Communication system and methods of estimating channel impulse responses therein

• US 7,149,239 B2
• Filed: 03/29/2002
• Issued: 12/12/2006
• Est. Priority Date: 04/03/2001
• Status: Expired due to Term

• Current Assignee: Malikie Innovations Limited (Key Patent Innovations Limited)
• Original Assignee: Nortel Networks Limited (Nortel Networks Corporation)
• Inventors: Hudson, John E
• Primary Examiner(s): Bayard, Emmanuel
• Assistant Examiner(s): Williams, Lawrence B.
• Application Number:
  US10/109,915
• Publication Number:
  US 20030043887A1
• Time in Patent Office: 1,719 Days
• Field of Search: 375/144, 375/229, 375/347, 370/342, 455/137, 700/52
• US Class Current: 375/144
• CPC Class Codes: H04L 1/0618 Space-time coding
• CPC Class Codes: H04L 2025/03414 Multicarrier
• CPC Class Codes: H04L 25/0212 of impulse response
• CPC Class Codes: H04L 25/0224 using sounding signals
• CPC Class Codes: H04L 25/025 using least-mean-square [LMS] method
• CPC Class Codes: H04L 25/03159 operating in the frequency domain H04L25/03165, H04L25/03178 take precedence

Abstract

Multiple Steiner codes are transmitted as bursts from multiple base stations (182, 184, 186) having one or more transmit elements (174, 176, 178, 180), with successive bursts providing an extended training sequence for use in channel estimation at an addressed unit (172), such as a mobile handset. Accurate channel estimation is possible through the use of Wiener frequency domain MMSE deconvolution (518) combined with frequency domain spatial decoupling matrices, with quasi-orthogonal pseudo-noise sequences (502, 504, 520, 522) allocated to base stations and their antenna elements. The use of Steiner codes to supplement Wiener frequency domain MMSE deconvolution and frequency domain spatial decoupling results in the possibility of allocating only a single training sequence to each base station provided that the training sequence is of sufficient length to encompass all multiple time-translated channel impulse responses (H). Estimates may be refined iteratively by minimising the MS error of demodulated pilot symbols. Estimates may also be refined by removing taps from the impulse response which are insignificant based on a relatively long-term power-delay profile for the channel.

Claims

1. A method of determining channel impulse responses of a plurality of channels to a communication device, the method comprising:

performing transform operations on both a replica of a signal sequence s_n and a received training sequence y_n received by the communication device in at least one burst, the received training sequence y_n being the signal sequence as received through a channel, the transform operations arranged to generate a multiplicity of signal sequence frequency bins and a multiplicity of training sequence frequency bins;

performing point-by-point operations between corresponding signal sequence frequency bins and training sequence frequency bins; and

concatenating the point-by-point operations associated with the channel to provide a composite frequency response for the channel, the composite frequency response allowing, in the time domain, generation of a channel impulse response H for the channel;

wherein multiple Steiner codes are transmitted as training sequences, the multiple Steiner codes sent from multiple transmit elements in multiple training bursts.

2. The method according to claim 1, further comprising:

separating training sequence bursts emanating from a single element transmitter by one of a cyclic prefix and a blank (zero) carrier.

3. The method according to claim 1, wherein the multiple Steiner codes are transmitted from multiple elements of a base station transmit antenna.

4. The method according to claim 1, further comprising using a set of matrix operations in the frequency domain to resolve channels to the communication device from multiple transmitting stations, the matrix operations providing solvable linear equations for the channel impulse response H and the signal sequence expressible in a matrix-vector form at each frequency bin in terms of the received training sequence.

5. The method according to claim 4, further comprising solving the linear equations using a minimum mean square error (MMSE) estimation technique.

