Frequency domain signal reconstruction compensating for phase adjustments to a sampling signal

• US 5,793,801 A
• Filed: 07/09/1996
• Issued: 08/11/1998
• Est. Priority Date: 07/09/1996
• Status: Expired due to Term

• Current Assignee: Telefonaktiebolaget LM Ericsson
• Original Assignee: Telefonaktiebolaget LM Ericsson
• Inventors: Fertner, Antoni
• Primary Examiner(s): Pham, Chi H.
• Assistant Examiner(s): LUTHER, WILLIAM A
• Application Number:
  US08/679,387
• Time in Patent Office: 763 Days
• Field of Search: 375/233, 375/232, 375/230, 375/229, 375/354, 375/219, 375/340, 375/346, 375/348, 375/350, 364/724.19, 364/724.2, 333/28 R, 333/28 T, 333/166, 333/167, 333/18, 379/406, 379/410, 379/411, 348/392, 348/424, 348/425
• US Class Current: 375/219
• CPC Class Codes: H04B 3/23 using a replica of transmitted signal in the time domain, e.g. echo cancellers

Abstract

In a digital communications system, a sampled signal is precisely reconstructed (rather than approximated) even in situations where the phase of the sampling signal is adjusted. More particularly, the sampled signal is compensated in the frequency domain for phase adjustments to the sampling instance in the time domain. In the context of echo cancellation, an echo transfer function is generated representing the echo channel impulse response. Aliased components present in the echo impulse response are specifically identified and compensated for in the frequency domain by treating each spectral filter coefficient of the echo transfer function as the sum of a baseband component and an aliased component. In addition to accurately compensating for sampling phase adjustments, this technique considerably relaxes traditional sampling constraints.

Claims

1. In a digital communications system where transceivers communicate over a communications channel, each transceiver sampling a signal received over the communications channel in accordance with a sampling signal and including an echo canceller for canceling an echo implemented as a digital filter, a method comprising the steps of:

inputting to the digital filter a signal transmitted from the transceiver and delaying the input signal through a plurality of delay stages to generate a series of input signals;

transforming the series of input signals into spectral components to create series of input spectral components;

multiplying each of the spectral components by a corresponding spectral filter coefficient to generate a corresponding product, the spectral filter coefficients representing an echo transfer function;

summing the products corresponding to each of the spectral components to generate an output signal from the digital filter corresponding to an expected echo;

subtracting the filter output signal from the received signal to cancel echo from the received signal;

determining an error based on an uncancelled echo;

adapting the spectral filter coefficients of the adaptive filter to minimize the error;

detecting that the phase of the sampling signal will be adjusted;

predicting new values of the spectral filter coefficients in the frequency domain to compensate for the sampling signal adjustment including analyzing each spectral filter coefficient as having at least two spectral components with a first spectral component including baseband frequency components and a second spectral component including aliased frequency components;

adjusting the phase of the sampling signal; and

using the predicted values of the spectral filter coefficients so that the digital filter output signal accurately models the received signal when the phase of the sampling signal is adjusted.

2. The method in claim 1, wherein the predicting step includes correcting the spectral filter coefficients in the frequency domain by an amount corresponding to the amount of time the sampling signal is adjusted.

3. The method in claim 1, further comprising:

determining the values of the first and second spectral components for each spectral filter coefficient, and

using the determined values of the first and second spectral components to generate the predicted value of each spectral filter coefficient.

4. The method of claim 3, wherein the predicting step includes the step of changing in the frequency domain the first and second spectral components of each spectral filter coefficient by an amount corresponding to the amount of time the sampling signal is adjusted.

5. The method in claim 4, wherein each spectral filter coefficient is predicted using two previously obtained values of the spectral filter coefficient, calculating the first and second spectral components for each spectral filter coefficient based on the two previously obtained values, arid then performing the changing step.

6. An adaptive echo canceller implemented in the frequency domain for use in a digital communications system where transceivers communicate over a communications channel, each transceiver sampling a signal received over the communications channel in accordance with a timing signal, comprising:

delay stages that receive a transmitted signal to generate a series of input signals;

a Fourier transformer that transforms the series of input signals into a frequency domain to create series of input spectral components;

a multiplier multiplying each of the spectral components by a corresponding spectral filter coefficient to generate a corresponding product;

a summer that sums the products corresponding to each of the spectral components to generate a summed signal corresponding to an expected echo;

a combiner that subtracts the summed signal from a signal received over the communications channel to cancel echo from the received signal and that generates an error based on an uncancelled echo;

a signal processor that adapts the spectral filter coefficients of the adaptive filter to minimize the error;

a compensator that analyzes each spectral filter coefficient as having at least two signal components with a first spectral component including baseband frequency components corresponding to the echo and a second spectral component including aliased frequency components, predicts new values of the spectral filter coefficients in the frequency domain corresponding to an amount of time the timing signal is to be adjusted, and provides the predicted values of the spectral filter coefficients to the one or more multipliers and the signal processor so that the summed signal accurately models an echo in a signal received by the transceiver when the timing signal is adjusted.

