Method and apparatus for channel identification using incomplete or noisy information
First Claim
1. A method of communication channel identification comprising the steps of:
- transmitting a test signal x over a communication channel to be identified, the test signal x having a signal span;
receiving a signal y, wherein signal y comprises test signal x after test signal x has passed through the communication channel; and
calculating a sequence of channel values corresponding to an estimated channel impulse response, the channel impulse response providing an estimate of the communication channel, wherein the sequence of channel values comprises a sampled discrete channel of a finite number of non-zero indices in an interval defmed by -L, M!, wherein the channel impulse response has an impulse response span equal to M+L, the impulse response span being less than the span of the test signal x, wherein said step of calculating the sequence of channel values further comprises using a least-squares (LS) estimator.
2 Assignments
0 Petitions
Accused Products
Abstract
A method and apparatus for channel identification utilizing two Least-Squares (LS) estimators. Each LS estimator is used for calculating a sequence of channel values, further for determining an estimated channel impulse response, over an entire frequency band thereof in light of the fact that information is incomplete or unreliable over part of the frequency band. Each LS estimator operates for the case when the estimated channel impulse response span is less than the span of a known test signal, the test signal having been transmitted over the channel for use in identifying the channel. In a TV ghost-cancellation system for removal of channel induced distortion from received signals, each LS estimator is used to compute channel impulse response coefficients, wherein the system includes ghost-cancellation filters responsive to the channel impulse response for removing the effects of the channel from the signals.
61 Citations
48 Claims
-
1. A method of communication channel identification comprising the steps of:
-
transmitting a test signal x over a communication channel to be identified, the test signal x having a signal span; receiving a signal y, wherein signal y comprises test signal x after test signal x has passed through the communication channel; and calculating a sequence of channel values corresponding to an estimated channel impulse response, the channel impulse response providing an estimate of the communication channel, wherein the sequence of channel values comprises a sampled discrete channel of a finite number of non-zero indices in an interval defmed by -L, M!, wherein the channel impulse response has an impulse response span equal to M+L, the impulse response span being less than the span of the test signal x, wherein said step of calculating the sequence of channel values further comprises using a least-squares (LS) estimator. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12)
- 3. The method of communication channel identification according to claim 1, wherein calculating the sequence of channel values still further comprises using a least-squares (LS) estimator hB given by
- space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
T.sup.T y,
where (TT T).sup.†
TT is approximately equal to T.sup.†
, wherein T.sup.†
is a Moore-Penmore (approximate) inverse of T and T is a Toeplitz matrix given by ##EQU8## further wherein xn, for n=0 to (N+L-1), is a finite sequence of sampled values of a reference test signal XREF, wherein the reference test signal XREF comprises the test signal x uncorrupted by the communication channel, further wherein N equals the number of sampled values of the finite sequence of xn, and still further wherein y is a finite sequence of sampled values of the received signal. - space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
-
-
4. The method of communication channel identification according to claim 1, wherein said step of calculating the sequence of channel values further comprises generating a number of singular-value-decomposition (SVD) singular values and truncating an optimal number of SVD singular values to zero.
-
5. The method of communication channel identification according to claim 4, wherein truncating the optimal number of SVD singular values to zero comprises the steps of (i) truncating a first number of smallest singular values and then (ii) continuing to truncate an increasing number of smallest singular values until a minimum squared-estimation error (SE) of a performance of the LS estimator is obtained.
-
6. The method of communication channel identification according to claim 4, wherein truncating the optimal number of SVD singular values to zero comprises truncating singular values of less than 1×
- 10-4.