6. The method according to claim 5, wherein the MMSE estimation technique employs a Weiner filtering operation providing:

{circumflex over (H)}=(S^HS+Cov(B))<sup>−

1</sup>S^HY

where;

Cov(B) is a covariance matrix of thermal noise;

S^HY is a matched filter operation arranged to minimize the signal to noise ratio but not to remove mutual interference between channel estimates; and

(S^HS+Cov(B))<sup>−

1</sup> is a decoupling matrix that removes the coupling between different channels caused by any non-ideal nature of the training sequences.

7. The method according to claim 2, wherein the number of bursts sent to the communication device from each transmitting unit in communication contact therewith is calculated as a multiplication of:

a number of transmitting elements in a transmit array of a transmitting unit; and

a number of transmitting units in communication contact with the communication device.

8. The method according to claim 7, wherein training sequence burst between transmitting units are time-aligned.

9. A method of determining channel impulse responses of channels incident to a communication device, the method comprising:

transmitting multiple quasi-orthogonal pseudo-noise sequences as bursts from multiple base stations each having at least one transmit element, successive bursts providing an extended training sequence for use in channel estimation at the communication device; and

applying a Wiener frequency domain MMSE deconvolution with frequency domain spatial decoupling matrices to generate channel impulse response estimates for the channels;

wherein the quasi-orthogonal pseudo-noise sequences are Steiner codes.

10. The method according to claim 9, further comprising allocating only a single training sequence to each base station, the single training sequence being of sufficient length to encompass all multiple time-translated channel impulse responses.

11. A method of determining channel impulse responses of a plurality of channels established between a plurality of transmitting elements and a communication device in a communication system, the method comprising:

substantially simultaneously transmitting different training bursts from each of the plurality of transmitting elements, each burst having a length at least as long as a maximum channel duration in the communication system multiplied by a number corresponding to the plurality of transmitting elements;

recovering at the communication device a signal sequence S_n from the different training bursts Y_n; and

resolving the plurality of channels to recover associated channel impulse responses H for each channel by solving an algebraic matrix operation expressed in matrix-vector form as Y=SH, where;

S is a matrix of partial training bursts for each channel, each training burst segmented into N pieces in the time domain;

Y is a vector of a received signal sequence; and

H is a concatenation of different channel impulse response vectors;

wherein multiple Steiner codes are transmitted as training sequences, the multiple Steiner codes sent from multiple transmit elements in multiple training bursts.

12. A computer program product residing in a computer readable medium and for a processor within a receiver device, the computer program product comprising:

code that performs transform operations an both a replica of a signal sequence s_n and a received training sequence y_n received by a communication device in at least one burst, the received training sequence y_n being one of multiple Steiner codes, the multiple Steiner codes sent from multiple transmit elements in multiple training bursts and being the signal sequence as received through a channel, the transform operations arranged to generate a multiplicity of signal sequence frequency bins and a multiplicity of training sequence frequency bins;

code that performs paint-by-point operations between corresponding signal sequence frequency bins and training sequence frequency bins; and

code that concatenates the point-by-point operations associated with the channel to provide a composite frequency response for the channel, the composite frequency response allowing, in the time domain, generation of the channel impulse response for the channel.

13. A communication device having a receiver coupled, in use, to receive a plurality of channels supporting a signal sequence s_n and training sequence y_n bursts, the communication device having:

a signal processing platform to perform transform operations on both a replica of a signal sequence s_n and a received training sequence y_n received by the communication device in at least one burst, the received training sequence y_n being the signal sequence as received through a channel, the transform operations arranged to generate a multiplicity of signal sequence frequency bins and a multiplicity of training sequence frequency bins;

the signal processing platform arranged to perform point-by-point operations between corresponding signal sequence frequency bins and training sequence frequency bins; and

the signal processing platform further arranged to concatenate the point-by-point operations associated with the channel to provide a composite frequency response for the channel, the composite frequency response allowing, in the time domain, generation of a channel impulse response for the channel;

whereby multiple Steiner codes are transmitted as the training sequences, the multiple Steiner codes sent to the receiver through multiple channels in multiple training bursts.

