JPH0738479A - Adaptive receiver - Google Patents

Abstract

PURPOSE:To obtain an adaptive receiver capable of making stable matching filtering even when a clock phase deviation takes place with respect to the adaptive receiver conducting the diversity synthesis by plural adaptive matching filters and eliminating inter-code interference by a discrimination feedback equalizer. CONSTITUTION:The adaptive receiver uses plural adaptive matching filters 101-10N receiving input signals received separately to input a signal maximizing the S/N to a discrimination feedback equalizer 30 through a synthesizer 20. Then the discrimination feedback equalizer 30 extracts discrimination data S from which an inter-code interference is eliminated and the discrimination data S correct the tap coefficient of the adaptive matching filters 101-10N. In each of the adaptive matching filters 101-10N, a reference tap complex multiplier 14 is provided in an output stage of a synthesizer 13. A signal r0 of a center tap is inputted directly to the synthesizer 13.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an adaptive receiver, and more particularly to an adaptive receiver which performs diversity combining by a plurality of adaptive matched filters and cancels intersymbol interference by a decision feedback equalizer.

[0002]

2. Description of the Related Art Conventionally, there is known an adaptive receiver which performs diversity combining by a plurality of adaptive matched filters and cancels intersymbol interference by a decision feedback equalizer (Kojiro Watanabe: "Adaptation in multipath transmission line". Reception system ", The Institute of Electronics and Communication Engineers, Communication System Study Group, February 1979 (CS78
-203)). As shown in FIG. 2, this conventional adaptive receiver has N adaptive matched filters (AMF) 40 ₁ to 40 _N.
, A combiner 20 for combining the output signals of these N adaptive matched filters 40 ₁ to 40 _N , and a decision feedback equalizer (DFE) to which the output combined signal of the combiner 20 is input.
30 and is configured to control the tap coefficients of the adaptive matched filters 40 ₁ to 40 _N by the output signal of the decision feedback equalizer 30.

This adaptive receiver performs optimum reception on a multipath line of a power limiting system, and the adaptive matched filters 40 ₁ to 40 _N of each diversity branch cause the signal-to-noise power ratio of the baseband input signal ( Since the signal strength is enhanced by maximizing the S / N ratio) and then maximal ratio combining is performed by the combiner, the residual intersymbol interference after diversity combining by the combiner 20 can be removed by the decision feedback equalizer 30. Enables digital transmission over severe multipath fading lines. This adaptive receiver has already been put to practical use in tropospheric scatter communication, which is a typical multipath line.

The adaptive matched filters 40 ₁ to 40 _{N, which} are the main parts of this adaptive receiver, have the same structure and are connected in cascade as shown in FIG. 2 and each have a delay time of T / 2 (T is a symbol period). The delay elements 41 _{1 to} 41 ₄ , the complex multipliers 42 _{1 to} 42 ₄ into which the input signals of the delay elements 41 _{1 to} 41 ₄ are branched and input, and the complex multipliers into which the output signals of the delay element 41 ₁ are input 42 ₅ and these complex multipliers 42
Combiner 43 for combining _{1-42 5} output signals, respectively
And a tap coefficient correction circuit 44.

Next, the adaptive matched filters 40 ₁ to 40
The operation of _N will be described with reference to FIG. In FIG.
The same reference numerals are given to the same components as. In FIG. 3, 41 is a delay element corresponding to the delay elements 41 _{1 to} 41 ₄ , and 42 is a complex multiplier corresponding to the complex multipliers 42 _{1 to} 42 ₄ . 3A is an example of a transmission path impulse response, FIG. 3B is an example of a tap coefficient distribution of an adaptive matched filter, and FIG. 3C is an example of a transmission system impulse response at the output of an adaptive matched filter. FIG. 6D shows an example of the tap coefficient distribution of the adaptive matched filter when the reference tap of the adaptive matched filter is shifted, and FIG. 6E shows the transmission system impulse response at the output of the adaptive matched filter when the reference tap is shifted. An example is shown respectively.

Now, the impulse response of the transmission line is shown in FIG.
If it is shown in (A), the transversal filter composed of the delay element 41, the complex multiplier 42, and the combiner 43 in this case, that is, the adaptive matched filter, is the impulse response time of FIG. The tap coefficient distribution shown in FIG. 6B, which is an inverted complex conjugate response, is estimated. The response of complex conjugate time reversal, 3 a sampling value of T / 2 interval of the impulse response of (A) h _{-2,
h -1, h 0, h 1} , h 2 Distant, similarly, the sampling value of T / 2 interval of the tap coefficient distribution of the adaptive matched filter as shown in FIG. _{(B), W -2, W} - ₁ , W ₀ , W ₁ , W
_{If it is 2} , it is expressed by the following equation.

