CA1310708C - Adaptive, digital filter including a non-recursive part and a recursive part - Google Patents
Adaptive, digital filter including a non-recursive part and a recursive partInfo
- Publication number
- CA1310708C CA1310708C CA000597705A CA597705A CA1310708C CA 1310708 C CA1310708 C CA 1310708C CA 000597705 A CA000597705 A CA 000597705A CA 597705 A CA597705 A CA 597705A CA 1310708 C CA1310708 C CA 1310708C
- Authority
- CA
- Canada
- Prior art keywords
- filter
- recursive
- adaptive
- filters
- signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Lifetime
Links
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H21/00—Adaptive networks
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H21/00—Adaptive networks
- H03H21/0012—Digital adaptive filters
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B3/00—Line transmission systems
- H04B3/02—Details
- H04B3/20—Reducing echo effects or singing; Opening or closing transmitting path; Conditioning for transmission in one direction or the other
- H04B3/23—Reducing echo effects or singing; Opening or closing transmitting path; Conditioning for transmission in one direction or the other using a replica of transmitted signal in the time domain, e.g. echo cancellers
- H04B3/237—Reducing echo effects or singing; Opening or closing transmitting path; Conditioning for transmission in one direction or the other using a replica of transmitted signal in the time domain, e.g. echo cancellers using two adaptive filters, e.g. for near end and for end echo cancelling
Abstract
ABSTRACT OF THE DISCLOSURE
An adaptive digital filter includes a non-recursive part and a recursive part and can be updated in a simple and reliable manner. The recursive part of the filter has a plurality of separate, permanently set recursive filters with different impulse responses, and a linear combination is formed with adaptive weighting factors from the output signals of the recursive filters. The filter is updated by a single signal utilized for updating the non-recursive part of the filter and the adaptive weighting factors in the recursive part of the filter.
An adaptive digital filter includes a non-recursive part and a recursive part and can be updated in a simple and reliable manner. The recursive part of the filter has a plurality of separate, permanently set recursive filters with different impulse responses, and a linear combination is formed with adaptive weighting factors from the output signals of the recursive filters. The filter is updated by a single signal utilized for updating the non-recursive part of the filter and the adaptive weighting factors in the recursive part of the filter.
Description
)I'J(3 The invention relates to an adaptive, digital filter including a non-recursive part and a recursive part. uses of the filter are for example as an echo canceller or equalizer in t~lecommunications equipment.
The impulse response from a ~ilter which is used for echo cancellation in talecommunication equipment must ss closely as possible imitate the impulse response of the transmission line in question. Included in the transmission line in such a case are two-wire to four-wire junctions, analogue-digital converters etc, which affect the impulse response. The latter generally has relatively long duration. It is therefore difficult to achieve a suitable impulse response with a filter which only has a finite impulse response. Such filters are called non-recursive filters or FIR filters (finite impulse response). To achieve a suitable impulse response, a filter for echo cancellation should comprise both a non-recursive part and a recursive part. Recursive filters are also called IIR filters (infinite impulse response).
There are known reliable methods for updating adaptive FIR
filters, i.e. by adjusting the coeEficients of such filters.
They can be updated by minimizing the square of an error signal, which constitutes the difference between a so- called desired signal and the output signal of the filter. In such a case the desired signal may be a signal occurring on the receiver side in communication equipment where the filter is included. The square of the error signal can be minimized, e.g. according to the so-called LMS method (least mean square). The LMS method is described inter alia in the book:
Widrow and Stearns, "Adaptive signal processing", pp 99-101.
Minimizing the square of an error signal according to the above is a so-called least square problem, due to the square of the error signal being a quadratic function of the filter coefficient values. This means that this square can be represented by a quadratic error surface, in an N-dimensional space where N is the number of coefficients, the optimum - 1 3 1 OI(~Ir3 filter setting corresponding to the minimum point on this surface.
The corresponding s~uare for an IIR filter is not represented by a quadratic error surface accordiny to the above, however, and the error surface can have local minimum points instead.
Known updating methods can fasten in such a local minimum pcint, resulting in the optimum setting never being obtained.
Recursive filters can also be unstable, as a result of the fact that the poles in the transform of the transfer function can at least temporarily be moved outside the unit circle.
For an IIR filter of the first degree, this means that the filter coefficient can be an amount greater than one, which makes the filter unstable.
It is known to use a so-called "equation errorl' structure to avoid the problem with local minimii. In such a case two FIR
filters are used, inter alia, of which one is connected to a transmitter side and the other to a receiver side in the same telecommunication equipment. An error signal is formed by the output signal of one filter being subtracted from the output signal of the othsr. The square of this error signal has a quadratic error surface, but a structure of this kind has the disadvantage that the minimized error signal does not represent the actual error. This is so, inter alia, when disturbances occur and when speech signals occur on the transmitter and receiver sides simultaneously. It has also been found difficult to adjust two filters which are connected in this way, due to the filters affecting each other. The equation error method is described, e.g. in the above-mentioned book "Adaptive signal processing", pp 250-253.
An object of the present invention is to provide an adaptive digital filter which includes a non-recursive part and a recursive part, and which can be updated in a simple and reliable way. This is achieved by the recursive part of the 1 31 n70~
filter having a plurality of separate, permanently set recursive filters with different impulse responses, and in that linear combination with adaptive weighting ~actors is formed by the output signals of the recursive filter. The filter is updated by a single signal being utilized for updating ths non-recursive part and the-adaptive weighting factors in the recursive part. A stable filter is also obtained in this way, due to the poles of the recursive filter not being displaced.