14. The communication device according to claim 13, wherein the signal processing platform operates to establish a set of matrix operations in the frequency domain to resolve channels to the communication device from multiple transmitting stations, the matrix operations providing solvable linear equations for the channel impulse response and the training sequence expressible in a matrix-vector form at each frequency bin.

15. The communication device according to claim 14, wherein the signal processing platform solves the linear equations using a minimum mean square error (MMSE) estimation technique.

16. The communication device according to claim 14, wherein the MMSE estimation technique employs a Weiner filtering operation providing:

{circumflex over (H)}=(S^HS+Cov(B))<sup>−

1</sup>S^HY

where;

Cov(B) is a covariance matrix of thermal noise;

S^HY is a matched filter operation arranged to maximise the signal to noise ratio but not to remove mutual interference between channel estimates; and

(S^HS+Cov(B))<sup>−

1</sup> is a decoupling matrix that removes the coupling between different channels caused by any non-ideal nature of the training sequences.

17. A communication receiver comprising:

means for receiving, in use, multiple quasi-orthogonal pseudo-noise sequences as bursts from multiple base stations each having at least one transmit element, successive bursts providing an extended training sequence for use in channel estimation at the communication receiver; and

means for applying a Wiener frequency domain MMSE deconvolution with frequency domain spatial decoupling matrices to generate channel impulse response estimates for the channels;

wherein the quasi-orthogonal pseudo-noise sequences are Steiner codes.

18. A communication device operational to receive a plurality of training sequences on a plurality of channels and a signal sequence S_n, the communication device comprising:

a receiver for substantially simultaneously receiving, in use, different training bursts from each of the plurality of channels emanating from a plurality of transmit elements, each burst having a length at least as long as a maximum channel duration multiplied by a number corresponding to the plurality of transmit elements;

recovery circuitry for recovering, in use, the signal sequence S_n from the different training bursts Y_n; and

a processor arranged to resolve the plurality of channels to recover associated channel impulse responses H for each channel by solving an algebraic matrix operation expressed in matrix-vector form as Y=SH, where;

S is a matrix of partial training bursts for each channel, each training burst segmented into N pieces in the time domain;

Y is a vector of a received signal sequence; and

H is a concatenation of different channel impulse response vectors;

wherein multiple Steiner codes are transmitted as training bursts, the multiple Steiner codes sent from multiple transmit elements in multiple training bursts.

19. A base station of a communication system, the base station comprising:

a transmitter chain arranged to transmit multiple quasi-orthogonal pseudo-noise sequences as training bursts Y_n from at least one transmit element and further arranged to transmit a signal sequence S_n, successive training bursts providing an extended training sequence for use in channel estimation at a communication device of the communication system, the transmitter chain substantially simultaneously transmitting, in use, different training bursts from each of the at least one transmit element, each training burst having a length at least as long as a maximum channel duration in the communication system multiplied by a number corresponding to a plurality of channels to the communication device, the extended training sequence and the signal sequence S_n providing a resolution mechanism to the communication device allowing the communication device to resolving the plurality of channels to recover associated channel impulse responses H for each channel by solving an algebraic matrix operation expressed in matrix-vector form as Y=SH, where;

S is a matrix of partial training bursts for each channel, each training burst segmented into N pieces in the time domain;

Y is a vector of a received signal sequence; and

H is a concatenation of different channel impulse response vectors;

multiple Steiner codes are transmitted as training bursts, the multiple Steiner codes sent from multiple transmit elements in multiple training bursts.

20. The base station of claim 19, wherein the transmitter chain is further arranged to separate training bursts emanating from the at least one transmit element by one of a cyclic prefix and a blank (zero) carrier.

21. The base station of claim 19, wherein the number of training bursts sent to the communication device from each base station in communication contact therewith is calculated as a multiplication of:

a number of transmitting elements in a transmit array of a base station; and

a number of base stations in communication contact with the communication device.