- 7. The method of communication channel identification according to claim 4, wherein calculating the sequence of channel values still further comprises using a least-squares (LS) estimator hA given by
- space="preserve" listing-type="equation">h.sub.A =T.sup.†
y,
where T.sup.†
is a Moore-Penmore (approximate) inverse of T and T is a Toeplitz matrix given by ##EQU9## further wherein xn, for n=0 to (N+L-1), is a finite sequence of sampled values of a reference test signal XREF, wherein the reference test signal XREF comprises the test signal x uncorrupted by the communication channel, further wherein N equals the number of sampled values of the finite sequence of xn, and still further wherein y is a finite sequence of sampled values of the received signal. - space="preserve" listing-type="equation">h.sub.A =T.sup.†
- 104.
- space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
T.sup.T y,
TT is approximately equal to T.sup.†
, wherein T.sup.†
is a Moore-Penmore (approximate) inverse of T and T is a Toeplitz matrix given by ##EQU10## further wherein xn, for n=0 to (N+L-1), is a finite sequence of sampled values of a reference test signal XREF, wherein the reference test signal XREF comprises the test signal x uncorrupted by the communication channel, further wherein N equals the number of sampled values of the finite sequence of xn, and still further wherein y is a finite sequence of sampled values of the received signal.
- 10-4.
-
13. An apparatus for communication channel identification comprising:
-
means for receiving a signal y, wherein signal y comprises a test signal x after test signal x has passed through a communication channel to be identified, test signal x having been transmitted over the communication channel and further having a signal span; means for storing a reference test signal XREF, wherein the reference test signal XREF comprises the test signal x uncorrupted by the communication channel; and means responsive to the received signal y and the reference test signal xREF for calculating a sequence of channel values corresponding to an estimated channel impulse response, the channel impulse response providing an estimate of the communication channel, wherein the sequence of channel values comprises a sampled discrete channel of a finite number of non-zero indices in an interval defmed by -L, M!, wherein the channel impulse response has an impulse response span equal to M+L, the impulse response span being less than the span of the test signal x, wherein said calculating means further comprises a least-squares (LS) estimator. - View Dependent Claims (14, 15, 16, 17, 18, 22, 23, 24)
- 15. The apparatus for communication channel identification according to claim 13, wherein the LS estimator of said calculating means comprises a LS estimator hB given by
- space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
T.sup.T y,
where (TT T).sup.†
T is approximately equal to T.sup.†
, wherein T.sup.†
is a Moore-Penmore (approximate) inverse of T and T is a Toeplitz matrix given by ##EQU12## further wherein xn, for n=0 to (N+L-1), is a finite sequence of sampled values of a reference test signal xREF, further wherein N equals the number of sampled values of the finite sequence of xn, and still further wherein y is a finite sequence of sampled values of the received signal. - space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
-
-
16. The apparatus for communication channel identification according to claim 13, wherein said calculating means further comprises means for generating a number of singular-value-decomposition (SVD) singular values and means for truncating an optimal number of SVD singular values to zero.
-
17. The apparatus for communication channel identification according to claim 16, wherein the truncating means (i) truncates a first number of smallest singular values and then (ii) continues to truncate an increasing number of
smallest singular values until a minimum squared-estimation error (SE) of a performance of the LS estimator is obtained. -
18. The apparatus for communication channel identification according to claim 16, wherein the truncating means truncates singular values of less than 1×
- 10-4.
- 22. The apparatus for communication channel identification according to claim 16, wherein the LS estimator of said calculating means comprises a LS estimator hB given by
- space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
T.sup.T y,
where (TT T).sup.†
TT is approximately equal to T.sup.†
, wherein T.sup.†
is a Moore-Penmore (approximate) inverse of T and T is a Toeplitz matrix given by ##EQU14## further wherein xn, for n=0 to (N+L-1), is a finite sequence of sampled values of a reference test signal XREF, further wherein N equals the number of sampled values of the finite sequence of xn, and still further wherein y is a finite sequence of sampled values of the received signal. - space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
- 10-4.