Accordingly therefore the present invention provides an adaptive digital ~ilter comprising a non-recursive part and a recursive part; said recursive part including a plurality of branches each having respective separate, permanently set recursive filters with mutual different impulse responses, respective multiplication means with an adaptive multiplication factor associated with each recursive filter, and summing means which in conjunction with said multiplication means forms a linear combination of the output signals of the recursive filters; and means for generating a single signal ~rom the output of said recursive and non-recursive parts to update the non-recursive part and the adaptive multiplication factors of said multiplication means in the recursive part.
The invention will now be described in more detail, by way of example only, with reference to the accompanying drawings in which:-Figure 1 illustrates a known apparatus for echo cancellation;Figure 2 illustrates an example of desired impulse response from a filter in accordance with the invention;
" 1 31 070~
Figure 3 illustrates a first embodiment of a filter in accordance with the invention;
Figure 4 illustrates a more detailed embodiment of the filter according to Figure 3;
Figure 5 is a series of graphs giving examples of different impulse responses in certain individual filters included in the filter in accordance with the invention; and - 3a -1 3 1 070~, Figure 6 illustrates a second embodiment of a filter in accordance with the inv~ntion.
A known apparatus for echo cancellation is illustrated in Figure 1. A digital input signal x(n) occurring on the transmission side of telecommunications equipment is applie~
to a two-wire to four-wire junction 2, i.e. a hybrid, which is connected to a receiver side in the telecommunications equipment and across a two-wire line to a telephone set ~.
Echo signals occur in the hybrid and in the two-wire line.
The output signal to the receiver side from -the hybrid 2 is denoted d(n) and consists solely of echo signals when no signal is received from the telephone set 4. This signal agrees with the above-mentioned desired signal.
The input signal x(n) is also applied to an adaptive ~IR
filter 1, which generates an expected echo signal y(n). An error signal e(n) is formed in summing means 3, this signal being the difference between the signals d(n) and y(n), and is utilized for updating the filter. As will be seen from the above, an FIR filter can be updated according to known methods, e.g. the LMS method. The impulse response of the filter is however generally too short for effective echo cancellation to be obtained.
In Figure 2 there is illustrated an example of a desired impulse response h(n) with relatively duration, where n denotes the sequential number for the respective sample value. The impulse response can be divided into two main parts. There is first a considerable transient containing the greater part of the signal energy of the impulse. There is then a long, and substantially exponentially decaying part, a so-called tail. Negative values can also occur in the impulse response.
A first embodiment example of a filter in accordance with the invention is illustrated in Figure 3. The filter receives a digital signal x(n) as inpuk signal, this signal 1 ~1 ()7~,~
corresponding, for example, to ~he siynal x(n) in the apparatus accordiny to Figure 1. The input signal is applied to an FIR filter 11 directly, and to a plurality of IIR
filters 13-16 after delay by a time ~vin a delay means 12.
The IIR filters are suitably of the first degree, and have permanently set filter coefficients having mutually differing values. The output signal from the FI~ filter 11 is supplied to a summing means 22, and the output signals from 20 the IIR
filters 13-16 are each applied to their respective multiplication means 18-21. Each of the latter has an adaptive multiplication factor. These multiplication factors are assumed to have the values WO-W3, and they are adjusted in the way given below. The output signals from the FIR
filter 11 and from the multiplication means 18-21 are finally added to the summing means.
In accordance with the inventive concept, the first part of the impulse response is generated in the FIR filter 11 and the second part is generated as a linear combination of the output signals from the IIR filters 13-16. The weightings in the linear combination are here determined by the adaptive multiplication factors, or weighting factors W0-W3. By suitable delay of the input signal xtn) to the IIR filters, both parts of the impulse response can be generated independently of each other. The filters in accordance with the invention thus comprise two separate filter parts, a non-recursive filter part and a recursive filter part, the output signals of which are added.
The filter output signal is denoted y(n) and is subtracted from an arbitrary desired signal d(n) in a summing means 3.
A difference signal e(n) thus obtained occurs on a line 17 and is utilized both for updating the non-recursive filter part, i.e. the FIR filter 11, and the recursive filter part.
Updating the latter takes place by updating the adaptive weighting factors W0-W3 of the multiplication means 18-21.
The signals y(n), d(n) and e(n) and the summing mean~ 3 agree with corresponding signals and means in Figure 1, for 1 31 070?) example, but the field of application of the filter is of course not limited -to echo cancellation.
For the sake of completeness, it is pointed out that updating means are required both for the FIR filter 11 and the multiplication means 18-21, these updating means being generally known in connection with digital filters.
There is shown in Figure 4 a more detailed embodiment of the filter according to Figure 3. The FIR filter 11 conventionally comprises delay means 38-40, multiplication means 34-37, and summing means 31-33. The IIR filters 13-16 are of the firs~ degree, and each has its permanently set filter coefficient. These filters are also implemented conventionally and each comprises a summing means, e.g. 131, a delay means, e.g. 132, and a multiplication means, e.g.
133.
The multiplication means are each allocated a permanently set coefficient PO-P3, which have mutually different values, and which are thus the filter coefficients of the IIR filters.
Each of the delay means 38-40 included in the FIR filter 11 delay the input signal x(n) by a sample value, and together these correspond to the delay means 12 illustrated in Figure 3. In the illustrated example, ~r =3 T. Such a separate delay means is thus not required in practise but can he included in the FIR filter 25 instead. The summing means 22 in Figure 3 is shown in Figure 4 as a plurality of separate summing means 221-224.