22. A method of estimating a channel impulse response comprising:

(a) receiving a signal including a training sequence of predetermined training symbols,

(b) passing the received signal through a channel equaliser to substantially remove distortion of the signal caused by transmission over a channel,

(c) demodulating at least one of the training symbols,

(d) calculating a non quadratic, non linear error metric of the demodulated training symbol against a locally stored record of a known correct value of the demodulated training symbol

(a) adjusting the estimated channel impulse response to substantially minimise a mean square error, and

(f) feeding the adjusted estimated channel impulse response back to the channel equaliser for use in subsequent equalisation operations;

whereby channel estimates used in the channel equaliser are iteratively refined; and

multiple transmissions are received over a plurality of channels each including a training sequence and wherein a summed mean square error of all training sequences is minimised by adjusting the estimated channel response h for each tap of each channel impulse response in turn according to the equation

θ

(h)=(h−

h_OPT)^HM(h−

h_OPT)+c

where h is a vector of channel impulse responses for all channels, h_OPT is an optimum channel impulse response, M is a matrix representing cross-coupling between channel components and a is the minimum error.

23. A method according to claim 22, wherein the training symbol error metric is calculated as a mean square error.

24. A method according to claim 22, wherein the received signal is a CDMA signal and wherein the training sequence is despread using at least one different spreading factor to recover training sequence bits or sub-bits.

25. A method according to claim 22, wherein multiple transmissions are received over a plurality of channels each including a training sequence and wherein the summed mean square error of all training sequences is minimised by adjusting the estimated channel response h using a technique selected from gradient descent methods, search methods, and direct solution algorithms.

26. A method according to claim 22, wherein θ

(h) is minimised by calculating θ

(h) and θ

(h±

δ

) where ±

δ

is a relatively small symmetric perturbation about h, fitting a curve to three points generated thereby and finding the minimum of a fitted curve in terms of h.

27. A method according to claim 22 wherein the channel equaliser is a Wiener (MMSE) equaliser.

28. A method according to claim 22, wherein a first estimate of the channel impulse response is generated by:

performing transform operations on both a replica of a signal sequence s_n and a received training sequence y_n received by the communication device in at least one burst, the received training sequence y_n being the signal sequence as received through a channel, the transform operations arranged to generate a multiplicity of signal sequence frequency bins and a multiplicity of training sequence frequency bins;

performing point-by-point operations between corresponding signal sequence frequency bins and training sequence frequency bins; and

concatenating the point-by-point operations associated with the channel to provide a composite frequency response for the channel, the composite frequency response allowing, in the time domain, generation of a first estimate of the channel impulse response for the channel.

29. A method according to claim 22, wherein the estimated impulse response is further refined by obtaining a long-term averaged power delay profile for the channel, setting a predetermined variance threshold, estimating a channel impulse response having a plurality of taps, and removing taps from the channel impulse response estimate equivalent to those which in the long-term averaged power delay profile, have a variance below the predetermined variance threshold.

30. A method according to claim 29 wherein the predetermined variance threshold is a power threshold.

31. A method according to claim 29 wherein the predetermined variance threshold is an accuracy threshold.

32. A method of estimating a channel impulse response comprising:

(a) obtaining a long-term averaged power delay profile for the channel,

(b) setting a predetermined variance threshold,

(c) estimating a channel impulse response having a plurality of taps, and

d) removing taps from the channel impulse response estimate equivalent to those which in the long-term averaged power delay profile, have a variance below the predetermined variance thresholds;

wherein the channel impulse response estimate is combined with an accuracy of the channel impulse response estimate as predicted point-by-point from a combined power delay profile end interference level, whereby a modified short term, channel estimate is produced in which points likely to have poor accuracy are deweighted and those likely to have good accuracy are retained such that a final equalise performance is optimised.

33. A method according to claim 32, wherein the predetermined variance threshold is dependent on the level of interference in the channel.

34. A method according to claim 32, wherein the long term power delay profile is obtained based on knowledge of a likely maximum time delay in any given channel.

35. A method according to claim 32, wherein the long term power delay profile is obtained using a simple estimate of the channel impulse response.

