Channel estimation method and system using fast fourier transforms

• US 20040196782A1
• Filed: 07/11/2003
• Published: 10/07/2004
• Est. Priority Date: 04/04/2003
• Status: Active Grant

• Current Assignee: Interdigital Technology Corporation (InterDigital, Inc.)
• Original Assignee: Interdigital Technology Corporation (InterDigital, Inc.)
• Inventors: Zeira, Ariela, Pan, Jung-Lin, Haim, John W.
• Granted Patent: US 7,639,729 B2
• Time in Patent Office: Days
• US Class Current: 370/210
• CPC Class Codes: H04B 2001/6912 using chirp
• CPC Class Codes: H04L 25/0232 by interpolation between sounding signals
• CPC Class Codes: H04L 27/103 Chirp modulation for spread spectrum techniques H04B1/69
• CPC Class Codes: H04L 27/2647 Arrangements specific to the receiver only equalisation H04L27/01

Abstract

A low cost method and system for efficiently implementing channel estimation in a wireless communication system using any desired length of a fast Fourier transform (FFT) independent of burst type or signal structure. The hardware complexity required to perform the channel estimation to process a plurality of different burst types is reduced. Simple tail zero-padding is used when the length of FFT is extended to a desired length for more efficient computation.

Claims

1. In a wireless communication system, a method of performing channel estimation, the method comprising:

(a) receiving reference signals having different lengths;

(b) processing the reference signals using a fast Fourier transform (FFT); and

(c) extending the FFT to a desired length L for more efficient computation.

2. The method of claim 1 wherein the FFT is extended to the length L to process a plurality of different burst types associated with the reference signals.

3. In a wireless communication system, a method of performing channel estimation, the method comprising:

(a) receiving a time domain signal r;

(b) multiplying, element-to-element, the sequences m and r by a chirp waveform, the chirp waveform being based on the length of the FFT and denoting the resulting sequences as m_W and r_W respectively, where m is a midamble sequence; and

(c) creating a chirp sequence v based on the chirp waveform.

4. The method of claim 3 wherein the chirp waveform is W^n2/2 for n=0,1,2, . . . ,P−

1 where P=456 for burst types 1/3 or P=192 for burst type 2, and

5. The method of claim 4 wherein the chirp sequence v=W⁻

(n−

P+1)2/2 for n=0,1,2, . . . ,2P−

2.

6. A wireless communication system for performing channel estimation, the system comprising:

(a) means for receiving reference signals having different lengths;

(b) means for processing the reference signals using a fast Fourier transform (FFT); and

(c) means for extending the FFT to a desired length L for more efficient computation.

7. The system of claim 6 wherein the FFT is extended to the length L to process a plurality of different burst types associated with the reference signals.

8. A wireless communication system for performing channel estimation, the system comprising:

(a) means for receiving a time domain signal r;

(b) means for multiplying, element-to-element, the sequences m and r by a chirp waveform, the chirp waveform being based on the length of the FFT and denoting the resulting sequences as m_W and r_W respectively, where m is a midamble sequence; and

(c) means for creating a chirp sequence v based on the chirp waveform.

9. The system of claim 8 wherein the chirp waveform is W^n2/2 for n=0,1,2, . . . ,P−

1 where P=456 for burst types 1/3 or P=192 for burst type 2, and

10. The system of claim 9 wherein the chirp sequence v=W⁻

(n−

P+1)2/2 for n=0,1,2, . . . ,2P−

2.

11. A wireless transmit/receive unit (WTRU) for performing channel estimation, the WTRU comprising:

(a) means for receiving reference signals having different lengths;

(b) means for processing the reference signals using a fast Fourier transform (FFT); and

(c) means for extending the FFT to a desired length L for more efficient computation.

12. The WTRU of claim 11 wherein the FFT is extended to the length L to process a plurality of different burst types associated with the reference signals.

13. A wireless transmit/receive unit (WTRU) for performing channel estimation, the WTRU comprising:

(a) means for receiving a time domain signal r;

(b) means for multiplying element-to-element the sequences m and r by a chirp waveform, the chirp waveform being based on the length of the FFT and denoting the resulting sequences as m_W and r_W respectively, where m is a midamble sequence; and

(c) means for creating a chirp sequence v based on the chirp waveform.

14. The WTRU of claim 13 wherein the chirp waveform is W^n2/2 for n=0,1,2, . . . ,P−

1 where P=456 for burst types 1/3 or P=192 for burst type 2, and

15. The WTRU of claim 14 wherein the chirp sequence v=W⁻

(n−

P+1)2/2 for n=0,1,2, . . . ,2P−

2.