As will be seen from above, the diffe.rence signal e(n) is used for both updating the FIR filter 11 and the adaptive weighting factors W0-W3 of the multiplication means 18-21 in the recursive filter part. The problem of minimizing the difference signal e(n) is equal to minimizing the sum of the square of the expression W0 X pOn ~ Wl x pln ~ W2 x p2n + W3 x p3n - f(n), where n goes from zero to infinity, P0-P3 are 1 ) 1 ()7 ~J r)) the permanent recursive filter coefficients and f(n) is the desired impulse response. This sum has a quadratic error area with only one minimum, since the weighting factors are only present linearly in the expression. This means that the recursive filter part can be updated according to the same method as the n~n-recursive filter part, e.g. according to the LMS method.
Some of the advantages achieved with the filter in accordance with the invention are that the difference signal is represented by a quadratic error area, simultaneously as the difference signal r~presents the actual error (as opposed to an equation error structure). In addition, the recursive filter part is always stable, since the poles of the individual IIR filters are not displaced. This depends in turn on the filter coefficients Po-P3 being permanent.
Some graphs are illustrated in Figure 5, and are examples of different impulse responses of the individual IIR filters in the recursive part of the filter. The transfer functions o~
the IIR filters 13-16 are denoted in turn by hO(n)-h3(n). It is assumed that the input signal to the filters is delayed by a plurality of sample values corresponding to the length of the impulse response of the FIR filter.
The filter coefficients PO-P3 are,`according to the example, 0.5, 0.75, 0.875 and 0.9375. The transfer functions will then be: hO(n) = 005n, hl(n) = 0.75n etc. Other coefficient values can of course be selected.
The part of the entire desired impulse response occurring to the left of the impulse responses illustrated in Figure 5, i.e. earlier than these, is generated in the FIR filter 11.
This is adapted such that its output signal ceases when the impulse responses according to Figure 5 start. It is pointed out, however, that the number of delay means in the FIR
filter included in the filter accordiny to Fiyure ~ is not adapted to the graphs in Figure 5.
1 -~,1 07C~
By linearly combining a plurality o~ given impulse responses in the way described above, it is possible to achi~ve impulse responses of very varying forms. Both positive and negative weighting ~actors WO-W3 can thus occur, of course. The long decaying part of the desired impulse response cannot always be imitated exactly. This does not make so much difference, however, since only a relatively small part of the energy of the entire desired impulse response is in this part. On the other hand, the first, major part of the impulse response which is generated by the FIR filter can be imitated rather precisely.
A second embodiment of a filter in accordance with the invention is illustrated in Figure 6. Further to the means included in the filter according to Figure 3, there is also a network denoted by 50 in this filter. The network 50 includes multiplication means and summing means, which are adapted to form linear combinations of the output signals of the IIR filters 13-16. These means are connected such that the multiplication means 18 obtains the output signal from the filter 13 in an unaltered condition. The multiplication means 19 obtains the sum of the output signal from khe filter 14 and the output signal from the filter 13 multiplied by a factor, and so on, according to the Figure. The linear combinations can be selected such that the input signals to the multiplications means 18, 21 will be orthogonal. These orthogonal impulse responses are then weighted by adaptive weighting factors, as with the filters according to figures 3 and 4. A change in a given weighting factor does not necessarily cause a change in the remaining weighting factors in this case. More rapid convergence is thus obtained. The number of calculations increases somewhat, however.
The filter in accordance with the invention can be used in different connections, when a relatively lony impulse response is desired and not only for adaptive echo cancellation. Of course, the number of IIR filters may be both more or less than just four, as illustrated in the 1 31 070~, examples. The implementation of the FIR and IIR filters can also be different from what has been shown in the examples.
Neither is it necessary to delay the input siynal to the IIR
filters. However, the delay results in the first part of the desired impulse response being generated solely by the FIR
filter, and the second part of the response being generated solely by the recursive filter part.
The impulse response from a ~ilter which is used for echo cancellation in talecommunication equipment must ss closely as possible imitate the impulse response of the transmission line in question. Included in the transmission line in such a case are two-wire to four-wire junctions, analogue-digital converters etc, which affect the impulse response. The latter generally has relatively long duration. It is therefore difficult to achieve a suitable impulse response with a filter which only has a finite impulse response. Such filters are called non-recursive filters or FIR filters (finite impulse response). To achieve a suitable impulse response, a filter for echo cancellation should comprise both a non-recursive part and a recursive part. Recursive filters are also called IIR filters (infinite impulse response).
There are known reliable methods for updating adaptive FIR
filters, i.e. by adjusting the coeEficients of such filters.
They can be updated by minimizing the square of an error signal, which constitutes the difference between a so- called desired signal and the output signal of the filter. In such a case the desired signal may be a signal occurring on the receiver side in communication equipment where the filter is included. The square of the error signal can be minimized, e.g. according to the so-called LMS method (least mean square). The LMS method is described inter alia in the book:
Widrow and Stearns, "Adaptive signal processing", pp 99-101.
Minimizing the square of an error signal according to the above is a so-called least square problem, due to the square of the error signal being a quadratic function of the filter coefficient values. This means that this square can be represented by a quadratic error surface, in an N-dimensional space where N is the number of coefficients, the optimum - 1 3 1 OI(~Ir3 filter setting corresponding to the minimum point on this surface.
The corresponding s~uare for an IIR filter is not represented by a quadratic error surface accordiny to the above, however, and the error surface can have local minimum points instead.
Known updating methods can fasten in such a local minimum pcint, resulting in the optimum setting never being obtained.
Recursive filters can also be unstable, as a result of the fact that the poles in the transform of the transfer function can at least temporarily be moved outside the unit circle.
For an IIR filter of the first degree, this means that the filter coefficient can be an amount greater than one, which makes the filter unstable.