- 19. The apparatus for communication channel identification according to clain 16, wherein the LS estimator of said calculating means comprises a LS estimator hA given by
- space="preserve" listing-type="equation">h.sub.A =T.sup.†
y,
where T.sup.†
is a Moore-Pemnore (approximate) inverse of T and T is a Toeplitz matrix given by ##EQU13## further wherein xn, for n=0 to (N+L-1), is a finite sequence of sampled values of a reference test signal xREF, further wherein N equals the number of sampled values of the finite sequence of xn, and still further wherein y is a finite sequence of sampled values of the received signal.- View Dependent Claims (20, 21)
- space="preserve" listing-type="equation">h.sub.A =T.sup.†
-
25. A method of removing channel-induced distortion from signals, said method comprising the steps of:
-
receiving the signals, wherein the signals include a signal y, further wherein signal y comprises a test signal x after test signal x has passed through and been distorted by a communication channel, test signal x having been transmitted over the communication channel and further having a signal span; identifying the communication channel in real time, wherein said identifying step comprises calculating a sequence of channel values corresponding to an estimated channel impulse response, the channel impulse response providing an estimate of the communication channel, wherein the sequence of channel values comprises a sampled discrete channel of a finite number of non-zero indices in an interval defmed by -L, M!, wherein the channel impulse response has an impulse response span equal to M+L, the impulse response span being less than the span of the test signal x, wherein calculating the sequence of channel values further comprises using a leastsquares (LS) estimator; and filtering the received signals in real time, in response to the sequence of channel values corresponding to the estimated channel impulse response, to remove channel-induced distortion from the signals. - View Dependent Claims (26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36)
- 27. The method of removing channel-induced distortion from signals according to claim 25, wherein calculating the sequence of channel values further comprises using a least-squares (LS) estimator hB given by
- space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
T.sup.T y,
where (TT T).sup.†
TT is approximately equal to T.sup.†
, wherein T.sup.†
is a Moore-Penmore (approximate) inverse of T and T is a Toeplitz matrix given by ##EQU16## further wherein xn, for n=0 to (N+L-1), is a finite sequence of sampled values of a reference test signal xREF, wherein the reference test signal xREF comprises the test signal x uncorrupted by the communication channel, further wherein N equals the number of sampled values of the finite sequence of xn, and still further wherein y is a finite sequence of sampled values of the received signal. - space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
-
-
28. The method of removing channel-induced distortion from signals according to claim 25, wherein calculating the sequence of channel values further comprises generating a number of singular-value-decomposition (SVD) singular values and truncating an optimal number of SVD singular values to zero.
-
29. The method of removing channel-induced distortion from signals according to claim 28, wherein truncating the optimal number of SVD singular values to zero comprises the steps of (i) truncating a first number of smallest singular values and then (ii) continuing to truncate an increasing number of smallest singular values until a minimum squared-estimation error (SE) of a performance of the LS estimator is obtained.
-
30. The method of removing channel-induced distortion from signals according to claim 28, wherein truncating the optimal number of SVD singular values to zero comprises truncating singular values of less than 1×
- 10-4.
- 31. The method of removing channel-induced distortion from signals according to claim 28, wherein calculating the sequence of channel values still further comprises using a least-squares (LS) estimator hA given by
- space="preserve" listing-type="equation">h.sub.A =T.sup.†
y,
where T.sup.†
is a Moore-Penmore (approximate) inverse of T and T is a Toeplitz matrix given by ##EQU17## further wherein xn, for n=0 to (N+L-1), is a finite sequence of sampled values of a reference test signal XREF, wherein the reference test signal XREF comprises the test signal x uncorrupted by the communication channel, further wherein N equals the number of sampled values of the finite sequence of xn, and still further wherein y is a finite sequence of sampled values of the received signal. - space="preserve" listing-type="equation">h.sub.A =T.sup.†
- 104.
- space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
T.sup.T y,
TT is approximately equal to T.sup.†
, wherein T.sup.†
is a Moore-Penmore (approximate) inverse of T and T is a Toeptitz matrix given by ##EQU18## further wherein xn, for n=0 to (N+L-1), is a finite sequence of sampled values of a reference test signal XREF, wherein the reference test signal XREF comprises the test signal x uncorrupted by the communication channel, further wherein N equals the number of sampled values of the finite sequence of xn, and still further wherein y is a finite sequence of sampled values of the received signal.