36. A method according to claim 32, wherein the long term power delay profile is obtained from historical channel impulse responses for the channel.

37. A method according to claim 32, wherein the long term power delay profile is obtained by deriving a channel impulse response from an autocorrelation function of the received signal.

38. A method according to claim 32, wherein the modified short term channel estimate is produced using a Bayesian method.

39. A method according to claim 32, wherein the channel impulse response Ĥ

is estimated according to the equation

$\begin{array}{c}{\hat{H}}_{\mathrm{MMSE}}=<br><br>{\mathrm{Diag}<br><br>\left(\frac{\mathrm{\lambda <br><br>}+{\mathrm{\rho <br><br>}}_k}{\mathrm{\lambda <br><br>}<br><br><br><br>{\mathrm{\rho <br><br>}}_k}\right)}^{-1}<br><br>\mathrm{Diag}<br><br>\left(\frac{\mathrm{\lambda <br><br>}+{\mathrm{\sigma <br><br>}}^2}{\mathrm{\lambda <br><br>}<br><br><br><br>{\mathrm{\sigma <br><br>}}^2}\right)<br><br>{\hat{H}}_1\\ =<br><br>\mathrm{Diag}<br><br>\left(\frac{{\mathrm{\rho <br><br>}}_k}{{\mathrm{\sigma <br><br>}}^2}·<br><br>\frac{\mathrm{\lambda <br><br>}+{\mathrm{\sigma <br><br>}}^2}{\mathrm{\lambda <br><br>}+{\mathrm{\rho <br><br>}}_k}\right)<br><br>{\hat{H}}_1\end{array}$

where ρ

_k is a long-term standard deviation of each channel estimate tap which provides a long-term power delay profile for the channel, λ

is the estimation standard deviation of a least squares estimator, and σ

is the standard deviation of thermal noise in the channel.

40. A method of estimating a channel impulse response comprising:

(a) taking long-term measurements to build a database of where in time, multipath interference is located in the channel to generate an incoherent power delay profile for the channel, and

(b) modelling a relatively short term channel impulse response by allocating variable complex amplitudes and phases to taps at locations in the short term model which have the multipath interference in the incoherent power delay profile which is above a predetermined threshold.

41. A method according to claim 40, wherein the channel impulse response having a plurality of taps is estimated by performing transform operations on both a replica of a signal sequence s_n and a received training sequence y_n received by the communication device in at least one burst, the received training sequence y_n being the signal sequence as received through a channel, the transform operations arranged to generate a multiplicity of signal sequence frequency bins and a multiplicity of training sequence frequency bins;

performing point-by-point operations between corresponding signal sequence frequency bins and training sequence frequency bins; and

concatenating the point-by-point operations associated with the channel to provide a composite frequency response for the channel, the composite frequency response allowing, in the time domain, generation of the channel impulse response for the channel.

42. A method according to claim 41 wherein the channel impulse response estimates are refined by receiving a signal including a training sequence of predetermined training symbols, passing the received signal through a channel equaliser to substantially remove distortion of the signal caused by transmission over a channel, demodulating at least one of the training symbols, calculating the mean square error of the demodulated training symbol against a locally stored record of a known correct value of the demodulated training symbol, adjusting the estimated channel impulse response to substantially minimise the mean square error, and feeding the adjusted estimated channel impulse response back to the channel equaliser for use in subsequent equalisation operations whereby the channel estimates used in the channel equaliser are iteratively refined.