It is known to use a so-called "equation errorl' structure to avoid the problem with local minimii. In such a case two FIR
filters are used, inter alia, of which one is connected to a transmitter side and the other to a receiver side in the same telecommunication equipment. An error signal is formed by the output signal of one filter being subtracted from the output signal of the othsr. The square of this error signal has a quadratic error surface, but a structure of this kind has the disadvantage that the minimized error signal does not represent the actual error. This is so, inter alia, when disturbances occur and when speech signals occur on the transmitter and receiver sides simultaneously. It has also been found difficult to adjust two filters which are connected in this way, due to the filters affecting each other. The equation error method is described, e.g. in the above-mentioned book "Adaptive signal processing", pp 250-253.
An object of the present invention is to provide an adaptive digital filter which includes a non-recursive part and a recursive part, and which can be updated in a simple and reliable way. This is achieved by the recursive part of the 1 31 n70~
filter having a plurality of separate, permanently set recursive filters with different impulse responses, and in that linear combination with adaptive weighting ~actors is formed by the output signals of the recursive filter. The filter is updated by a single signal being utilized for updating ths non-recursive part and the-adaptive weighting factors in the recursive part. A stable filter is also obtained in this way, due to the poles of the recursive filter not being displaced.
Accordingly therefore the present invention provides an adaptive digital ~ilter comprising a non-recursive part and a recursive part; said recursive part including a plurality of branches each having respective separate, permanently set recursive filters with mutual different impulse responses, respective multiplication means with an adaptive multiplication factor associated with each recursive filter, and summing means which in conjunction with said multiplication means forms a linear combination of the output signals of the recursive filters; and means for generating a single signal ~rom the output of said recursive and non-recursive parts to update the non-recursive part and the adaptive multiplication factors of said multiplication means in the recursive part.
The invention will now be described in more detail, by way of example only, with reference to the accompanying drawings in which:-Figure 1 illustrates a known apparatus for echo cancellation;Figure 2 illustrates an example of desired impulse response from a filter in accordance with the invention;
" 1 31 070~
Figure 3 illustrates a first embodiment of a filter in accordance with the invention;
Figure 4 illustrates a more detailed embodiment of the filter according to Figure 3;
Figure 5 is a series of graphs giving examples of different impulse responses in certain individual filters included in the filter in accordance with the invention; and - 3a -1 3 1 070~, Figure 6 illustrates a second embodiment of a filter in accordance with the inv~ntion.
A known apparatus for echo cancellation is illustrated in Figure 1. A digital input signal x(n) occurring on the transmission side of telecommunications equipment is applie~
to a two-wire to four-wire junction 2, i.e. a hybrid, which is connected to a receiver side in the telecommunications equipment and across a two-wire line to a telephone set ~.
Echo signals occur in the hybrid and in the two-wire line.
The output signal to the receiver side from -the hybrid 2 is denoted d(n) and consists solely of echo signals when no signal is received from the telephone set 4. This signal agrees with the above-mentioned desired signal.
The input signal x(n) is also applied to an adaptive ~IR
filter 1, which generates an expected echo signal y(n). An error signal e(n) is formed in summing means 3, this signal being the difference between the signals d(n) and y(n), and is utilized for updating the filter. As will be seen from the above, an FIR filter can be updated according to known methods, e.g. the LMS method. The impulse response of the filter is however generally too short for effective echo cancellation to be obtained.
In Figure 2 there is illustrated an example of a desired impulse response h(n) with relatively duration, where n denotes the sequential number for the respective sample value. The impulse response can be divided into two main parts. There is first a considerable transient containing the greater part of the signal energy of the impulse. There is then a long, and substantially exponentially decaying part, a so-called tail. Negative values can also occur in the impulse response.
A first embodiment example of a filter in accordance with the invention is illustrated in Figure 3. The filter receives a digital signal x(n) as inpuk signal, this signal 1 ~1 ()7~,~
corresponding, for example, to ~he siynal x(n) in the apparatus accordiny to Figure 1. The input signal is applied to an FIR filter 11 directly, and to a plurality of IIR
filters 13-16 after delay by a time ~vin a delay means 12.
The IIR filters are suitably of the first degree, and have permanently set filter coefficients having mutually differing values. The output signal from the FI~ filter 11 is supplied to a summing means 22, and the output signals from 20 the IIR
filters 13-16 are each applied to their respective multiplication means 18-21. Each of the latter has an adaptive multiplication factor. These multiplication factors are assumed to have the values WO-W3, and they are adjusted in the way given below. The output signals from the FIR
filter 11 and from the multiplication means 18-21 are finally added to the summing means.
In accordance with the inventive concept, the first part of the impulse response is generated in the FIR filter 11 and the second part is generated as a linear combination of the output signals from the IIR filters 13-16. The weightings in the linear combination are here determined by the adaptive multiplication factors, or weighting factors W0-W3. By suitable delay of the input signal xtn) to the IIR filters, both parts of the impulse response can be generated independently of each other. The filters in accordance with the invention thus comprise two separate filter parts, a non-recursive filter part and a recursive filter part, the output signals of which are added.
The filter output signal is denoted y(n) and is subtracted from an arbitrary desired signal d(n) in a summing means 3.
A difference signal e(n) thus obtained occurs on a line 17 and is utilized both for updating the non-recursive filter part, i.e. the FIR filter 11, and the recursive filter part.
Updating the latter takes place by updating the adaptive weighting factors W0-W3 of the multiplication means 18-21.
The signals y(n), d(n) and e(n) and the summing mean~ 3 agree with corresponding signals and means in Figure 1, for 1 31 070?) example, but the field of application of the filter is of course not limited -to echo cancellation.
