JPH05110617A - Adaptive maximum likelihood series estimating device - Google Patents

Abstract

PURPOSE:To compensate a phase variation amount over a frequency offset of a wide range. CONSTITUTION:The device is provided with a phase rotating section 40 rotating a phase of a reception signal Yn to compensate a phase variation of the reception signal, a Viterbi algorithm processing section 60, a transmission line estimating section 70 estimating the impulse response of a transmission line and a phase estimating section 80A estimating the phase to give the phase estimating value to the phase rotating section 40. Then the phase estimating section 80A and the phase rotating section 40 compensate the phase variation amount. In this case, an integration means 90 is provided in a loop filter 82A to implement in the compensation of the phase fluctuation by a frequency offset and phase jitter by using a quadratic phase synchronization loop, thereby compensating a steady-state phase rotation by the frequency offset. Thus, the transmission symbol is estimated with less error ratio over the wide range of frequency offset.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an adaptive maximum likelihood sequence estimator used in an equalizer or the like for compensating for distortion of a transmission path and obtaining a correct transmission signal in a receiver for digital communication, and more particularly, a carrier frequency offset. The present invention relates to an adaptive maximum likelihood sequence estimator applied to a phase compensation type adaptive equalizer or the like that performs equalization while compensating for the phase fluctuation due to.

[0002]

2. Description of the Related Art Conventionally, as a technique in such a field,
For example, some documents were described in the following documents. Literature; IEE Trance Action On
Communications (IEEE Transaction on Communic
ations), COM-22 [5] (1974-5) (US) G.I. U
ngerboeck “Adaptive Maximum-Like Lifted Receiver Fore-Carry-Modulated Data-Transmission Systems (A
daptive Maximum-Likelihood Receiver for Carrier-Mod
Calculated Data-Transmission Systems) "P.624-6
36 In recent years, digital mobile communication has been rapidly developed, but in land mobile communication, a large number of interfering waves with delays and high-speed movement of mobile terminals cause frequency-selective fading, resulting in reception signals. Since the waveform is significantly distorted, it is necessary to compensate for this distortion by the equalizer. The maximum likelihood sequence estimation applied to this equalizer is the most effective equalization for obtaining correct transmission data from the received signal waveform distorted due to the delay characteristic of the transmission line, such as frequency selective fading. It is one of the methods.

FIG. 2 is a block diagram showing a configuration example of a transmitter / receiver in a conventional digital mobile communication. The transceiver input data b _m has a transmitter 10 which transmits signals s _c (t) is based on, the signal s _c (t) is the transmission line 20
The signal is received by the receiver 30 via. The transmitter 10 includes an encoder 11, a transmission low-pass filter (hereinafter, referred to as a transmission LPF) 12, and a modulator 13. Further, the receiver 30 includes a demodulator 31, a reception low-pass filter (hereinafter, referred to as reception LPF) 32, an adaptive equalizer 33, and a decoder 34.

In the transmitter 10, the input data b _m is converted into a transmission symbol x _n by the encoder 11, and the transmission LPF 12 limits the band to generate a transmission complex baseband signal s (t), which is sent to the modulator 13. send. In the modulator 13, the signal s
Modulate (t) with a carrier of frequency f _{c to} obtain the signal s
_c (t) is transmitted to the receiver 30 via the transmission line 20.
In the receiver 30, the demodulator 31 performs synchronous detection at a frequency equal to the transmission carrier frequency f _c , converts the signal r _c (t) passing through the transmission line 20 into a complex baseband signal r (t), and further receives the signal. A band-limited reception complex baseband signal y (t) is obtained through the LPF 32. This signal y
(T) is sampled at symbol intervals T.

The adaptive equalizer 33 compensates the characteristics of the transmission line 20 due to frequency selective fading from the sample value y _n of the signal y (t) and estimates the transmission symbol. Here, since the phase of the received signal changes due to the frequency offset and the phase jitter of the transmission carrier frequency and the demodulation frequency of the receiver 30, the adaptive equalizer 33 uses the maximum likelihood sequence estimation to
The transmission symbol is estimated while compensating for the phase fluctuation of the received signal. Finally, the decoder 34 estimates the transmitted symbol E
Decode X _n (where E represents the estimate) to obtain the transmitted data Eb _m .