7. The echo canceller according to claim 6, wherein each compensator modifies the spectral filter coefficients in the frequency domain by an amount corresponding to the amount of time the timing signal is adjusted.

8. The echo canceller according to claim 6, wherein the compensator determines the values of the first and second spectral components for each spectral filter coefficient, and uses the determined values of the first and second spectral components to generate the predicted value of each spectral filter component.

9. The echo canceller according to claim 8, wherein the compensator corrects the first and second spectral components of each spectral filter coefficient in the frequency domain by an amount corresponding to the amount of time the timing signal is adjusted.

10. The echo canceller according to claim 9, wherein each spectral filter coefficient is predicted using two previously obtained values of the spectral filter coefficient, calculating the first and second spectral components for the spectral filter coefficient based on the two previous values of the spectral filter coefficient, and then performing the correction of the first and second spectral components.

11. The echo canceller according to claim 7, wherein the compensator includes:

a first filter for receiving the amount of time that the timing signal is being changed;

a memory storing a set of coefficients for correcting the spectral filter coefficients;

a selector for receiving an input from the first filter to generate an address to the memory; and

a plurality of second filters, each second filter corresponding to and receiving as an input one of the spectral filter coefficients including;

a delay for delaying the input signal;

a first multiplier for multiplying the delayed input signal by one of the coefficients selected from the memory;

a second multiplier for multiplying the input signal by another of the coefficients selected from the memory; and

a summer for summing the products from the first and second multipliers, the sum being the predicted spectral filter coefficient.

12. In a digital communications system where transceivers communicate over a communications channel, a method implemented in one or more of the transceivers comprising the steps of:

sampling a signal received over the communications channel in accordance with a sampling signal;

developing a transfer function corresponding to the impulse response of the communications channel; and

compensating the transfer function in the frequency domain for a phase adjustment in the sampling signal that occurs in the time domain including correcting the spectral coefficients of the transfer function in the frequency domain by an amount corresponding to the adjustment in the sampling signal,

wherein aliased signal content is taken into account in correcting spectral coefficients of the transfer function by identifying for spectral coefficients of the transfer function at least two spectral components, a first spectral component including baseband frequency components and a second spectral component including aliased frequency components.

13. The method in claim 12, wherein the transfer function corresponds to a Fourier transform of an echo impulse response of the communications channel.

14. The method in claim 12, the method further comprising:

determining the values of the first and second spectral components for each spectral coefficient, and

using the determined values of the first and second spectral components to generate the predicted value of each spectral coefficient.

15. In a digital communications system where transceivers communicate over a communications channel, each transceiver sampling a signal received over the communications channel and sampled in accordance with a timing signal, the received signal having frequency components greater than one half of the frequency of the timing signal, and including an adaptive digital filter modeling an impulse response of the communications channel, a method of operating the adaptive digital filter including determining aliased frequency components generated as a result of the sampling frequency being less than twice as high as the highest frequency in the sampled signal in the modeling of the communications channel impulse response and modifying filter coefficients of the adaptive digital filter using the determined, aliased frequency components.

16. The method in claim 15, wherein the filter coefficients represent the communications channel model, and each filter coefficient is processed using a first vector component representing nonaliased frequency components and a second vector component representing aliased frequency components.

17. The method in claim 16, wherein at least one of the filter coefficients is adjusted taking into account both its first and second vector components.

18. The method in claim 16, wherein the adaptive digital filter is recursively adapted to model a transfer function of the communications channel, the method further comprising:

converging the adaptive digital filter a first and second time to generate first and second relationships of filter coefficients representing the communications channel impulse response at first and second times, and

solving a system of two equations using the first and second relationships to determine values of the filter coefficients.

19. In a digital communications system where transceivers communicate over a communications channel, a method implemented in one of the transceivers comprising the step of:

sampling a signal received over the communications channel in accordance with a timing signal having a sampling rate that is less than twice the highest frequency component of the received signal;

identifying in the sampled signals aliased frequency components generated as a result of the timing signal having a frequency that is less than twice the highest frequency component of the received signal; and

reconstructing the received signal from the signal samples using the identified aliased frequency components.

20. The method in claim 19, wherein the reconstructed signal corresponds to a transfer function in the frequency domain of the impulse response of the communications channel, the method further comprising the step of:

compensating the transfer function in the frequency domain for a phase adjustment to the timing signal that occurs in the time domain.