basis set wherein the matrix containing the crosstalk basis set is partitioned into sub-matrices pertaining to each type of possible crosstalk disturber, and a singular value decomposition is performed on each sub-matrix to produce a reduced crosstalk basis set.
7. The method of claim 1 further comprising the step of sending the identity of the selected crosstalk disturber to a system for cancellation of the crosstalk disturber.
8. The method of claim 1 further comprising the step of sending the estimated pair-to-pair coupling of the selected crosstalk disturber to a system for cancellation of the crosstalk disturber.
9. The methods of claim 1 wherein the disturber type is identified using a reduced crosstalk basis set wherein the matrix containing the crosstalk basis set is partitioned into sub-matrices pertaining to each type of possible crosstalk disturber, and a singular value decomposition is performed on each sub-matrix to produce a reduced crosstalk basis set.
10. A method for identifying crosstalk disturbers on a subscriber loop comprising the steps of:
measuring the power spectral density of the noise present on a subscriber loop;
matching the power spectral density for said subscriber loop with a predetermined set of power spectral densities for a group of possible crosstalk disturbers using multiple regression to identify crosstalk disturbers.
11. The method of claim 10 wherein all non-zero coefficients in the multiple regression must be positive.
12. The method of claim 10 wherein in the application of the multiple regression technique selected crosstalk disturbers are disqualified from the model in pre-specified groups based upon a group's total power contribution to the modeled spectra.
13. The method of claim 10 wherein the disturber type is identified using a reduced crosstalk basis set wherein the matrix containing the crosstalk basis set is partitioned into sub-matrices pertaining to each type of possible crosstalk disturber, and a singular value decomposition is performed on each sub-matrix to produce a reduced crosstalk basis set.
14. In a system for identifying the source of crosstalk disturbance in a subscriber loop wherein the measured power spectral density of the noise on the subscriber loop is compared with a predetermined set of power spectral densities for a group of possible disturbers resulting from use of a group of various types of DSL signal, the method of determining the predetermined crosstalk basis set of power spectral densities for a group of possible disturbers comprising the steps of:
transmitting a first type of DSL signal over a test loop in a bundle of test loops;
measuring the pair-to-pair crosstalk couplings to generate a set of canonical crosstalk couplings (PSDs);
repeating the transmitting and measuring steps for each test loop;
reiterating the transmitting, measuring and repeating steps for each DSL type in the group of various types of DSL signals; and
multiplying the power spectral density for each type of DSL signal by the set of canonical crosstalk couplings (PSDs) to generate a matrix consisting the crosstalk basis set of power spectral densities for each a possible disturber.
15. The method of claim 14 wherein the matrix containing the crosstalk basis set is partitioned into sub-matrices pertaining to each type of possible disturber, and a singular value decomposition is performed on each sub-matrix to produce a reduced crosstalk basis set.
16. A method for identifying crosstalk disturbers on a subscriber loop based on the availability of a crosstalk basis set comprising the steps of:
measuring the power spectral density of the noise present on a subscriber loop;
performing a singular value decomposition on the matrix of the crosstalk basis set to reduce its size;
identifying the disturbers for said subscriber loop with said reduced size crosstalk basis set using a multiple regression technique;
identifying the closest pair-to-pair coupling pertaining to the disturber using a second mathematical technique on the portion of the original crosstalk basis set pertaining to said identified disturber types.