Frequency domain training of prefilters for receivers

• US 7,248,849 B1
• Filed: 06/03/2003
• Issued: 07/24/2007
• Est. Priority Date: 06/03/2003
• Status: Active Grant

• Current Assignee: Radia Communications, Inc. (Texas Instruments, Inc.)
• Original Assignee: Texas Instruments, Inc.
• Inventors: Singh, Manoneet, Ariyavisitakul, Sirikiat L.
• Primary Examiner(s): Urban; Edward F.
• Assistant Examiner(s): Dean; Raymond S.
• Application Number:
  US10/454,421
• Time in Patent Office: 1,512 Days
• Field of Search: 455/506, 455/504, 455/133, 455/502, 455/101, 455/503, 455/132, 455/566, 455/63.1, 455/65, 455/67.13, 455/213, 455/286, 455/307, 455/334, 455/339, 375346-350, 375/316, 375/341, 375/343, 375/348
• US Class Current: 455/307
• CPC Class Codes: H04L 25/022 of frequency response
• CPC Class Codes: H04L 25/03261 with impulse-response shortening filters
• CPC Class Codes: H04L 25/03292 with channel estimation circuitry

Abstract

A prefilter is trained as follows. The frequency response B of a conditioned channel is determined without reference to the prefilter, and the frequency response W of the prefilter is computed from the frequency response B of the conditioned channel.

Claims

1. A method for operating a receiver to recover data contained in a signal received from a communications channel, the method comprising:

determining a frequency response of a conditioned channel without reference to a prefilter, wherein the conditioned channel has a delay spread that is shorter than a delay spread of the communications channel;

determining a frequency response of the communications channel;

computing a frequency response of the prefilter from the frequency response of the conditioned channel and the frequency response of the communications channel;

receiving the signal transmitted across the communications channel;

applying the prefilter to the received signal; and

processing the prefiltered signal to estimate the data contained in the signal.

2. The method of claim 1 wherein the step of determining the frequency response of the conditioned channel comprises:

determining the frequency response of the conditioned channel, assuming a length of the prefilter is infinite.

3. The method of claim 1 wherein the step of determining the frequency response of the conditioned channel comprises:

determining an impulse response of the conditioned channel without reference to the prefilter; and

transforming the impulse response of the conditioned channel to obtain the frequency response of the conditioned channel.

4. The method of claim 1 wherein the step of determining a frequency response of a conditioned channel comprises:

determining a conditioned channel as a linear predictive filter of a residual error between (i) an original signal before transmission across the communications channel; and

(ii) a received signal after transmission across the communications channel as filtered by an optimum linear equalizer.

5. The method of claim 4 wherein the step of determining the frequency response of the conditioned channel comprises:

determining a frequency response of the communications channel;

computing a frequency-dependent SNR from the frequency response of the communications channel;

inverse transforming a function of the frequency-dependent SNR to obtain an autocorrelation function;

computing an impulse response of the conditioned channel from the autocorrelation function; and

transforming the impulse response of the conditioned channel to obtain the frequency response of the conditioned channel.

6. The method of claim 1 wherein the step of determining the frequency response of the conditioned channel comprises:

computing the frequency response of the conditioned channel from an estimate of a response of the communications channel.

7. The method of claim 1 wherein the step of determining the frequency response of the communications channel comprises:

computing the frequency response of the communications channel directly from a training signal received over the communications channel.

8. The method of claim 1 wherein the step of determining the frequency response of the conditioned channel is non-iterative.

9. The method of claim 1 wherein the delay spread of the conditioned channel is one symbol duration.

10. The method of claim 1 further comprising:

setting a value for the delay spread of the conditioned channel, wherein the steps of determining a frequency response of a conditioned channel and computing a frequency response of the prefilter are performed assuming the set value for the delay spread.

11. The method of claim 1 wherein the step of processing the prefiltered signal comprises:

processing the prefiltered signal according to maximum likelihood sequence estimation.

12. The method of claim 1 wherein the step of processing the prefiltered signal comprises:

determining a confidence level of the estimate of the data.

13. The method of claim 12 further comprising:

decoding the estimate of the data, based in part on the confidence level of the estimate.

