Suppression of multipath effects in the transmission of analog TV signals
View Patent ↗A method for compensating for multipath effects when receiving a vestigial sideband modulated TV signal, one portion of the spectrum of the undistorted TV signal being symmetrical with respect to a video carrier frequency, and deviations from the symmetry of the relevant portion of the spectrum that exists in the undistorted state being analyzed in order to suppress the multipath effects.
1. A method for compensating for multipath effects of a vestigial sideband modulated TV signal, a portion of a spectrum of the undistorted TV signal being symmetrical with respect to a video carrier frequency, the method comprising:
receiving the vestigial sideband modulated TV signal; and
analyzing deviations from symmetry of a relevant portion of the spectrum that exists in the undistorted state so as to suppress the multipath effects,
wherein the TV signal includes an imaginary component and a real component and further comprising minimizing the imaginary component in proportion to the real component in the baseband so as to suppress the multipath effects,
wherein the minimizing is performed using a digital filter, and further comprising determining filter coefficients using a stochastic gradient method including the following steps:
defining an error function e(k) from the filtered signal y(k)
calculating the gradient of the error function e(k)
superposing the gradient of the error function
defining a step size for adapting the filter coefficients
defining the filter coefficients of the digital filter as components of a distortion correction vector and
calculating the new filtered TV signal using the distortion correction vector.
2. The method as recited in claim 1 , wherein
the error function e(k) has the form
e
(
k
)
=
α
·
∑
k
N
H
arctan
(
imag
(
y
(
k
)
)
/
real
(
y
(
k
)
)
)
and the distortion correction vector is calculated using the expression
w
(
k
+
1
)
=
w
(
k
)
-
μ
e
(
k
)
Δ
e
(
k
)
Δ
w
(
k
)
.
3. The method as recited in claim 1 , wherein
the error function e(k) has the form
e
(
k
)
=
α
·
∑
k
N
H
b
(
k
)
·
arctan
(
imag
(
y
(
k
)
)
/
real
(
y
(
k
)
)
)
where
b
(
k
)
=
y
(
k
)
∑
k
N
H
y
(
k
)
and the distortion correction vector is calculated using the expression
w
(
k
+
1
)
=
w
(
k
)
-
μ
e
(
k
)
Δ
e
(
k
)
Δ
w
(
k
)
.