Maximum-likelihood sequence estimation is performed when a received signal sequence BY _n = {y ₁ , y ₂ , ..., Y _N } (where B represents a vector) in a finite interval is obtained. With the 20 impulse responses h (t) known, the transmission symbol sequence BX _N = which has the highest probability (likelihood) of achieving BY _N.
{X ₁ , x ₂ , ..., X _N } are estimated and are efficiently calculated using the Viterbi algorithm known as the decoding method of the convolutional code, as described in the above-mentioned document.

FIG. 3 is a functional block diagram of a conventional adaptive maximum likelihood sequence estimator adapted to the adaptive equalizer 33 in FIG. This adaptive maximum likelihood sequence estimator is configured by an individual circuit using an integrated circuit or the like, or program control using a processor, and the phase rotating unit 40 and the delay unit 5 are provided.
0, Viterbi algorithm processing unit 60, transmission path estimation unit 7
0 and the phase estimation part 80 are provided. A solid line connecting the functional blocks represents a real number, a one-dot chain line represents a complex number, and a two-dot chain line represents a complex vector.

The phase rotation unit 40 phase-rotates the sample value Y _n of the received signal based on the estimated phase value Eφ _n to compensate the phase variation due to the frequency offset and the phase jitter cr _n (however, c represents a compensation) and is composed of an arithmetic means 41 and a multiplication means 42. The delay means 50 is for compensating the decision delay of the Viterbi algorithm, delays the sample value cr _n of the received signal for a predetermined time, and delays the delayed value cr _n.
It has a function of giving _nM to the transmission path estimation unit 70 and the phase estimation unit 80. The Viterbi algorithm processing unit 60
The sample value cr _n is input, the transmission symbol is estimated according to the Viterbi algorithm based on the impulse response estimation value Eh _j of the transmission line 20 from the transmission line estimation unit 70,
It has a function of outputting the estimated transmission symbol sequence EX _nM .

In the transmission path estimation unit 70, the actual transmission path 20
Has a function of estimating it and giving the impulse response estimated value Eh _j of the transmission path to the Viterbi algorithm processing unit 60, the received signal reproducing means 71, the impulse response adaptive updating means 72, And subtraction means 73.

The received signal reproducing means 71 uses the following equation (1) from the estimated transmission symbol sequence {EX _n , EX _n-1 , ..., EX _nL } and the impulse response estimation value Eh _{j of the} transmission line 20.
An estimated value Er _n of the received signal is generated. This is subtracted by the subtraction means 73 from the received signal cr _n which has been compensated for the phase fluctuation by the following equation (2), and the error signal e _{n is} obtained.
To get e _n = cr _n -Er the _n · · · (2) the impulse response adaptive update means 72, by LMS (Least Mean Sukeyazu) algorithm represented by the following formula (3), the impulse response estimate Eh of the transmission line 20
Update _j (j = 0, 1, ..., L). ^{_{Eh j n + 1 = Eh j}} n + β · e n · EX * nj ··· (3) where, j = 0,1, ..., L *; positive constant phase estimator called step size; complex conjugate beta Reference numeral 80 estimates the amount of phase fluctuation due to frequency offset and phase jitter by the following equation (4), gives the phase estimation value Eφ _{n + 1} to the phase rotation unit 40, and the phase rotation unit 40
Works to compensate for the amount of phase fluctuation. _{Eφ n + 1 = Eφ n +} α · Im [e n · cr * n] ··· (4) where, Im []; the imaginary part of the complex alpha; positive constant phase estimator 80 and phase by the phase rotation section 40 The fluctuation compensation is equivalent to a first-order phase locked loop. Therefore, it can be said that the phase estimation unit 80 includes a phase error detection circuit 81 of a phase locked loop (PLL), a loop filter 82, and a voltage controlled oscillator (hereinafter, referred to as VCO) 83.

[0011] The phase error detecting circuit 81, the multiplication that multiplies a complex conjugate calculation unit 81a for obtaining the complex conjugate cr ^* _nM from the output cr _nM delay means 50, the complex conjugate cr ^* _nM and the estimated error e _nM Means 81b and imaginary part extracting means 81c for extracting the imaginary part from the multiplication result and outputting the phase error Δφ _n . The filter 82 is
Multiplier 82 for multiplying the phase error Δφ _n by a multiplier α
a. The VCO 83 is composed of an adding means 83a for adding the phase estimation value Eφ _n-1 at the previous time to the output of the multiplying means 82a, and a delay means 83b such as a register for delaying the output of the adding means 83a. ing.