For the sake of completeness, it is pointed out that updating means are required both for the FIR filter 11 and the multiplication means 18-21, these updating means being generally known in connection with digital filters.
There is shown in Figure 4 a more detailed embodiment of the filter according to Figure 3. The FIR filter 11 conventionally comprises delay means 38-40, multiplication means 34-37, and summing means 31-33. The IIR filters 13-16 are of the firs~ degree, and each has its permanently set filter coefficient. These filters are also implemented conventionally and each comprises a summing means, e.g. 131, a delay means, e.g. 132, and a multiplication means, e.g.
133.
The multiplication means are each allocated a permanently set coefficient PO-P3, which have mutually different values, and which are thus the filter coefficients of the IIR filters.
Each of the delay means 38-40 included in the FIR filter 11 delay the input signal x(n) by a sample value, and together these correspond to the delay means 12 illustrated in Figure 3. In the illustrated example, ~r =3 T. Such a separate delay means is thus not required in practise but can he included in the FIR filter 25 instead. The summing means 22 in Figure 3 is shown in Figure 4 as a plurality of separate summing means 221-224.
As will be seen from above, the diffe.rence signal e(n) is used for both updating the FIR filter 11 and the adaptive weighting factors W0-W3 of the multiplication means 18-21 in the recursive filter part. The problem of minimizing the difference signal e(n) is equal to minimizing the sum of the square of the expression W0 X pOn ~ Wl x pln ~ W2 x p2n + W3 x p3n - f(n), where n goes from zero to infinity, P0-P3 are 1 ) 1 ()7 ~J r)) the permanent recursive filter coefficients and f(n) is the desired impulse response. This sum has a quadratic error area with only one minimum, since the weighting factors are only present linearly in the expression. This means that the recursive filter part can be updated according to the same method as the n~n-recursive filter part, e.g. according to the LMS method.
Some of the advantages achieved with the filter in accordance with the invention are that the difference signal is represented by a quadratic error area, simultaneously as the difference signal r~presents the actual error (as opposed to an equation error structure). In addition, the recursive filter part is always stable, since the poles of the individual IIR filters are not displaced. This depends in turn on the filter coefficients Po-P3 being permanent.
Some graphs are illustrated in Figure 5, and are examples of different impulse responses of the individual IIR filters in the recursive part of the filter. The transfer functions o~
the IIR filters 13-16 are denoted in turn by hO(n)-h3(n). It is assumed that the input signal to the filters is delayed by a plurality of sample values corresponding to the length of the impulse response of the FIR filter.
The filter coefficients PO-P3 are,`according to the example, 0.5, 0.75, 0.875 and 0.9375. The transfer functions will then be: hO(n) = 005n, hl(n) = 0.75n etc. Other coefficient values can of course be selected.
The part of the entire desired impulse response occurring to the left of the impulse responses illustrated in Figure 5, i.e. earlier than these, is generated in the FIR filter 11.
This is adapted such that its output signal ceases when the impulse responses according to Figure 5 start. It is pointed out, however, that the number of delay means in the FIR
filter included in the filter accordiny to Fiyure ~ is not adapted to the graphs in Figure 5.
1 -~,1 07C~
By linearly combining a plurality o~ given impulse responses in the way described above, it is possible to achi~ve impulse responses of very varying forms. Both positive and negative weighting ~actors WO-W3 can thus occur, of course. The long decaying part of the desired impulse response cannot always be imitated exactly. This does not make so much difference, however, since only a relatively small part of the energy of the entire desired impulse response is in this part. On the other hand, the first, major part of the impulse response which is generated by the FIR filter can be imitated rather precisely.
A second embodiment of a filter in accordance with the invention is illustrated in Figure 6. Further to the means included in the filter according to Figure 3, there is also a network denoted by 50 in this filter. The network 50 includes multiplication means and summing means, which are adapted to form linear combinations of the output signals of the IIR filters 13-16. These means are connected such that the multiplication means 18 obtains the output signal from the filter 13 in an unaltered condition. The multiplication means 19 obtains the sum of the output signal from khe filter 14 and the output signal from the filter 13 multiplied by a factor, and so on, according to the Figure. The linear combinations can be selected such that the input signals to the multiplications means 18, 21 will be orthogonal. These orthogonal impulse responses are then weighted by adaptive weighting factors, as with the filters according to figures 3 and 4. A change in a given weighting factor does not necessarily cause a change in the remaining weighting factors in this case. More rapid convergence is thus obtained. The number of calculations increases somewhat, however.
The filter in accordance with the invention can be used in different connections, when a relatively lony impulse response is desired and not only for adaptive echo cancellation. Of course, the number of IIR filters may be both more or less than just four, as illustrated in the 1 31 070~, examples. The implementation of the FIR and IIR filters can also be different from what has been shown in the examples.
Neither is it necessary to delay the input siynal to the IIR
filters. However, the delay results in the first part of the desired impulse response being generated solely by the FIR
filter, and the second part of the response being generated solely by the recursive filter part.
Claims (6)
- THE EMBODIMENTS OF THE INVENTION IN WHICH AN EXCLUSIVE
PROPERTY OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS:
l. An adaptive digital filter comprising a non-recursive part and a recursive part; said recursive part including a plurality of branches each having separate, permanently set recursive filters with mutually different impulse responses, respective multiplication means with an adaptive multiplication factor associated with each recursive filter, and summing means which in conjuction with said multiplication means form a linear combination of the output signals of the recursive filters; and means for generating a single signal from the output of said recursive and non-recursive parts to update the non-recursive part and the adaptive multiplication factors of said multiplication means in the recursive part. - 2. An adaptive filter as claimed in claim 1, wherein the recursive filters are of the first degree.