W _-2 = h ₂ ^* (1) W _-1 = h ₁ ^* (2) W ₀ = h ₀ ^* (3) W ₁ = h _-1 ^* (4) W ₂ = h _-2 ^* ( 5) In the above formula, * represents a complex conjugate. The main response of the transmission path impulse response is h ₀ at time t = ₀ , and the energy is maximum as shown in FIG. 3 (A).
The corresponding adaptive matched filter response is W ₀ in FIG. 3B, which is the tap coefficient of the adaptive matched filter, and is the central complex multiplier (42 _{3 in} FIG. 2) of the complex multipliers 42. ). This center tap is referred to as a reference tap because it serves as a reference timing for impulse response. Similarly, other sampling values W _i (i = −
2, -1, 1, 2) is also multiplied as a tap coefficient by a complex multiplier (42 _{i in} FIG. 2) in a corresponding order among the plurality of complex multipliers 42.

Thus, the impulse response of the input transmission line (FIG. 3A) is a time-reversed complex conjugate response (FIG. 3A).
The convolution of (B)) is called matched filtering in communication theory, and the impulse response at the output of the adaptive matched filter has a symmetrical shape as shown in FIG. That is, the time-dispersed impulse response (FIG. 3A) converges at the reference time (t = 0), and the signal is strengthened. This is the operation of maximizing the SN ratio by the adaptive matched filter.

The first point in matched filtering is to estimate its tap coefficient, and the usual method is as follows.

First, the decision data signal is input to the replica filter and the received input signal is estimated. Next, the error between this estimated signal and the actual input signal is taken, and the tap coefficient of the matched filter is estimated by an algorithm that minimizes this root mean square. However, each of the adaptive matched filters 40 ₁ to 40 _{N in} the adaptive receiver shown in FIG. 2 does not use a replica filter in order to more easily estimate the matched filter tap coefficient, but uses a correlation method based on the decision data signal. I am using.

That is, in order to explain this correlation method, the transmission side generates a transmission symbol sequence {S _n } ..., S _-2,
When transmitted every T cycle in the order of S _-1 , S ₀ , S ₁ , S ₂ , ..., the received signal is a convolution operation of this transmission symbol sequence and the channel impulse response (Fig. 3 (A)). Given in. At this time, the received signals r _-2 , r _-1 , r ₀ , r ₁ , r ₂ on each tap of the adaptive matched filter 40 ₁ of FIG.

R ₋₂ =. ． ． + H ₂ · S ₀ + h ₀ · S ₁ + h ₋₂ · S ₂ +. ． ． (6) r ₋₁ =. ． ． + H ₃ · S ₋₁ + h ₁ · S ₀ + h ₋₁ · S ₁ +. ． ． (7) r ₀ =. ． ． + H ₂ · S ₋₁ + h ₀ · S ₀ + h ₋₂ · S ₁ +. ． ． (8) r ₁ =. ． ． + H ₁ · S ₋₁ + h ₋₁ · S ₀ + h _-3 · S ₁ +. ． ． (9) r ₂ =. ． ． + H ₂ · S ₋₂ + h ₀ · S ₋₁ + h ₋₂ · S ₀ +. ． ． (10) The tap coefficient correction circuit 44 of FIG.
Received signal r _i on the tap (i = -2, -1, 0, 1,
2) and the decision data signal S from the decision feedback equalizer 30 are subjected to correlation calculation, and the obtained correlation value is used as the tap coefficient of the corresponding tap. That is, the i-th tap coefficient Wi is expressed by the following equation.

W _i = E [r _i ^* · S] (11) Here, E [] indicates time average (expected value) processing, and *
Indicates a complex conjugate. Further, when the decision data signal S from the decision feedback equalizer 30 is S ₀ at time t = 0, the above tap coefficient W _i is expressed by the following equation.

W _i = E [r _i ^* · S ₀ ] (12) Substituting equations (6) to (10) into equation (12) above, the autocorrelation power for S is set to “1”, If the autocorrelation when one symbol or more is separated is set to "0" and the calculation is advanced, the following equation is obtained.