- 10-4.
-
37. An apparatus for removing channel-induced distortion from signals, said means for receiving the signals, wherein the signals include a signal y, further wherein signal y comprises a test signal x after test signal x has passed through and been distorted by a communication channel, test signal x having been transmitted over the communication channel and further having a signal span;
-
means for storing a reference test signal xREF, wherein the reference test signal XREF comprises a copy of the test signal x uncorrupted by the comlmunication channel; means responsive to the received signal y and the reference test signal XREF for identifying the communication channel in real time, wherein said identifying means comprises means for calculating a sequence of channel values corresponding to an estimated channel impulse response, the channel impulse response providing an estimate of the communication channel, wherein the sequence of channel values comprises a sampled discrete channel of a finite number of non-zero indices in an interval defined by -L, M!, wherein the channel impulse response has an impulse response span equal to M+L, the impulse response span being less than the span of the test signal x, wherein said calculating means further comprises a least-squares (LS) estimator; and means for filtering the received signals in real time, in response to the sequence of channel values corresponding to the estimated channel impulse response, to remove channel-induced distortion from the signals. - View Dependent Claims (38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48)
- 39. The apparatus for removing channel-induced distortion from signals according to claim 37, wherein the LS estimator of said calculating means comprises a LS estimator hB given by
- space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
T.sup.T y,
where (TT T).sup.†
TT is approximately equal to T.sup.†
, wherein T.sup.†
is a Moore-Penmore (approximate) inverse of T and T is a Toeplitz matrix given by ##EQU20## further wherein xn, for n=0 to (N+L-1), is a finite sequence of sampled values of a reference test signal xREF, further wherein N equals the number of sampled values of the finite sequence of xn, and still further wherein y is a finite sequence of sampled values of the received signal. - space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
-
-
40. The apparatus for removing channel-induced distortion from signals according to claim 37, wherein said calculating means further comprises means for generating a number of singular-value-decomposition (SVD) singular values and means for truncating an optimal number of SVD singular values to zero.
-
41. The apparatus for removing channel-induced distortion from signals according to claim 40, wherein the truncating means (i) truncates a first number of smallest singular values and then (ii) continues to truncate an increasing number of smallest singular values until a minimum squared-estimation error (SE) of a performance of the LS estimator is obtained.
-
42. The apparatus for removing channel-induced distortion from signals according to claim 40, wherein the truncating means truncates singular values of less than 1×
- 10-4.
- 43. The apparatus for removing channel-induced distortion from signals according to claim 40, wherein the LS estimator of said calculating means comprises a LS estimator hA given by
- space="preserve" listing-type="equation">h.sub.A =T.sup.†
y,
where T.sup.†
is a Moore-Penmore (approximate) inverse of T and T is a Toeplitz matrix given by ##EQU21## further wherein xn, for n=0 to (N+L-1), is a finite sequence of sampled values of a reference test signal xRu, further wherein N equals the number of sampled values of the finite sequence of xn, and still further wherein y is a finite sequence of sampled values of the received signal. - space="preserve" listing-type="equation">h.sub.A =T.sup.†
- 10-4.
- space="preserve" listing-type="equation">h.sub.B =(T.sup.T T).sup.†
T.sup.T y,
TT is approximately equal to T.sup.†
, wherein T.sup.†
is a Moore-Penmore (approximate) inverse of T and T is a Toeplitz matrix given by ##EQU22## further wherein xn, for n=0 to (N+L-1), is a finite sequence of sampled values of a reference test signal xREF, further wherein N equals the number of sampled values of the finite sequence of xn, and still further wherein y is a finite sequence of sampled values of the received signal.
- 10-4.
Specification