- 3. An adaptive filter as claimed in claim 2, wherein the filter also includes summing means for summing the output signal of the non-recursive part and said linear combination.
- 4. An adaptive filter as claimed in claim 3, wherein the filter also includes a delay means adapted such that an input signal applied to the filter is applied to the recursive filters after a predetermined delay.
- 5. An adaptive filter as claimed in any one of claims 1-4, wherein the filter also includes a network inserted between the recursive filters and said multiplication means, and which is adapted to form a linear combinations of the output signals of the recursive filters.
- 6. An adaptive filter as claimed in claim 3, wherein said single signal generating means is connected to the output of said summing means for summing the output signal of the non-recursive part and said linear combination.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| SE8802076A SE461308B (en) | 1988-06-03 | 1988-06-03 | ADAPTIVE DIGITAL FILTER INCLUDING A NON-RECURSIVE PART AND A RECURSIVE PART |
| SE8802076-3 | 1988-06-03 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CA1310708C true CA1310708C (en) | 1992-11-24 |
Family
ID=20372511
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA000597705A Expired - Lifetime CA1310708C (en) | 1988-06-03 | 1989-04-25 | Adaptive, digital filter including a non-recursive part and a recursive part |
Country Status (18)
| Country | Link |
|---|---|
| US (1) | US5014232A (en) |
| EP (1) | EP0347394B1 (en) |
| JP (1) | JPH03502634A (en) |
| KR (1) | KR960000843B1 (en) |
| CN (1) | CN1014288B (en) |
| AU (1) | AU609611B2 (en) |
| BR (1) | BR8906966A (en) |
| CA (1) | CA1310708C (en) |
| DE (1) | DE68905246T2 (en) |
| DK (1) | DK170319B1 (en) |
| ES (1) | ES2038449T3 (en) |
| FI (1) | FI93409C (en) |
| GR (1) | GR3007321T3 (en) |
| MX (1) | MX170248B (en) |
| NO (1) | NO301203B1 (en) |
| SE (1) | SE461308B (en) |
| TR (1) | TR24214A (en) |
| WO (1) | WO1989012360A1 (en) |
Families Citing this family (42)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5278552A (en) * | 1989-10-23 | 1994-01-11 | Jeco Company Limited | Indicator control circuit |
| DE69020889T2 (en) * | 1990-03-28 | 1996-03-14 | Sel Alcatel Ag | Method for determining the coefficients of an FIR filter in equalizers. |
| US5247474A (en) * | 1991-04-18 | 1993-09-21 | Fujitsu Ten Limited | Coefficients setting method of a reverberation unit |
| EP0511698B1 (en) * | 1991-04-23 | 1998-07-08 | Laboratoires D'electronique Philips S.A.S. | Semi-recursive adaptive equalizer |
| ES2038904B1 (en) * | 1991-09-10 | 1995-01-16 | Alcatel Standard Electrica | PROCEDURE AND DEVICE FOR ADAPTIVE CANCELLATION OF ACOUSTIC ECOS. |
| EP0543568A2 (en) * | 1991-11-22 | 1993-05-26 | AT&T Corp. | High resolution filtering using low resolution processors |
| US5402520A (en) * | 1992-03-06 | 1995-03-28 | Schnitta; Bonnie S. | Neural network method and apparatus for retrieving signals embedded in noise and analyzing the retrieved signals |
| US5337264A (en) * | 1992-06-01 | 1994-08-09 | Levien Raphael L | Time reversal gaussian approximation filter |
| US5615233A (en) * | 1992-07-22 | 1997-03-25 | Motorola, Inc. | Method for channel estimation using individual adaptation |
| US5416799A (en) * | 1992-08-10 | 1995-05-16 | Stanford Telecommunications, Inc. | Dynamically adaptive equalizer system and method |
| EP0660958B1 (en) * | 1992-09-21 | 1999-06-23 | Noise Cancellation Technologies, Inc. | Sampled-data filter with low delay |
| US6009082A (en) * | 1993-01-08 | 1999-12-28 | Multi-Tech Systems, Inc. | Computer-based multifunction personal communication system with caller ID |
| US5453986A (en) * | 1993-01-08 | 1995-09-26 | Multi-Tech Systems, Inc. | Dual port interface for a computer-based multifunction personal communication system |
| US5617423A (en) * | 1993-01-08 | 1997-04-01 | Multi-Tech Systems, Inc. | Voice over data modem with selectable voice compression |
| US5754589A (en) * | 1993-01-08 | 1998-05-19 | Multi-Tech Systems, Inc. | Noncompressed voice and data communication over modem for a computer-based multifunction personal communications system |
| US5452289A (en) * | 1993-01-08 | 1995-09-19 | Multi-Tech Systems, Inc. | Computer-based multifunction personal communications system |
| US5535204A (en) * | 1993-01-08 | 1996-07-09 | Multi-Tech Systems, Inc. | Ringdown and ringback signalling for a computer-based multifunction personal communications system |
| US5546395A (en) * | 1993-01-08 | 1996-08-13 | Multi-Tech Systems, Inc. | Dynamic selection of compression rate for a voice compression algorithm in a voice over data modem |
| US5812534A (en) * | 1993-01-08 | 1998-09-22 | Multi-Tech Systems, Inc. | Voice over data conferencing for a computer-based personal communications system |