14. The method of claim 13 further comprising:

weighting the confidence level of the estimate of the data by a function of a mean square error between (i) an original signal as filtered by the conditioned channel; and

(ii) the received signal as filtered by the prefilter.

15. The method of claim 1 further comprising:

inverse transforming the frequency response of the prefilter to obtain tap weights for an impulse response of the prefilter;

and wherein the step of applying the prefilter comprises;

applying the tap weights to the received signal.

16. The method of claim 1 wherein the signal transmitted across the communications channel comprises a multicarrier signal;

and wherein the processing step comprises;

demodulating the prefiltered multicarrier signal.

17. The method of claim 16 wherein the multicarrier signal comprises multicarrier blocks separated by guard intervals;

wherein the delay spread of the communications channel is longer than the guard intervals;

and wherein the delay spread of the conditioned channel is shorter than the guard intervals.

18. The method of claim 1 wherein the receiver has at least two input branches.

19. The method of claim 18 wherein:

the step of determining a frequency response of a conditioned channel comprises determining a frequency response of one conditioned channel for all input branches; and

the step of computing a frequency response of the prefilter comprises computing a frequency response of a branch-specific prefilter for each input branch.

20. The method of claim 18 wherein the step of determining a frequency response of a conditioned channel comprises weighting each input branch according to a noise level in a signal received by the input branch.

21. The method of claim 18 wherein the step of determining a frequency response of a conditioned channel comprises:

determining an impulse response of one conditioned channel for all input branches without reference to the prefilters, wherein the conditioned channel is a linear predictive filter of a residual error between (i) an original signal before transmission across the communications channel; and

(ii) a sum of each branch-specific received signal after transmission across the branch-specific communications channel as filtered by a branch-specific optimum linear equalizer; and

transforming the impulse response of the conditioned channel to obtain the frequency response of the conditioned channel;

wherein the step of computing a frequency response of the prefilter comprises, for each input branch;

computing the frequency response of the branch-specific prefilter from the frequency response of the conditioned channel and the frequency response of the branch-specific communications channel in order to optimize a preselected performance metric;

and wherein the method further comprises;

inverse transforming the frequency response of the branch-specific prefilters to obtain branch-specific tap weights for impulse responses of the prefilters;

for each input branch, applying the branch-specific tap weights to a signal received by the input branch;

combining the prefiltered signals from the input branches of the receiver to generate a composite, prefiltered signal; and

processing the composite, prefiltered signal to estimate the data contained in the signals.

22. The method of claim 1 wherein the data is transmitted in packets across the communications channel.

23. The method of claim 22 wherein computing a frequency response of the prefilter occurs not more than once per packet.

24. The method of claim 1 wherein the delay spread of the communications channel is a result of multi-path effects in the communications channel.

25. The method of claim 1 wherein the communications channel is wireless.

26. A receiver for receiving a signal transmitted across a communications channel, the signal containing data, the receiver comprising:

a training module comprising;

a first module for determining a frequency response of a conditioned channel without reference to a prefilter, wherein the conditioned channel has a delay spread that is shorter than a delay spread of the communications channel; and

a second module for computing a frequency response of the prefilter from the frequency response of the conditioned channel and a frequency response of the communications channel;

a prefilter module coupled to the training module for applying the prefilter to the received signal transmitted across the communications channel; and

circuitry for recovering data from the prefiltered signal.

27. The receiver of claim 26 wherein the first module determines the frequency response of the conditioned channel, assuming a length of the prefilter is infinite.

28. The receiver of claim 26 wherein the first module comprises:

a module for determining an impulse response of the conditioned channel without reference to the prefilter; and

an FFT for transforming the impulse response of the conditioned channel to obtain the frequency response of the conditioned channel.

29. The receiver of claim 26 wherein the first module determines a conditioned channel that is a linear predictive filter of a residual error between (i) an original signal before transmission across the communications channel;

and (ii) a received signal after transmission across the communications channel as filtered by an optimum linear equalizer.

30. The receiver of claim 29 wherein the first module comprises:

a module for determining a frequency response of the communications channel;

a first computational module for computing a frequency-dependent SNR from the frequency response of the communications channel;

an inverse FFT for inverse transforming a function of the frequency-dependent SNR to obtain an autocorrelation;

a second computational module for computing an impulse response of the conditioned channel from the autocorrelation function; and

a second FFT for transforming the impulse response of the conditioned channel to obtain the frequency response of the conditioned channel.

