Telephonic communications – Echo cancellation or suppression
Reexamination Certificate
2000-03-23
2004-02-24
Barnie, Rexford (Department: 2643)
Telephonic communications
Echo cancellation or suppression
C379S406020, C379S406080, C379S406070, C379S406090, C370S290000
Reexamination Certificate
active
06697486
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to an echo canceler and to an operating method therefor.
2. Description of Related Art
In long-distance telephone lines, part of the reception signal output from the sending party's end passes through a two-wire-to-four-wire conversion hybrid (or echo path) on the receiving party's end and circles back to the sending party, resulting in an echo which interferes significantly with the call. One apparatus for preventing this is a conventional echo canceler. One example of such an echo canceler is disclosed in Japanese Patent Application Laid-Open No. H9-93088/1997.
FIG. 27
is a configuration diagram for an echo canceler disclosed in the cited literature.
This echo canceler
100
comprises an adaptive filter modulus computing system
102
, and adaptive filter
104
, and an adder
106
. The adaptive filter modulus computing system
102
produce an adaptive filter modulus H
m
(i) for eliminating an echo E(i) that comes back with a delay to the sender's side SS. The adaptive filter
104
, using that adaptive filter modulus H
m
(i), produces a pseudo-echo signal GE(i) for that echo E(i). The adder
106
subtracts the pseudo-echo signal GE(i) from the echo E(i) to eliminate the echo E(i). Accordingly, the power of the echo component in the voice of the receiving party that comes back after a delay to the receiving side RS is reduced.
The symbol i (i=1, 2, 3, . . . ) represents a sample number for each signal. For example, the reception signal X(i) for the sample number 2 becomes X(2). The symbol m (m=1, 2, 3, . . . ) is a number (or convolution number) associated with each of a plurality of the delayers (or tap devices) that configure the adaptive filter modulus computer (described below) contained in the adaptive filter modulus computing system
102
. For example, the adaptive filter modulus H
m
(i) for the sample number 3 in the delayer having the number 5 becomes H
5
(3).
A concrete description is now given for the operation of the adaptive filter modulus computing system
102
, the adaptive filter
104
, and the adder
106
.
This adaptive filter modulus computing system
102
only operates when the call status is in a simplex status, that is, when a reception signal X(i) directed toward the receiving party's side RS by the sending party exists, but no transmission signal exists that is directed toward the sending party's side SS by the receiving party. This adaptive filter modulus computing system
102
estimates the delay characteristics (that is, impulse response in the echo path) of the two-wire-to-four-wire conversion hybrid HB at the receiving party's side RS by a commonly known least mean square method (hereinafter LMS method). That estimated impulse response becomes the adaptive filter modulus H
m
(i) described above. With the LMS method, the power of the echo cannot be reduced to or below the noise level. A method for resolving this problem is disclosed in the prior art literature cited earlier. Specifically, this adaptive filter modulus computing system
102
computes the adaptive filter modulus H
m
(i) according to formula 16 or formula 17 below, each of which is a computation formula in the LMS method.
A
m
(
i
)
=
∑
j
=
iBL
(
i
)
+
1
(
i
+
1
)
BL
(
i
)
ER
(
j
)
X
(
j
-
m
)
XP
m
=
∑
i
=
1
T
(
∑
j
=
iBL
(
i
)
+
1
(
i
+
1
)
BL
(
i
)
X
2
(
j
-
m
)
)
R
m
(
i
)
=
A
m
(
i
)
XP
m
H
m
(
i
+
1
)
=
H
m
(
i
)
+
KR
m
(
i
)
}
(
16
)
A
m
(
i
)
=
∑
j
=
iBL
(
i
)
+
1
(
i
+
1
)
BL
(
i
)
ER
(
j
)
X
(
j
-
m
)
XP
m
(
i
)
=
∑
j
=
iBL
(
i
)
+
1
(
i
+
1
)
BL
(
i
)
X
2
(
j
-
m
)
R
m
(
i
)
=
A
m
(
i
)
XP
m
(
i
)
H
m
(
i
+
1
)
=
H
m
(
i
)
+
KR
m
(
i
)
}
(
17
)
In these formulas 16 and 17, ER(i) represents the echo remainder, that is, the difference between the echo E(i) and the pseudo-echo signal GE(i). This echo remainder ER(i) is the echo component that cannot be eliminated, and hence remains, even after subtracting the pseudo-echo signal GE(i) from the echo E(i). R
m
(i) is the modulus update amount. K (K>0) is the step gain. BL(i) is the block length (or number of computation terms). T is the number of delayers (or tap devices) contained in the adaptive filter modulus computing system
102
that are used. This number used is called the tap length T, and is preset. The maximum value of this tap length T is the total number of delayers.
The parameters for the adaptive filter modulus H
m
(i) described in the foregoing are the echo remainder ER(i), the reception signal X(i), and the block length BL(i). Of these parameters, the echo remainder ER(i) and reception signal X(i) are both observed values, while the block length BL(i) is an artificial parameter.
The method of determining the block length BL(i) is now described.
The adaptive filter modulus computing system
102
described in the foregoing compares the size relationship between the modulus update amount in the formulas 16 and 17 and the R
m
(i) after one sample from that modulus update amount R
m
(i), and thereby determines the block length BL(i) which it outputs to the adaptive filter
104
.
Now, according to the LMS method, by producing a pseudo-echo signal GE(i) using the adaptive filter modulus Hm(i) produced according to formulas 16 and 17, and subtracting that pseudo-echo signal GE(i) from the echo E(i), the power of the echo remainder ER(i) is reduced to the level of the power of the noise N(i). This noise N(i), however, is assumed not to contain an echo component.
According to formula 16 or formula 17, when the modulus update amount is Rm(i+1)≦Rm(i), H
m
(i) converges. When this happens, this adaptive filter modulus computing system
102
estimates that the power of the echo remainder ER(i) can be reduced to the level of the noise N(i) power, and, based on that decision, holds the block length BL(i).
When the modulus update amount is Rm(i+1)>Rm(i), the adaptive filter modulus H
m
(i) diverges. When that happens, the adaptive filter modulus computing system
102
described earlier estimates that the power of the echo remainder ER(i) cannot be reduced to the level of the noise N(i) power, and, based on that decision, lengthens the block length BL(i).
This adaptive filter modulus computing system
102
, however, does not lengthen the block length BL(i) using the block length BL(i) at that current point in time as a reference value, but sequentially lengthens it from the predetermined minimum value of the block length. For example, if the current block length BL(i) is 30 and the modulus update value R
m
(i+1)>R
m
(i), this block length BL(i) will first return from 30 to the minimum value (10, for example) of the block length BL(i). Next, this adaptive filter modulus computing system
102
starts increasing that block length BL(i) at a constant ratio (adding 1 at a time, for example), from that minimum value. This computation of the modulus update amount R
m
(i) by lengthening the block length BL(i) is continued until R
m
(i+1)≦R
m
(i).
Next, the adaptive filter
104
described earlier produces a pseudo-echo signal GE(i) for the echo E(i) that circles back to the sending party's side SS, using formula 3 below. That is, the adaptive filter
104
produces the pseudo-echo signal GE(i) by convoluting the reception signal X(i) with the adaptive filter modulus H
m
(i) output from the adaptive filter modulus computing system
102
, and outputs that to the adder
106
.
GE
(
i
)
Barnie Rexford
Burdett James R.
Oki Electric Industry Co. Ltd.
Venable
LandOfFree
Echo canceler and operating method does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Echo canceler and operating method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Echo canceler and operating method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3347075