[0012]

However, in the conventional adaptive maximum likelihood sequence estimator, the phase fluctuation amount is compensated by using the first-order phase-locked loop. Of the fluctuation amount, steady phase rotation due to frequency offset cannot be sufficiently compensated. Therefore, if the frequency offset amount increases,
Since the error rate increases, there is a problem in that the range of applicable frequency offset (shift of carrier wave of transceiver) is narrow, and thereby the decoding accuracy deteriorates, which is difficult to solve.

The present invention solves the problem that the above-mentioned prior art has, that the stationary phase rotation due to the frequency offset of the carrier cannot be sufficiently compensated, and the phase fluctuation amount can be compensated over a wide range of frequency offsets. An adaptive maximum likelihood sequence estimator is provided.

[0014]

In order to solve the above-mentioned problems, a first aspect of the present invention includes a phase rotating section for rotating the phase of a received signal based on a phase estimation value to compensate for a phase fluctuation of the received signal. The received signal after the compensation is input, the transmission symbol is estimated according to the Viterbi algorithm based on the impulse response estimation value of the transmission line, and the Viterbi algorithm processing unit that outputs the estimated transmission symbol sequence and the impulse response of the transmission line are output. A transmission path estimation unit for estimating and a phase estimation unit,
The adaptive maximum likelihood sequence estimator provided is provided with an integrating means and an adding means.

Here, the transmission path estimation unit has a reception signal reproduction means for calculating an estimation value of a reception signal from the estimated transmission symbol sequence and an impulse response estimation value of the transmission path,
An estimated error is calculated by subtracting the estimated value of the received signal from the compensated received signal, and the impulse response of the transmission line is updated according to an adaptive algorithm based on the estimated error and the estimated transmission symbol sequence, and the new transmission line is updated. It has a function of giving the impulse response estimated value of to the Viterbi algorithm processing section and the received signal reproducing means. The phase estimation unit detects a phase error by using the received signal after the compensation and the estimation error, adds the phase error as a phase correction value to the phase estimation value, and gives a new phase estimation value to the phase rotation unit. It is a thing.

The integrating means has a function of integrating the phase error, and the adding means has a function of adding the output of the integrating means to the phase error to generate the phase correction value. In a second aspect based on the first aspect, a normalizing means is provided for normalizing the phase error with the short-time power of the compensated received signal.

[0017]

According to the first aspect of the present invention, since the adaptive maximum likelihood sequence estimator is configured as described above, the integrating means and the adding means provided in the phase estimating section use a quadratic phase-locked loop to generate the frequency. It works to compensate for phase fluctuations due to offset.

According to the second aspect of the invention, the normalizing means functions to normalize the phase error extracted by the phase locked loop with the short-time average power of the received signal. Therefore, the above problem can be solved.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First Embodiment FIG. 1 is a functional block diagram of an adaptive maximum likelihood sequence estimator showing a first embodiment of the present invention, in which elements common to those in FIG. Is attached. This adaptive maximum likelihood sequence estimator is adapted to the adaptive equalizer 33 in the transmitter / receiver in the digital mobile communication shown in FIG. 2 as in the conventional case. The adaptive maximum likelihood sequence estimator is an individual circuit using an integrated circuit or a program by a processor. It is composed of controls and the like.

In this adaptive maximum likelihood sequence estimator, the conventional maximum likelihood sequence estimator shown in FIG.
Instead of the phase estimation unit 80 in the inside, a phase estimation unit 80A having a different configuration is provided. The phase estimation unit 80A is composed of a phase error detection circuit 81 and a VCO 83 similar to the conventional one, and a loop filter 82A different from the conventional one.

The phase error detection circuit 81, conventional manner, the estimated error e from the complex conjugate calculation unit 81a for obtaining the complex conjugate cr ^* _nM from the output cr _nM delay means 50, the complex conjugate cr ^* _nM subtraction means 73 It is composed of a multiplication means 81b for multiplying _nM and an imaginary part extraction means 81c for extracting an imaginary part from the output of the multiplication means 81b and outputting a phase error Δφ _n .