16. A base station (BS) for performing channel estimation, the BS comprising:

(a) means for receiving reference signals having different lengths;

(b) means for processing the reference signals using a fast Fourier transform (FFT); and

(c) means for extending the FFT to a desired length L for more efficient computation.

17. The BS of claim 16 wherein the FFT is extended to the length L to process a plurality of different burst types associated with the reference signals.

18. A base station (BS) for performing channel estimation, the BS comprising:

(a) means for receiving a time domain signal r;

(b) means for multiplying element-to-element the sequences m and r by a chirp waveform, the chirp waveform being based on the length of the FFT and denoting the resulting sequences as m_W and r_W respectively, where m is a midamble sequence; and

(c) means for creating a chirp sequence v based on the chirp waveform.

19. The BS of claim 18 wherein the chirp waveform is W^n2/2 for n=0, 1, 2, . . . , 2P−

1 where

P=456 for burst types 1/3 or P=192 for burst type 2, and

$W={\mathrm{<br><br>}}^{-j<br><br>\frac{2<br><br>\mathrm{\pi <br><br>}}{P}}.$

20. The BS of claim 19 wherein the chirp sequence v=W⁻

(n−

P+1)2/2 for n=0, 1, 2, . . . , 2P−

2.

21. In a wireless communication system, a method for performing channel estimation, the method comprising:

(a) receiving a time domain signal r;

(b) expressing r=m{circle over (×

)}h in the frequency domain, resulting in an output signal R=M·

H where m is a midamble sequence, h is a channel impulse response, {circle over (×

)} is a circular convolution operator, R is the fast Fourier transform (FFT) of time domain signal r, M is the FFT of midamble sequence m, and H is the FFT of channel impulse response h, and R=F(r), M=F(m) and H=F(h) where F( ) is defined as the operator of forward or inverse FFT;

(c) calculating H is calculated by dividing R by M, where R/M is the element-to-element division of the corresponding two FFT sequences; and

(d) estimating the impulse response by inverse FFT of H by calculating h=F<sup>−

1</sup>(H) where F<sup>−

1</sup>( ) is defined as the operator of forward or inverse FFT and h=F<sup>−

1</sup>(F(r)/F(m)) and F(r)/F(m) denotes the element-to-element division of FFT sequences F(r) and F(m).

22. The method of claim 21 wherein the forward or inverse FFT are exchangeable in the following form:

23. The method of claim 22 wherein the FFT sequences F(r) and F(m) are calculated by extended FFT and divided element-to-element, the method further comprising:

(e) multiplying, element-to-element, the sequences m and r by a chirp waveform and denoting the resulting sequences as m_W and r_W respectively;

(f) creating a chirp sequence based on the chirp waveform;

(g) zero padding the sequences m_W, r_W and v in the tail until the length of the sequences achieves L, and denoting the resulting sequences as m_W,Z, r_W,Z and

(h) performing an L-point FFT on m_W,Z, r_W,Z and v_Z each such that F(m_W,Z), F(r_W,Z) and F(v_Z);

(i) multiplying, element-to-element, the FFT of m_W,Z and r_W,Z each with FFT of v_Z such that the products are F(m_W,Z)·

F(v_Z) and F(r_W,Z)·

F(v_Z) respectively;

(j) performing an L-point inverse on the results in step (i) such that F<sup>−

1</sup>(F(m_W,Z)·

F(v_Z)) and F<sup>−

1</sup>(F(r_W,Z)·

F(v_Z)) respectively; and

(k) dividing, element-to-element, the results in step (j) and denoting the result as H such that

$\underset{\_}{H}=\frac{F^{-1}(F<br><br>\left({\underset{\_}{r}}_{W,Z}\right)·<br><br>F<br><br>\left({\underset{\_}{v}}_Z\right)}{F^{-1}(F<br><br>\left({\underset{\_}{m}}_{W,Z}\right)·<br><br>F<br><br>\left({\underset{\_}{v}}_Z\right)}$

wherein only the first P elements of sequence H are computed and used.

24. The method of claim 23 wherein F(H*) is computed by extended FFT and the result is conjugated and scaled, the method further comprising:

(l) conjugating the sequence H;

(m) multiplying, element-to-element, the conjugate sequence H* by the chirp waveform and denoting the result as H*_W;

(n) zero padding the conjugate sequences H*_W in the tail until the length of the sequence achieves L, and denoting the resulting sequence as H*_W,Z;

(o) performing an L-point FFT on H*_W,Z;

(p) multiplying, element-to-element, the FFT of H*_W,Z by FFT of zero-padded chirp sequence v_Z such that the product is F(H*_W,Z)·

F(v_Z);

(q) performing L-point inverse FFT on F(H*_W,Z)·

F(v_Z);

(r) multiplying, element-to-element, the sequence F<sup>−

1</sup>(F(H*_W,Z)·

F(v_Z)) by the chirp waveform;

(s) conjugating the result in step (r); and

(t) scaling the result in step (s) by factor

$\frac{1}{P}$

to obtain the estimated channel response.