43. A computer program product residing in a computer readable medium which when executed by a computer causes the computer to carry out the steps of:

(a) receiving a signal including a training sequence of predetermined training symbols,

(b) passing the received signal through a channel equaliser to substantially remove distortion of the signal caused by transmission over a channel,

(c) demodulating at least one of the training symbols,

(d) calculating a non quadratic, non linear error metric of the demodulated training symbol against a locally stored record of a known correct value of the demodulated training symbol,

(e) adjusting an estimated channel impulse response to substantially minimise the mean square error, and

(f) feeding the adjusted estimated channel impulse response back to the channel equaliser for use in subsequent equalisation operations

whereby channel estimates used in the channel equaliser are iteratively refined; and

wherein multiple transmissions are received over a plurality of channels each including a training sequence and wherein a summed mean square error of all training sequences is minimised by adjusting the estimated channel response h for each tap of each channel impulse response in turn according to the equation

θ

(h)=(h−

h_OPT)^HM(h−

h_OPT)+c

where h is the vector of channel impulse responses for all channels, h_OPT is the optimum channel impulse response, M is a matrix representing cross-coupling between channel components and c is the minimum error.

44. A product according to claim 43, wherein the training symbol error metric is calculated as a mean square error.

45. A product according to claim 43 including discarding channel estimate taps which have poor accuracy as predicted from a channel power delay profile.

46. A computer program product residing in a computer readable medium which when executed by a computer causes the computer to carry out the steps of:

(a) obtaining a long-term averaged power delay profile for the channel,

(b) setting a predetermined variance threshold,

(c) estimating a channel impulse response having a plurality of taps, and

(d) removing taps from the channel impulse response estimate equivalent to those which in the long-term averaged power delay profile, have a variance below the predetermined variance threshold;

wherein the channel impulse response estimate is combined with the accuracy of the channel impulse response as predicted point-by-point from the combined power delay profile and interference level, whereby a modified short tern channel estimate is produced in which points likely to have poor accuracy are deweighted and those likely to have good accuracy are retained such that final equalise performance is optimised.

47. A product according to claim 46 including iterative adjustment of the channel estimates to minimise a mean square error on at least one training symbol.

48. A communications receiver arranged to receive signals from a plurality of sources and arranged to estimate a channel impulse response for a channel between the receiver and each source by:

(a) receiving a signal including a training sequence of predetermined training symbols,

(b) passing the received signal through a channel equaliser to substantially remove distortion of the signal caused by transmission over a channel,

(c) demodulating at least one of the training symbols,

(d) calculating a non quadratic, non linear error metric of the demodulated training symbol against a locally stored record of its known correct value,

(e) adjusting an estimated channel impulse response to substantially minimise the mean square error, and

(f) feeding the adjusted estimated channel impulse response back to the channel equaliser for use in subsequent equalisation operations;

whereby the channel estimates used in the channel equaliser are iteratively refined; and

multiple transmissions are received over a plurality of channels each including a training sequence and wherein a summed mean square error of all training sequences is minimised by adjusting the estimated channel response h for each tap of each channel impulse response in turn according to the equation

θ

(h)=(h−

h_OPT)^HM(h−

h_OPT)+c

where h is a vector of channel impulse responses for all channels, h_OPT is an optimum channel impulse response, M is a matrix representing cross-coupling between channel components and c is the minimum error.

49. A receiver according to claim 48 including discarding channel estimate taps which have poor accuracy as predicted from a channel power delay profile.

50. A receiver according to claim 48, wherein the training symbol error metric is calculated as a mean square error.

51. A communications receiver arranged to receive signals from a plurality of sources and arranged to estimate a channel impulse response for the channel between the receiver and each source by:

(a) obtaining a long-term averaged power delay profile for the channel,

(b) setting a predetermined variance threshold,

(c) estimating a channel impulse response having a plurality of taps, and

(d) removing taps from the channel impulse response estimate equivalent to those which in the long-term averaged power delay profile, have a variance below the predetermined variance threshold;

wherein the channel impulse response estimate is combined with the accuracy of the channel impulse response as predicted point-by-point from a combined power delay profile and interference level, whereby a modified short term channel estimate is produced in which points likely to have poor accuracy are deweighted and those likely to have good accuracy are retained such that final equalise performance is optimised.

52. A receiver according to claim 51 including iterative adjustment of the channel estimates to minimise the mean square error on at least one training symbol.