The loop filter 82A includes a multiplication means 82a which constitutes the conventional loop filter 82 and an integration means 90.
And addition means 94 are added. Integrating means 90
Is the phase error Δφ output from the imaginary part extraction means 81c.
_It has a function of integrating _n, and comprises a multiplication means 91 for multiplying the phase error Δφ _n by a multiplier α1, an addition means 92, and a delay means 93 such as a register for delaying the output of the addition means 92. There is. The adding means 94 has a function of adding the phase error Δφ _n and the output of the integrating means 90 and outputting the result to the multiplying means 82a. The multiplication means 82a multiplies the output of the addition means 94 by a multiplier α2 to obtain the phase correction amount ΔEφ _{n + 1} as VC.
It has a function of outputting to O83.

The VCO 83 has a phase correction amount Δ as in the conventional case.
Eφ _{n + 1} is input and the phase estimation value Eφ _n is input to the phase rotation unit 40.
It has a function of giving to, and is composed of an adding means 83a and a delay means 83b such as a register.

Next, the operation will be described. For example, the sample value y _n of the received signal is _expressed as y _n = r _n exp [jφ _n ] ... (5). Here, phi _n is the phase deviation due to frequency offset and phase jitter, using the sample values of the received signal when r _n is not have these phase fluctuations, the impulse response vector Bh _n, and a transmission symbol vector BX _n following It is expressed as in Expression (6). r _n = Bh _n ^T · Bx _n (6) Here, Bh _n = {h ₀ ⁿ , h ₁ ⁿ , ..., h _L ⁿ } ^T Bx _n = {x _n , x _n-1 , ..., x _nL } ^T T; Transposition of vector In the receiver 30 of FIG. 2, the sample value Y _n of the received signal sampled at the symbol interval T is the phase rotation unit 40.
Is input to the phase rotation unit 40, the phase estimation unit 8
The phase rotation amount exp [−jEφ _n ] is calculated by the calculating means 41 from the phase estimated value Eφ _n estimated at 0 A, and is multiplied by the sample value Y _n of the received signal by the multiplying means 42 to compensate the phase ( Sample value cr of received signal like 7)
_n is the delay means 50 and the Viterbi algorithm processing unit 6
Output to 0. _{cr n = r n · exp [} j (φ n -Eφ n)] ··· (7) the delay means 50 delays the sample value cr _n of the received signal,
The delayed value cr _{n- M is given} to the complex conjugate calculating means 81a in the phase error detecting circuit 81 and the subtracting means 73 in the transmission path estimating section 70.

The Viterbi algorithm processing unit 60 inputs the sample value cr _n of the received signal and estimates the transmission symbol sequence based on the impulse response estimation value Eh _j of the transmission path 20 estimated by the impulse response adaptive updating means 72. Then
The estimated transmission symbol sequence {Ex _n , Ex _n-1 , ..., E
x _nL } is output.

Received signal reproducing means 7 in the transmission path estimating section 70
1 is the estimated transmission symbol sequence {Ex _n , Ex _n-1 , ...,
Ex _nL } and impulse response estimation value Eh _{j of the} transmission line 20
Based on and, the estimated value Er _nM of the received signal is obtained and given to the subtraction means 73. Subtracting means 73 subtracts the estimated value Er _nM from the sample value cr _nM of the received signal from the delay means 50, the estimation error determined for e _nM, it multiplication means 81b and the impulse response adaptive update in the phase error detection circuit 81 And means 72.

In the phase error detection circuit 81, the delay means 50
In calculating the complex conjugate cr ^* _nM complex conjugate calculation unit 81a from the sample value cr _nM of the delayed received signals, multiplying the estimation error e _nM of the transmission line and its complex conjugate cr ^* _nM multiplication unit 81b. From this multiplication result, the imaginary part extracting means 81c extracts the imaginary part, obtains the phase error Δφ _n as in the following equation (8), and supplies it to the loop filter 82A. In _{△ φ n = φ n -Eφ n} ··· (8) integrating means 90 in the loop filter 82A, multiplies the phase error △ phi _n and multipliers α1 multiplication unit 91, the multiplication result of the delay means 93 The output is added by the addition means 92,
The addition result is delayed by the delay means 93 by the delay means 93.
Give feedback to. Thereby, the integration means 90
Then, the phase error Δφ _n is integrated to obtain its average value, and the integration result and the phase error Δφ _n are added by the adding means 94. The output of the adding means 94 is the multiplier α in the multiplying means 82a.
It is multiplied by 2 to obtain the phase correction amount ΔEφ _{n + 1} and VC
It is sent to O83.

In the VCO 83, the adding means 83a adds the phase correction amount ΔEφ _{n + 1} to the estimated phase value Eφ _n , and the delaying means 83b delays it to the phase rotating section 40 as a new estimated phase value Eφ _{n + 1} . Output.