25. The method of claim 21 wherein the FFT is extended to a proper length L to process a plurality of different burst types by using a chirp transform algorithm (CTA) to compute F(r) and F(m).

30. The system of claim 25 wherein the FFT is extended to a proper length L to process a plurality of different burst types by using a chirp transform algorithm (CTA) to compute F(r) and F(m).

26. A wireless communication system for performing channel estimation, the system comprising:

(a) means for receiving a time domain signal r;

(b) means for expressing r=m{circle over (×

)}h in the frequency domain, resulting in an output signal R=M·

H, where m is a midamble sequence, h is a channel impulse response, {circle over (×

)} is a circular convolution operator, R is the fast Fourier transform (FFT) of time domain signal r, M is the FFT of midamble sequence m, and H is the FFT of channel impulse response h, and R=F(r), M=F(m) and H=F(h) where F( ) is defined as the operator of forward or inverse FFT;

(c) means for calculating H is calculated by dividing R by M, where R/M is the element-to-element division of the corresponding two FFT sequences; and

(d) means for estimating the impulse response by inverse FFT of H by calculating h=F<sup>−

1</sup>(H) where F<sup>−

1</sup>( ) is defined as the operator of forward or inverse FFT and h=F<sup>−

1</sup>(F(r)/F(m)) and F(r)/F(m) denotes the element-to-element division of FFT sequences F(r) and F(m).

27. The system of claim 26 wherein the forward or inverse FFT are exchangeable in the following form:

28. The system of claim 27 wherein the FFT sequences F(r) and F(m) are calculated by extended FFT and divided element-to-element, the method further comprising:

(e) means for multiplying, element-to-element, the sequences m and r by a chirp waveform and denoting the resulting sequences as m_W and r_W respectively;

(f) means for creating a chirp sequence based on the chirp waveform;

(g) means for zero padding the sequences m_W, r_W and v in the tail until the length of the sequences achieves L, and denoting the resulting sequences as m_W,Z, r_W,Z and v_Z;

(h) means for performing L-point FFT on m_W,Z, r_W,Z and v_Z each such that F(m_W,Z), F(r_W,Z) and F(v_Z);

(i) means for multiplying, element-to-element, the FFT of m_W,Z and r_W,Z each with FFT of v_Z such that the products are F(m_W,Z)·

F(v_Z) and F(r_W,Z)·

F(v_Z) respectively;

(j) means for performing an L-point inverse on F(m_W,Z)·

F(v_Z) and F(r_W,Z);

F(v_Z) such that F<sup>−

1</sup>(F(m_W,Z)·

F(v_Z)) and F<sup>−

1</sup>(F(r_W,Z)·

F(v_Z)) respectively; and

(k) means for dividing, element-to-element, F<sup>−

1</sup>(F(m_W,Z)·

F(v_Z)) by F<sup>−

1</sup>(F(r_W,Z)·

F(v_Z)) and denoting the result as H such that

$\underset{\_}{H}=\frac{F^{-1}(F<br><br>\left({\underset{\_}{r}}_{W,Z}\right)·<br><br>F<br><br>\left({\underset{\_}{v}}_Z\right)}{F^{-1}(F<br><br>\left({\underset{\_}{m}}_{W,Z}\right)·<br><br>F<br><br>\left({\underset{\_}{v}}_Z\right)}$

wherein only the first P elements of sequence H are computed and used.

29. The system of claim 28 wherein F(H*) is computed by extended FFT and the result is conjugated and scaled, the system further comprising:

(l) means for conjugating the sequence H;

(m) means for multiplying, element-to-element, the conjugate sequence H* by the chirp waveform and denoting the result as H*_W;

(n) means for zero padding the conjugate sequences H*_W in the tail until the length of the sequence achieves L, and denoting the resulting sequence as H*_W,Z;

(o) means for performing L-point FFT on H*_W,Z;

(p) means for multiplying, element-to-element, the FFT of H*_W,Z by FFT of zero-padded chirp sequence v_Z such that the product is F(H*_W,Z)·

F(v_Z);

(q) means for performing an L-point inverse FFT on F(H*_W,Z)·

F(v_Z);

(r) means for multiplying, element-to-element, the sequence F<sup>−

1</sup>(F(H*_W,Z)·

F(v_Z)) by the chirp waveform;

(s) means for conjugating the output of means (r); and

(t) means for scaling the output of means (s) by factor

$\frac{1}{P}$

to obtain the estimated channel response.