W ₋₂ = E [(... + h ₂ S ₀ + h ₀ S ₁ + h ₋₂ S ₂ + ...) ^* · S ₀ ] = h ₂ (13) W ₋₁ = E [(. _{.. + h 3 S -1 + h} 1 S 0 + h -1 S 1 + ...) * · S 0] = h 1 (14) W 0 = E [(... + h 2 S -1 + h 0 S 0 + H ₋₂ S ₁ + ...) ^* · S ₀ ] = h ₀ (15) W ₁ = E [(... + h ₁ S ₋₁ + h ₋₁ S ₀ + h ₋₃ S ₁ + ...) ^* · S ₀ ] = h ₋₁ (16) W ₂ = E [(... + h ₂ S ₋₂ + h ₀ S ₋₁ + h ₋₂ S ₀ + ...) ^* · S ₀ ] = h ₋₂ (17) It can be seen that the correlation calculation results of the above equations (13) to (17) match the equations (1) to (5) indicating the tap coefficient for operating as a matched filter. Therefore,
Even if the replica filter is not used, the tap correction circuit 44
The tap coefficient of the adaptive matched filter is obtained by the correlation calculation expression ((11) expression) in.

The second point in matched filtering is that it has a timing control function. It is well known that this timing function is realized by setting transversal filters at fractional intervals. Here, the fractional interval means that the delay time of the delay elements 41 _{1 to} 41 _{4 in} FIG. 2 is set to a value which is half the symbol period T.

The timing control function means to match the received signal timing which has been changed in the propagation path with the clock timing phase of the receiver. Specifically, the transmitted symbols propagate in independent diversity branches. The propagation path of each branch changes with time,
The propagation delay time is changing every moment. Therefore, the reception timing of the reception symbol of each diversity branch undergoes independent fluctuation and does not match.

Therefore, if the diversity combining is performed as it is, the main responses of the respective branches do not coincide with each other, and the combined impulse response includes not only the delay dispersion due to the multipath but also the dispersion due to the timing deviation between the branches. . This means that the combined impulse response deviates further from the Nyquist distortion free condition. Therefore, it is necessary to absorb the timing deviation between the branches. Therefore, the adaptive matched filters 40 ₁ to 40 _N of FIG. 2 which are fractionally spaced transversal filters absorb the timing phase shift by the following operation.

The received signals r _-2 , r _-1 , r ₀ , r ₁ and r ₂ are distributed in this order on each tap composed of the delay elements 41 _{1 to} 41 ₄ . These are received signals sampled at T / 2 intervals with respect to each other. Therefore, when focusing on the received signal r ₀ of the center tap, the symbol S _{0 in the} equation (8) is
Is also distributed in the received signals r ₋₁ and r ₁ on the front and rear taps as shown in Expressions (7) and (9).

Here, the received signal r _-1 shown in equation (7)
S ₀ in is only a time delay T / 2 than S ₀ in the received signal r ₀ of the center tap. On the other hand, S ₀ in the received signal r ₁ expressed by the equation (9) is the received signal r of the center tap.
It has advanced only in time T / 2 than the S ₀ in _0. In this case, if the tap coefficient W ₁ is made larger than the tap coefficient W ₀ , the timing of S ₀ can be advanced equivalently at the output of the combiner 43. Conversely, tap coefficient W _-1
When is larger than W ₀ , S ₀ delayed in time from the center tap is output from the combiner 43, so that the timing of S ₀ can be delayed equivalently.

Therefore, the adaptive matched filter utilizes this property to perform timing control to match the reception timings of the reception symbols of the diversity branches and enable effective diversity combining.

By the way, the tap coefficient W _i for timing control is obtained by the correlation method of the first point. In the correlation method described above, the decision feedback equalizer 30
The output determination data S of is used. This judgment data S
Is based on the timing of the clock recovery circuit built in the decision feedback equalizer 30. Therefore, all the correlation processing of the adaptive matched filter is governed by the received clock extracted from the decision data S, that is, the timing of the clock phase of the clock recovery circuit in the decision feedback equalizer 30.

Explaining this with reference to FIG. 3, the rising timing of the reception clock is the time t = in FIG.
Corresponds to 0. In addition, as shown in FIG. 3C, in the matched filter output, the main response is the reference timing (t) as described above in the normal operation by the matched filtering by the taps arranged at T / 2 intervals around this. =
0). This operation is exactly the maximization of the SN ratio by the adaptive matched filter, and at the same time, it is the timing function of adjusting the phase with the reference timing (t = 0).