31. The receiver of claim 26 wherein the first module computes the frequency response of the communications channel directly from a training signal received over the communications channel.

32. The receiver of claim 26 wherein the first module determines the frequency response of the conditioned channel in a non-iterative manner.

33. The receiver of claim 26 wherein the delay spread of the conditioned channel is one symbol duration.

34. The receiver of claim 26 wherein the training module operates with a set value for the delay spread of the conditioned channel.

35. The receiver of claim 26 wherein the recovering circuitry comprises:

a sequence estimator coupled to the prefilter module for processing the prefiltered signal to estimate the data contained in the signal.

36. The receiver of claim 35 wherein the sequence estimator comprises a maximum likelihood sequence estimator.

37. The receiver of claim 35 wherein the sequence estimator determines a confidence level of the estimate of the data.

38. The receiver of claim 37 further comprising:

a decoder coupled to the sequence estimator for decoding the estimate of the data, based in part on the confidence level of the estimate.

39. The receiver of claim 38 further comprising:

a module for weighting the confidence level of the estimate of the data by a function of a mean square error between (i) the original signal as filtered by the conditioned channel; and

(ii) the received signal as filtered by the prefilter.

40. The receiver of claim 26 further comprising:

an inverse FFT for inverse transforming the frequency response of the prefilter to obtain tap weights for an impulse response of the prefilter, wherein the prefilter module is coupled to receive the tap weights from the inverse FFT.

41. The receiver of claim 26 wherein the signal transmitted across the communications channel comprises a multicarrier signal;

and wherein the recovering circuitry comprises;

a multicarrier demodulator.

42. The receiver of claim 41 wherein the multicarrier signal comprises multicarrier blocks separated by guard intervals;

wherein the delay spread of the communications channel is longer than the guard intervals;

and wherein the delay spread of the conditioned channel is shorter than the guard intervals.

43. The receiver of claim 26 further comprising:

a front-end with at least two input branches.

44. The receiver of claim 43 wherein:

the first module determines a frequency response of one conditioned channel for all input branches; and

the second module computes a frequency response of a branch-specific prefilter for each input branch.

45. The receiver of claim 43 wherein the first module comprises:

a module for determining an impulse response of one conditioned channel for all input branches without reference to the prefilters, wherein the conditioned channel is a linear predictive filter of a residual error between (i) an original signal before transmission across the communications channel; and

(ii) a sum of each branch-specific received signal after transmission across the branch-specific communications channel as filtered by a branch-specific optimum linear equalizer; and

an FFT for transforming the impulse response of the conditioned channel to obtain the frequency response of the conditioned channel;

wherein, for each input branch, the second module computes the frequency response of the branch-specific prefilter from the frequency response of the conditioned channel and the frequency response of the branch-specific communications channel in order to optimize a preselected performance metric;

and wherein the receiver further comprises;

an inverse FFT for inverse transforming the frequency response of the branch-specific prefilters to obtain branch-specific tap weights for impulse responses of the prefilters;

a summer coupled to the branch-specific prefilter modules for combining the prefiltered signals from the input branches of the receiver to generate a composite, prefiltered signal; and

a sequence estimator coupled to the summer for processing the composite, prefiltered signal to estimate the data contained in the signals.

46. The receiver of claim 26 wherein the data is transmitted in packets across the communications channel.

47. The receiver of claim 46 wherein the training module computes a frequency response of the prefilter not more than once per packet.

48. The receiver of claim 26 wherein the training module is adapted for a wireless communications channel.

49. The receiver of claim 26 wherein the training module and the prefilter module are implemented as circuitry on a single integrated circuit.

50. A receiver for receiving a signal transmitted across a communications channel, the signal containing data, the receiver comprising:

means for determining a frequency response of a conditioned channel without reference to a prefilter, wherein the conditioned channel has a delay spread that is shorter than a delay spread of the communications channel;

means for computing a frequency response of the prefilter from the frequency response of the conditioned channel and a frequency response of the communications channel; and

means for applying the prefilter to the received signal; and

means for recovering data from the prefiltered signal.