As described above, the first embodiment has the following advantages. In the compensation of the amount of phase fluctuation by the phase estimation unit 80A and the phase rotation unit 40, the average component of the phase error calculated by the integration means 90 is converted into the instantaneous phase error Δ.
A fixed increment is added to the phase estimate by adding to φ _n . Therefore, the stationary phase rotation due to the frequency offset can be compensated, and it is equivalent to the secondary phase locked loop. In this way, since the phase variation due to the frequency offset and the phase jitter is compensated by using the secondary phase locked loop, it is possible to compensate the steady phase rotation due to the frequency offset. Therefore, the transmission symbol can be estimated with a small error rate over a wide range of frequency offsets.

As an example of this effect, FIG. 4 shows a simulation result of frequency offset versus bit error rate in the case of a frequency selective fading transmission line of a two-wave model.
Waveform 201 in FIG. 4 shows the characteristic of this embodiment, waveform 202 shows the conventional characteristic, and waveform 203 shows the characteristic when phase fluctuation is not compensated.

The simulation conditions in FIG. 4 are as follows: symbol interval T41 μsec, maximum Doppler frequency 80 Hz,
The delay of the delay wave is 1.0T and Eb / N0 (signal power to noise power density ratio per bit) is 20 dB. As is clear from FIG. 4, the range of the frequency offset at which the bit error rate of 10 ⁻² or less is obtained is about ± 150 Hz in the conventional adaptive maximum likelihood sequence estimator, but is ± 1 k in the present embodiment.
The error rate does not deteriorate at all up to the frequency offset in the range of Hz as compared with the case without the frequency offset. Therefore, the effect of this embodiment is very large.

Second Embodiment FIG. 5 is a functional block diagram of an adaptive maximum likelihood sequence estimator showing a second embodiment of the present invention. Elements common to those in FIG. It is attached. In this adaptive maximum likelihood sequence estimator, a phase estimator 80A-1 having a different configuration is provided instead of the phase estimator 80A of FIG. Phase estimation unit 80
A-1 is a phase error detection circuit 81A different from FIG. 1, and a loop filter 82A and a VCO 83 similar to FIG.
It is configured. The phase error detection circuit 81A includes a complex conjugate calculation means 81 which constitutes the phase error detection circuit 81 of FIG.
a, multiplication means 81b, and imaginary part extraction means 81c,
A normalizing means 100 based on the short-time average power of the received signal is added.

The normalizing means 100 calculates the complex conjugate cr ^* _nM.
Means 10 for multiplying the output cr _nM of the delay means 50 by
1, an averaging means 102 for averaging the multiplication results, and a dividing means 103 for dividing the output of the imaginary part extracting means 81c by the output of the averaging means 102.

[0034] Considering the output of the phase error detection circuit 81 in FIG. 1, (2) from the ^{_{equation, Im [* e n · cr}} n] = Im [(cr n -Er n) cr n *] = Im [Er _n ^* · cr _n ] ... (9) Here, assuming that the estimation operations of the transmission path estimation unit 70 and the phase estimation unit 80A are sufficiently performed, (9)
After all, the formula is Im [e _n · cr _n ^* ] = A · H ² · sin (Δφ _n ) ≈A · H ² · Δφ _n (10) However, A is the root mean square value of the transmission symbol vector and is a value determined by the modulation method. H is the absolute value of the impulse response vector of the transmission line 20,
For example, in a frequency selective fading transmission line in land mobile communication, it is an amount that changes with time. Therefore,
In the phase error extraction by the equation (10), when the multiplier α in the equation (4) is fixed, the variation of the frequency selective facing transmission line unrelated to the phase error is included as the correction amount of the estimated phase. On the other hand, the power of the received signal to compensate for phase (7) and ^{_{(6), | cr n | 2 = cr}} n · cr n * = r n · r n * = A · H 2 ··· (11 ), The phase error Δφ can be obtained by dividing equation (10) by equation (11).
Only _n can be extracted. However, in the case of frequency-selective fading, the level of the received signal may become very small instantaneously, and at this time, the division by equation (11) cannot be calculated. Instead, the normalization means of FIG. 100
Short-time average power A [cr _n
· Cr _n ^*] (where, A [·] represents an average) with, by the following equation (12) can be extracted by the phase error.