[0024]

However, the conventional adaptive receiver described above feeds back the judgment data retimed by the reception clock as described above to feed back the adaptive matched filter 40.
_{Since 1} to 40 _N are operated and controlled by the clock phase of the clock recovery circuit in the decision feedback equalizer 30, a feedback system using a clock as a medium is constructed. When the clock phase of the circuit is slightly shifted due to fading or some other reason,
The clock phase of the determination data S fed back to the adaptive matched filters 40 ₁ to 40 _N is also slightly shifted, and the timing control of each of the adaptive matched filters 40 ₁ to 40 _N is shifted.

This timing control deviation causes the reference tap coefficient W _0, which is the main response, to be made slightly smaller, and instead, one of the preceding and following tap coefficients W _-1 and W ₁ is set to be smaller than the original value. Also make it a little larger. Then, the main response of the transmission system impulse response at the outputs of the adaptive matched filters 40 ₁ to 40 _N is time t = 0 shown in FIG.
There is some deviation from the front or back. Then, this shift is input to the clock recovery circuit inside the decision feedback equalizer 30, and the reception clock phase shifts again.

As described above, in the conventional adaptive receiver, since the feedback loop system using the clock as a medium is constructed, the deviation of the clock phase of the clock recovery circuit propagates through the loop and is fed back to the clock recovery circuit again. An increase phenomenon of clock phase shift occurs. Adaptive matched filter 4
Since each of 0 ₁ to 40 _N has taps formed at T / 2 intervals, endless timing control operation is possible. Therefore, the phenomenon of increasing the clock phase shift is
It may continue until the main response settles on one end of the matched filter.

FIG. 3D shows an example of the transmission system impulse response at the output of the adaptive matched filter when the clock phase is shifted by T due to the clock phase shift.
As shown in FIG. 6D, the reference tap is located not at the center of the transversal filter but at W ₋₂ of the first tap.

In this case, the impulse response of FIG. 3D is h ^* (-t + T), which is offset by T from the original time-reversed complex conjugate h ^* (-t) for the adaptive matched filter. Therefore, the trailing edge of the transmission line response (Postcurso
The information of h ₁ and h ₂ which is r) is not included in the impulse response of FIG. 3D and cannot be multiplied by the complex multiplier. Therefore, correct matched filtering is not performed, and the output of the adaptive matched filter does not perform symmetrization of the impulse response and convergence of time-dispersed energy, as shown in FIG. This action creates a bad effect by itself, more than the effect of multipath distortion due to propagation.

As described above, in the conventional adaptive receiver, the feedback system using the reception clock phase as a medium is constructed,
Adaptive matched filter 40 ₁ to 40 _N reference tap position of the adaptive matched filter 40 ₁ to 40 _N by the interaction of the timing functions may become unstable, and the shift from the center tap,
There is a problem that the normal matched filter function may be impaired.

The present invention has been made in view of the above points,
An object of the present invention is to provide an adaptive receiver that solves the above problems by providing a complex multiplier at the reference tap position in the final output stage of an adaptive matched filter.

[0031]

SUMMARY OF THE INVENTION To achieve the above object, the present invention combines signals with maximized signal to noise power ratio by a plurality of adaptive matched filters supplied with separately received input signals. Is input to the decision feedback equalizer through the input / output unit, the decision feedback equalizer removes the decision data from which intersymbol interference is removed, and the tap coefficient of the plurality of adaptive matched filters is corrected by the decision data. In the receiver, each of the plurality of adaptive matched filters includes a delay unit, a first multiplication unit, a combination unit, a second multiplication unit, and a tap coefficient correction circuit.

Here, the above delay means outputs the input signal T /
It delays by 2 (where T is the symbol period of the input signal) and outputs a plurality of delayed signals having different delay times. Further, the first multiplication means respectively corresponds to the input signal and the other delayed signals except the delayed signal of the reference tap whose delay time is the median value among the plurality of delayed signals and the input signal. Multiply the tap coefficient of separately. Also,
The synthesizing means synthesizes the output signal of the first multiplying means and the delayed signal of the reference tap, respectively. The second multiplication means multiplies the output signal of the combining means by a second tap coefficient to be multiplied by the delay signal of the reference tap, and outputs the multiplied signal to the combiner. Further, the tap coefficient correction circuit calculates the first value by performing a correlation calculation between the determination data and the plurality of delay signals.
And the second tap coefficient are sequentially calculated.