| US5864560A (en) * | 1993-01-08 | 1999-01-26 | Multi-Tech Systems, Inc. | Method and apparatus for mode switching in a voice over data computer-based personal communications system |
| JPH0784993A (en) * | 1993-09-17 | 1995-03-31 | Fujitsu Ltd | Signal suppressing device |
| JP2872547B2 (en) * | 1993-10-13 | 1999-03-17 | シャープ株式会社 | Active control method and apparatus using lattice filter |
| US5757801A (en) * | 1994-04-19 | 1998-05-26 | Multi-Tech Systems, Inc. | Advanced priority statistical multiplexer |
| US5682386A (en) | 1994-04-19 | 1997-10-28 | Multi-Tech Systems, Inc. | Data/voice/fax compression multiplexer |
| FR2729024A1 (en) * | 1994-12-30 | 1996-07-05 | Matra Communication | ACOUSTIC ECHO CANCER WITH SUBBAND FILTERING |
| FI98015C (en) * | 1995-05-05 | 1997-03-25 | Unto Kalervo Laine | A method for modeling the signal spectrum and an apparatus for implementing the method |
| FI98177C (en) | 1995-06-01 | 1997-04-25 | Nokia Mobile Phones Ltd | Method and circuit arrangement for processing a signal containing interference |
| EP0896481B1 (en) * | 1997-08-05 | 2006-08-23 | Micronas Semiconductor Holding AG | Adaptive filter |
| US7242782B1 (en) | 1998-07-31 | 2007-07-10 | Onkyo Kk | Audio signal processing circuit |
| US6745218B1 (en) * | 1999-03-16 | 2004-06-01 | Matsushita Electric Industrial Co., Ltd. | Adaptive digital filter |
| US6813352B1 (en) * | 1999-09-10 | 2004-11-02 | Lucent Technologies Inc. | Quadrature filter augmentation of echo canceler basis functions |
| DE19955596A1 (en) | 1999-11-18 | 2001-06-13 | Infineon Technologies Ag | Device and method for echo cancellation in equilibrium transmission methods in duplex operation over a two-wire line |
| US6980592B1 (en) * | 1999-12-23 | 2005-12-27 | Agere Systems Inc. | Digital adaptive equalizer for T1/E1 long haul transceiver |
| US6480151B2 (en) | 2000-12-29 | 2002-11-12 | Lockheed Martin Corporation | GPS receiver interference nuller with no satellite signal distortion |
| US7079574B2 (en) | 2001-01-17 | 2006-07-18 | Radiant Networks Plc | Carrier phase recovery system for adaptive burst modems and link hopping radio networks |
| US6628707B2 (en) | 2001-05-04 | 2003-09-30 | Radiant Networks Plc | Adaptive equalizer system for short burst modems and link hopping radio networks |
| CN1774862A (en) * | 2003-04-17 | 2006-05-17 | 皇家飞利浦电子股份有限公司 | Adaptive filtering |
| US7406493B2 (en) * | 2004-03-17 | 2008-07-29 | Tektronix, Inc. | Up-sampling half-band reconstruction filtering |
| US7199964B2 (en) * | 2005-06-29 | 2007-04-03 | Seagate Technology Llc | Adaptive voltage-mode controller for a voice coil motor |
| NO322301B1 (en) * | 2005-07-13 | 2006-09-11 | Tandberg Telecom As | Small delay echo cancellation method and system. |
| WO2008041612A1 (en) * | 2006-09-29 | 2008-04-10 | Panasonic Corporation | Waveform equalizing device |
| RU198305U1 (en) * | 2020-02-26 | 2020-06-30 | Федеральное государственное бюджетное образовательное учреждение высшего образования "МИРЭА - Российский технологический университет" | ADAPTIVE DIGITAL FILTER FOR THE SUPPRESSION OF NON-FLUCTUATION INTERFERENCE |
Family Cites Families (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4495591A (en) * | 1981-02-27 | 1985-01-22 | The Regeants Of The University Of California | Pipelined digital filters |
| DE3116266A1 (en) * | 1981-04-24 | 1982-11-11 | TE KA DE Felten & Guilleaume Fernmeldeanlagen GmbH, 8500 Nürnberg | METHOD FOR EQUALIZING A DATA SIGNAL |
| DE3120434A1 (en) * | 1981-05-22 | 1982-12-16 | Standard Elektrik Lorenz Ag, 7000 Stuttgart | ADAPTIVE ECHOCOMPENSATION DEVICE FOR DIGITAL DUPLEX TRANSFER ON TWO-WIRE CABLES |
| JPS5834615A (en) * | 1981-08-24 | 1983-03-01 | Victor Co Of Japan Ltd | Iir digital filter |
| FR2515901A1 (en) * | 1981-11-04 | 1983-05-06 | Trt Telecom Radio Electr | MIC-DIFFERENTIAL TRANSMISSION SYSTEM WITH ADAPTIVE PREDICTION |
| US4791390A (en) * | 1982-07-01 | 1988-12-13 | Sperry Corporation | MSE variable step adaptive filter |
| DE3610382A1 (en) * | 1986-03-27 | 1987-10-01 | Ant Nachrichtentech | Circuit arrangement for adaptive echo cancellation in terminals for duplex transmission |
| US4803647A (en) * | 1986-05-30 | 1989-02-07 | Rca Licensing Corporation | Sampled data audio tone control apparatus |
| CA1271530A (en) * | 1986-07-14 | 1990-07-10 | Masaki Kobayashi | Adaptive digital filter |
| US5042026A (en) * | 1987-03-03 | 1991-08-20 | Nec Corporation | Circuit for cancelling whole or part of a waveform using nonrecursive and recursive filters |
| US4811360A (en) * | 1988-01-14 | 1989-03-07 | General Datacomm, Inc. | Apparatus and method for adaptively optimizing equalization delay of data communication equipment |
-
1988