[0035]

[Equation 1]

Therefore, an effect similar to that of the first embodiment can be obtained.

The present invention is not limited to the above embodiment,
Various modifications are possible. Examples of such modifications include the following. (A) In FIGS. 1 and 5, the position at which the multipliers α1 and α2 of the loop filter 82A are multiplied is not limited to the position shown in the figure. , Α2 may be multiplied. For example, the multiplication means 82a is provided on the output side of the imaginary part extraction means 81c,
For example, the output of the imaginary part extraction means 81c is multiplied by a multiplier α2.

(B) As a configuration of the adaptive maximum likelihood sequence estimator,
A matched filter or a whitened matched filter for reducing the signal-to-noise ratio (S / N ratio) is provided on the input side of the Viterbi algorithm processing unit 60 of FIG. 1 or 5, and the output of the filter is subjected to the Viterbi algorithm processing. There is a configuration example of inputting to the unit 60. The adaptive maximum likelihood sequence estimator having these configurations is different only in the calculation in the Viterbi algorithm processing unit 60, and the above embodiment can be applied to the adaptive maximum likelihood sequence estimator having these configurations in exactly the same manner.

(C) The adaptive maximum likelihood sequence estimator of the above embodiment has been described as applied to the adaptive equalizer 33 of digital mobile communication, but as an adaptive equalizer of data communication in a fixed communication network, etc. Is, of course, also applicable.

[0040]

As described above in detail, according to the first aspect of the present invention, the integrating means and the adding means are provided to compensate for the phase fluctuation due to the frequency offset and the phase jitter by using the secondary phase locked loop. Since this is done, it is possible to compensate for the stationary phase rotation due to the frequency offset. Therefore, it is possible to accurately estimate a transmission symbol with a small error rate over a wide range of frequency offsets.

According to the second invention, the standardizing means is provided,
Since the phase error extracted by the phase-locked loop is standardized by the short-time average power of the received signal, the transmission symbol can be accurately estimated in the frequency-selective fading transmission line or the like, almost like the first invention. It can be carried out.

[Brief description of drawings]

FIG. 1 is a functional block diagram of an adaptive maximum likelihood sequence estimator showing a first embodiment of the present invention.

FIG. 2 is a functional block diagram of a transceiver in general digital mobile communication.

FIG. 3 is a functional block diagram of a conventional adaptive maximum likelihood sequence estimator.

FIG. 4 is a diagram showing simulation results of frequency offset vs. bit error rate of the conventional example and the present example.

FIG. 5 is a functional block diagram of an adaptive maximum likelihood sequence estimator showing a second embodiment of the present invention.

[Explanation of symbols]

40 phase rotation part 50 delay means 60 Viterbi algorithm processing part 70 transmission path estimation part 71 received signal reproduction means 72 impulse response adaptive updating means 73 subtraction means 80A, 80A-1 phase estimation part 81, 81A phase error detection circuit 82A loop filter 83 VCO 90 integrating means 100 normalizing means

Claims (2)

[Claims] 1. A phase rotation unit for rotating a phase of a received signal based on the phase estimated value to compensate for a phase fluctuation of the received signal, and a received signal after the compensation, which is input to an impulse response estimated value of a transmission line. Based on the Viterbi algorithm, the transmission symbol is estimated, and the estimated value of the received signal is calculated from the Viterbi algorithm processing unit that outputs the estimated transmission symbol sequence and the estimated transmission symbol sequence and the impulse response estimation value of the transmission path. An impulse response of the transmission line according to an adaptive algorithm based on the estimated error and the estimated transmission symbol sequence, the estimated signal error is calculated by subtracting the estimated value of the received signal from the compensated received signal. To provide a new impulse response estimation value of the transmission line to the Viterbi algorithm processing unit and the received signal reproducing means. A transmission path estimation unit, a phase error is detected using the received signal after the compensation and the estimation error, the phase error is added to the phase estimation value as a phase correction value, and a new phase estimation value is added to the phase rotation unit. An adaptive maximum likelihood sequence estimator provided with a phase estimator for giving an integrating means for integrating the phase error, and an adding means for adding the output of the integrating means to the phase error to generate the phase correction value. , An adaptive maximum likelihood sequence estimator provided. 2. The adaptive maximum likelihood sequence estimator according to claim 1, further comprising a normalizing means for normalizing the phase error by a short time power of the received signal after the compensation. Likelihood sequence estimator.