[0033]

According to the present invention, the second tap coefficient to be multiplied by the delay signal of the reference tap among the plurality of delay signals from the delay means and the input signal is output by the second multiplying means to the output of the synthesizing means. By multiplying the combined signal, the maximum ratio combining of each diversity branch is realized. Further, a first tap coefficient obtained by normalizing the conventional tap coefficient with the second tap coefficient is applied to the delay signal at the tap position other than the reference tap among the plurality of delay signals by the first multiplication means. By multiplying, the matched filtering in each diversity branch is realized.

Therefore, in the present invention, the first tap coefficient multiplied by the first multiplying means can be set to a value smaller than the second tap coefficient multiplied by the second multiplying means. The reference tap position can always be held at the center tap position of the adaptive matched filter.

[0035]

Next, an embodiment of the present invention will be described.
FIG. 1 shows a block diagram of an embodiment of the present invention. In FIG.
The same components as in FIG. In FIG. 1, N (N is a natural number of 2 or more, usually a power of 2) adaptive matched filters 10 ₁ to 10 _N are N.
Each input signal is provided in a one-to-one correspondence. The input signal is, for example, a baseband signal obtained by separately frequency-converting signals received by N receiving antennas in the microwave band.

In this adaptive receiver, the signal-to-noise power ratio (SN ratio) of the baseband input signal is maximized by the adaptive matched filters 10 ₁ to 10 _N of each diversity branch, and then the maximum ratio combining is performed by the combiner. The decision feedback equalizer 30 removes residual intersymbol interference after diversity combining by the combiner 20, and the diversity combined and demodulated data (decision data) from the decision feedback equalizer 30 is removed. the basic operation is configured to output is the same as the conventional configuration of the adaptive matched filter 10 ₁ to 10 _N is different from the conventional adaptive matched filter 40 ₁ to 40 _N.

Since the adaptive matched filters 10 ₁ to 10 _N have the same configuration, the adaptive matched filter 10 ₁ will be described as a representative example. As shown in FIG. 1, the adaptive matched filters 10 ₁ are connected in cascade. Delay time is T / 2
(T is a symbol period) delay elements 11 _{1 to} 11 ₄ , complex multipliers 12 ₁ and 12 ₂ into which input signals of the delay elements 11 ₁ and 11 ₂ are branched, and input / output of the delay element 11 ₄ . Complex multipliers 12 ₃ and 1 to which signals are respectively input
2 ₄ , a combiner 13 for combining the output signals of the complex multipliers 12 _{1 to} 12 _{4 and} the output signal of the delay element 11 ₂ , respectively, and a complex multiplier 1 provided on the output side of the combiner 13.
4 and a tap coefficient correction circuit 15.

That is, the adaptive matched filter 1 of the present embodiment.
0 ₁ , delay elements 11 ₁ to ₁ 1 constituting the delay means
1 wherein the _fourth input signal or output signal is input first
The complex arithmetic unit constituting the multiplication means of _{No. 1} is composed of 12 _{1 to} 12 ₄ and is not provided in the center tap to be the reference tap, and the output signal of the delay element 11 ₂ which is the delay signal of the reference tap is always combined as it is. Since it is input to the device 13, the reference tap is secured.

However, if the complex multiplier is left removed from the reference tap as it is, the amplitude and phase of the signal corresponding to the reference tap cannot be controlled and the signal does not function as a matched filter, or even the maximum diversity synthesis is performed. It will be impossible.

Therefore, in this embodiment, instead of removing the complex multiplier from the reference tap, the complex multiplier 14 is provided at the final output stage of the adaptive matched filter, that is, the output side of the combiner 13. The complex multiplier 14 constitutes the second multiplication means, and the second tap coefficient W _{0 to} be multiplied is obtained by the tap coefficient correction circuit 15 by the following equation.

W ₀ = E [r ₀ ^* · S] (18) (where, r ₀ is a delay signal of the reference tap) Therefore, when only the reference tap is focused, the amplitude / phase control as in the conventional case is performed. Is possible.

However, since the complex multiplier 14 multiplies the output combined signal of the combiner 13 by the tap coefficient W _{0 of} equation (18), the tap coefficient described above is applied to signals from taps other than the reference tap. W ₀ is commonly multiplied. That is, the tap coefficients of the adaptive matched filter 10 ₁ W _k (except k = -2, -1, 1) will be multiplied by a tap coefficient multiplied by W ₀ on the signal of each tap to, the intact normal adaptive Matched filtering becomes impossible.

Therefore, in this embodiment, the complex multipliers 12 ₁ ...
The first tap coefficient W _{k to} be multiplied by 12 ₄ is calculated by the tap coefficient correction circuit 15 based on the following equation.