- 1988-06-03 SE SE8802076A patent/SE461308B/en not_active IP Right Cessation
-
1989
- 1989-04-05 JP JP1505721A patent/JPH03502634A/en active Pending
- 1989-04-05 EP EP89850107A patent/EP0347394B1/en not_active Expired - Lifetime
- 1989-04-05 DE DE8989850107T patent/DE68905246T2/en not_active Expired - Fee Related
- 1989-04-05 WO PCT/SE1989/000176 patent/WO1989012360A1/en active IP Right Grant
- 1989-04-05 ES ES198989850107T patent/ES2038449T3/en not_active Expired - Lifetime
- 1989-04-05 BR BR898906966A patent/BR8906966A/en not_active IP Right Cessation
- 1989-04-05 AU AU35670/89A patent/AU609611B2/en not_active Ceased
- 1989-04-05 KR KR1019900700205A patent/KR960000843B1/en not_active IP Right Cessation
- 1989-04-07 US US07/334,712 patent/US5014232A/en not_active Expired - Lifetime
- 1989-04-19 TR TR89/0338A patent/TR24214A/en unknown
- 1989-04-25 CA CA000597705A patent/CA1310708C/en not_active Expired - Lifetime
- 1989-05-19 MX MX016126A patent/MX170248B/en unknown
- 1989-06-03 CN CN89103704A patent/CN1014288B/en not_active Expired
-
1990
- 1990-01-25 FI FI900406A patent/FI93409C/en not_active IP Right Cessation
- 1990-01-25 NO NO900357A patent/NO301203B1/en unknown
- 1990-01-26 DK DK021990A patent/DK170319B1/en not_active IP Right Cessation
-
1993
- 1993-03-11 GR GR930400407T patent/GR3007321T3/el unknown
Also Published As
| Publication number | Publication date |
|---|---|
| DK170319B1 (en) | 1995-07-31 |
| AU609611B2 (en) | 1991-05-02 |
| AU3567089A (en) | 1990-01-05 |
| DK21990D0 (en) | 1990-01-26 |
| EP0347394B1 (en) | 1993-03-10 |
| KR900702646A (en) | 1990-12-08 |
| NO900357D0 (en) | 1990-01-25 |
| BR8906966A (en) | 1990-12-18 |
| US5014232A (en) | 1991-05-07 |
| FI900406A0 (en) | 1990-01-25 |
| NO900357L (en) | 1990-01-25 |
| SE461308B (en) | 1990-01-29 |
| DK21990A (en) | 1990-01-26 |
| DE68905246D1 (en) | 1993-04-15 |
| ES2038449T3 (en) | 1993-07-16 |
| FI93409C (en) | 1995-03-27 |
| TR24214A (en) | 1991-07-02 |
| DE68905246T2 (en) | 1993-06-17 |
| EP0347394A1 (en) | 1989-12-20 |
| SE8802076D0 (en) | 1988-06-03 |
| GR3007321T3 (en) | 1993-07-30 |
| KR960000843B1 (en) | 1996-01-13 |
| MX170248B (en) | 1993-08-12 |
| FI93409B (en) | 1994-12-15 |
| NO301203B1 (en) | 1997-09-22 |
| JPH03502634A (en) | 1991-06-13 |
| SE8802076L (en) | 1989-12-04 |
| CN1038193A (en) | 1989-12-20 |
| CN1014288B (en) | 1991-10-09 |
| WO1989012360A1 (en) | 1989-12-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA1310708C (en) | Adaptive, digital filter including a non-recursive part and a recursive part | |
| CA2010652C (en) | Echo canceller having fir and iir filters for cancelling long tail echoes | |
| US4268727A (en) | Adaptive digital echo cancellation circuit | |
| AU698609B2 (en) | Echo canceller having Kalman filter for optimal adaptation | |
| US5204854A (en) | Adaptive hybrid | |
| CA1168775A (en) | All digital lsi line circuit for analog lines | |
| Kawamura et al. | A tap selection algorithm for adaptive filters | |
| NZ198653A (en) | Digital recursive automatic equalizer | |
| US4467441A (en) | Adaptive filter including controlled tap coefficient leakage | |
| US4894864A (en) | Interface circuit | |
| US4628157A (en) | Bidirectional adaptive voice frequency repeater | |
| US5315621A (en) | Adaptive nonrecursive digital filter and method for forming filter coefficients therefor | |
| GB2086197A (en) | Digital two-to-four wire converter for full duplex signals | |
| US5111418A (en) | Method and network configuration for obtaining the gradient of the output signals of a given network for processing discrete-time signals relating to the network parameters | |
| IE62809B1 (en) | Adaptive digital filter including a non-recursive part and a recursive part | |
| Hoya et al. | Application of the leaky extended LMS (XLMS) algorithm in stereophonic acoustic echo cancellation | |
| KR100475771B1 (en) | Device and method for echo compensation in a two-wire full duplex channel transmission method | |
| CA2137130A1 (en) | Method for the adaptive control of a digital echo canceller in a telecommunication system | |
| US4852036A (en) | Adaptive digital filter and an echo canceler incorporating the same | |
| Park et al. | On acoustic-echo cancellation implementation with multiple cascadable adaptive FIR filter chips | |
| CA1233254A (en) | Two terminal impedance circuit | |
| Perez-Meana et al. | A continuous time structure for filtering and prediction using Hopfield Neural Networks | |
| Abousaada et al. | Performance analysis of an efficient AIFIR echo-tail canceller | |
| Tahernezhadi et al. | A DSP-based lattice-pole-zero acoustic echo canceller | |
| Higa et al. | A gradient type algorithm for blind system identification and equalizer based on second order statistics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| MKLA | Lapsed |