Multiplex communications – Duplex – Transmit/receive interaction control
Reexamination Certificate
1998-11-09
2002-02-26
Yan, Kwang B. (Department: 2664)
Multiplex communications
Duplex
Transmit/receive interaction control
C379S406080
Reexamination Certificate
active
06351457
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to methods of and apparatuses for identifying unknown systems such as transmission lines and spatial acoustical coupling.
As applications of identification of unknown systems with adaptive filters, echo cancellers, noise cancellers, hauling cancellers and adaptive equalizers are well known in the art. Pertaining prior art technique will now be described in connection with an echo canceller as an example, which cancels echo leaking from the transmitting side to the receiving side on the 4-line side of a 2-to4-wire hybrid transformer.
An echo canceller uses an adaptive filter having tap coefficients equal to more than those of the impulse response of the echo path. In operation, the echo canceller generates a psuedo echo (or echo replica) corresponding to the transmitted signal, and thus suppresses echo leaking from the transmitting circuit to the receiving circuit on the 4-line side of the 2-to4-wire hybrid transformer. Each tap coefficient of the adaptive filter is corrected by taking correlation between an error signal, obtained by subtracting the echo replica from the mixed signal including the echo and the received signal, and the transmitted signal. Typical examples of the tap coefficient adaptation algorithm of such an adaptive filter, are the LMS algorithm as disclosed in Proceedings of IEEE, Vol. 63, No. 12, December 1975, pp. 1962-1716 (Literature
1
) and a learning identification method (LIM) disclosed in IEEE Transactions on Automatic Control, Vol. 12, No. 3, March 1967, pp. 282-287 (literature 2).
When a fixed delay is present between the point on the 4-line side, where the echo canceller is inserted, and the point where the hybrid transformer is located, the echo canceller requires a sufficient number of taps for covering both the maximum estimated fixed delay and the actual significant part of the impulse response. Thus, when the fixed delay is long, an enormous number of taps are necessary, thereby increasing the hardware size and the convergence time due to mutual interference of the coefficients. To solve this problem, an adaptive control method of coefficient arrangement is disclosed in IEEE Transactions on Circuits and Systems-11, “Analog and Digital Signal Processing”, Vol.43, No.9, September 1996, pp. 629-636 (Literature 3). In this method, the position of the significant part of the impulse response is estimated, around which tap coefficient of the adaptive filter are adaptively located.
The gist of the method shown in Literature 3 is to roughly estimate the position of part position and arrange the tap coefficients in a limited neighborhood of the estimated position, thus reducing the convergence time. The convergence time until the taps are arranged at right positions is also reduced by dividing a queue, in which the indexes of taps without any coefficient are stored, into two queue, one corresponding to the waveform response part neighborhood and the other corresponding to the significant part and the other part than the significant part. The position of the significant is estimated from the absolute maximum value of the tap coefficients, and only a single tap coefficient arrangement range is estimated. Therefore, where a plurality of significant parts are present (these parts being collectively called multi-echo when echoes are concerned), the tap coefficient arrangement range should be expanded so as to cover all these parts. Therefore, where a long fixed delay is involved between two parts, the limiting effect on the coefficient arrangement is reduced, inevitably increasing the convergence time. To solve this problem, methods which permit fast convergence even with multi-echoes and concentration tap coefficients in the significant parts, are disclosed in Japanese Patent Disclosure No. 7-202766 (Literature 4) and IEEE Proceedings of International Conference on Acoustics, Speech and Signal Processing, Vol. III, pp. 41-44, April 1994 (Literature 5).
FIG. 24
is a block diagram showing the construction of the echo canceller proposed in Literature
4
. In the adaptive filter shown in
FIG. 24
, (N−1) delay elements
20
1
to
20
N−1
are provided for delaying a transmitted signal supplied from a transmitted signal input terminal
1
. The taps provided are N in number including the one zero delay. L (N>L) coefficient generators
30
1
to
30
L
are provided for generating tap coefficients of the adaptive filter. Specifically, the adaptive FIR filter shown in
FIG. 24
, unlike the usual adaptive FIR filters, has a sufficient number of tap coefficients for realizing the substantial waveform response part excluding the flat delay part or parts, and adaptively arranges the tap coefficients therein. A routing switch
7
is provided for switching the connection between the delay elements and the coefficient generators. A tap control circuit
9
is provided for controlling the routing switch
7
. Specifically, the routing switch
7
selectively transmits the outputs of L delay elements to the coefficient generators according to the data received from an output terminal
900
of the tap control circuit
9
. The routing switch
7
supplies the outputs of L delay elements with the tap coefficient generators
30
1
to
30
L
and also to multipliers
40
1
to
40
L
, respectively. The multipliers
40
1
to
40
L
multiply tap coefficient values provided from the coefficient generators
30
1
to
30
L
and the delayed signals from the routine switch
7
, and supplies the products to an adder
8
. The adder
8
adds together the products from the multipliers
40
1
to
40
L
and provides the sum as an echo replica. The tap control circuit
9
supplies from its output terminals
901
1
to
901
L
step size, which are used for coefficient adaptation. The tap control circuit
9
also supplies coefficient-clear signals from its output terminals
902
1
to
902
L
to the coefficient generators
30
1
to
31
L
for clearing, i.e., zero resetting, the coefficients therein. Furthermore, the tap control circuit
9
receives from its input terminals
903
1
to
903
L
tap position control signals, and uses these signals for generating the step size and the coefficient-clear signals.
The transmitted signal supplied from the transmitted signal input terminal
1
is transmitted from a transmitted signal output terminal
2
via a transmission line to a hybrid transformer
3
, is coupled to the 2-line side thereof. Due to impedance mismatch, the transmitted signal partly leaks as an echo to the receiving side. The echo from a received signal input terminal
4
is supplied to a subtracter
5
. The subtracter
5
subtracts the echo replica from the adder
8
, and transmits the difference result to a received signal output terminal
6
. The difference is also supplied as an error signal for coefficient adaptation to the tap coefficient generators
30
1
to
30
L
.
Assuming as the coefficient adaptation algorithm the LMS algorithm shown in “Adaptive Signal Processing”, 1985, Prentice-Hall Inc., USA (Literature 6), the value c
i
(k+1) of i-th tap coefficient after the (k+1)-th adaptation is given, using value c
i
(k) after k-th adaptation, as:
c
i
(k+1)=c
i
(k)+&mgr;
i
e(k)×(k−a(i)) (1)
where &mgr;
i
is the step size corresponding to the i-th coefficient, e(k) is the residual echo, x(k−a(i)) is the input signal sample at (k−a(i) )-th coefficient adaptation, a(i) is a set of indexes to the delay elements selected by the routing switch
7
, and L is the number of delay elements.
The coefficient generators
30
i
(i being 1 to L) may have a construction as shown in FIG.
25
. As shown, a multiplier
31
multiplies the error signal and the step size by each other. A multiplier
32
multiplies the product output of the multiplier
31
and delayed signal supplied from the routing switch
7
. An adder
33
adds the output of the multiplier
32
, which represents a coefficient correction amount, and a coefficient value stored in a memory
34
. The sum result of
LandOfFree
Method of and apparatus for identifying unknown system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of and apparatus for identifying unknown system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of and apparatus for identifying unknown system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2960394