W _k = E [r _k ^* · S] / W ₀ (19) As can be seen from the above equation, the first tap coefficient W _k is normalized by the second tap coefficient W ₀ .

As a result, the signals other than the reference taps are multiplied by the tap coefficient W _k expressed by the equation (19) in the complex multipliers 12 _{1 to} 12 ₄ , and the complex multiplier 14 further calculates the tap coefficient W _k. Multiplying by ₀ is equivalent to finally being multiplied by the tap coefficient for regular adaptive filtering.

The tap coefficient W ₀ is a correlation value between the delay signal r ₀ of the reference tap and the judgment data S, as shown in the equation (18), and originally includes the main response component of the transmission path impulse response. Therefore, this correlation value is larger than the values of other tap coefficients. Further, the tap coefficient W _k is a value obtained by normalizing the correlation value between the signal r _k other than the reference tap and the determination data S by W ₀ as shown in the equation (19). Therefore, the output signals of the complex multipliers 12 _{1 to} 12 ₄ have smaller amplitudes than the delay signal r ₀ of the reference tap.

This means that the reference tap position is always held at the center tap position of the adaptive matched filter 10 ₁ , and the other taps are the leading edge (Pre) of the impulse response other than the main response.
current) or the trailing edge with a time-reversed complex conjugate. Therefore, according to this embodiment, even if the phase of the output clock of the clock recovery circuit in the decision feedback equalizer 30 is slightly shifted, the reference tap position does not shift from the central position of the adaptive matched filter. Since the main response of the impulse response of the adaptive matched filter output can always be held at the reference timing position, it is possible to prevent the above phenomenon of increasing the clock phase shift.

[0048]

As described above, according to the present invention,
Even if the phase of the output clock of the clock recovery circuit in the decision feedback equalizer is slightly deviated, the reference tap position can always be held at the center tap position of the adaptive matched filter. This can be prevented in advance, so that stable adaptive matched filtering can always be performed. Therefore, according to the present invention, stable adaptive equalization and timing control can be performed even for severe multipaths such as minimum phase transition or non-minimum phase transition fading, and even in severe multipath fading lines, it is even more effective than before. A stable reception operation can be performed.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an embodiment of the present invention.

FIG. 2 is a configuration diagram of a conventional example.

FIG. 3 is a diagram for explaining the operation of an adaptive matched filter and conventional problems.

[Explanation of Codes] 10 ₁ to 10 _N Adaptive Matched Filter 11 _{1 to} 11 ₄ Delay Element 12 _{1 to} 12 ₄ First Complex Multiplier 13, 20 Combiner 14 Second Complex Multiplier 15 Tap Coefficient Correction Circuit 30 Judgment Feedback equalizer

Claims (2)

[Claims] 1. A decision feedback equalizer is provided with a signal whose signal-to-noise power ratio is maximized by a plurality of adaptive matched filters to which separately received input signals are supplied, and which is input to the decision feedback equalizer. In the adaptive receiver configured to correct the tap coefficients of the plurality of adaptive matched filters according to the determination data while extracting the determination data from which the intersymbol interference is removed by the form equalizer, each of the plurality of adaptive matched filters is A delay unit for delaying the input signal in units of T / 2 (where T is a symbol period of the input signal) and outputting a plurality of delay signals having different delay times; and a plurality of delay signals from the delay unit, Of the input signals, other delay signals except the delay signal of the reference tap whose delay time is the median value and the input signal are respectively multiplied by the corresponding first tap coefficients. Deviating first multiplying means, combining means for respectively combining the output signal of the first multiplying means and the delay signal of the reference tap, and the output signal of the combining means should be multiplied by the delay signal of the reference tap Second multiplication means for multiplying by a second tap coefficient and outputting to the combiner, and a tap coefficient for sequentially calculating the first and second tap coefficients by a correlation calculation between the determination data and the plurality of delay signals. An adaptive receiver having a configuration including a correction circuit. 2. The tap coefficient correction circuit is configured such that W _k = E [r _k ^* · S for a signal r _{k at} a tap position k (where k is a positive or negative integer other than 0) by the delay means. ] / W ₀ (where E [] is the time averaging process, * is the complex conjugate, S is the determination data, and W ₀ is the second tap coefficient.) Tap coefficient of W _k
And W ₀ = E [r ₀ ^* · S] (where, r ₀ is a delay signal of the reference tap), and the second tap coefficient W ₀ is calculated based on the formula Receiving machine.