JP6600187B2 - Communication device - Google Patents

Description

  The present invention relates to a communication apparatus and a frequency offset correction method, and more particularly, to a communication apparatus and a frequency offset applied to a communication system that switches and transmits a frequency offset training signal and an adaptive equalizer training signal on the transmission side. This relates to the correction method.

  In digital power transfer using a transmission line as a transmission line, a frequency offset between the carrier wave of the transmission device and the demodulation carrier wave of the reception device due to variations in accuracy of the crystal oscillators mounted on the transmission device and the reception device. This is a factor that deteriorates characteristics such as bit error rate. The frequency offset refers to a phenomenon in which the frequency of a crystal oscillator mounted for transmission / reception causes a frequency error of about ± 30 ppm at maximum in transmission / reception due to variations in manufacturing accuracy. As a result, phase rotation occurs at the output of the demodulator, and the adaptive equalizer cannot perform normal symbol determination, so that transmission quality deteriorates.

For this reason, the structure which correct | amends the frequency offset which arises between transmission / reception is provided.
For example, Patent Document 1 discloses an equalizer for increasing the convergence speed with a relatively small circuit scale. This equalizer is obtained by calculating the absolute value of each of the reference symbols di (n) and dq (n) and calculating the absolute value of each of the reference symbols di (n) and dq (n). A weighting table in which a calculation process for obtaining a weight W (n) obtained by calculation of a least square error algorithm with respect to a reference symbol point, a distance between the reference symbol point and the origin is tabulated in advance for each modulation scheme; A weighting multiplier that multiplies W (n) and the output of the adder to obtain weighted equalization errors Ei (n) and Eq (n) is between the adder and the first complex multiplier. Inserted and weighted equalization errors Ei (n), Eq (n) and input signals ui (n), uq (n) are respectively subjected to complex multiplications to correspond to the I-phase and Q-phase respectively. Configured to obtain a multiplication output signal .

  Patent Document 2 discloses a digital radio receiver using a carrier offset compensation circuit. In this digital radio receiver, an initial phase correction unit determines a phase shift from a known signal, corrects an initial phase, multiplies a tap coefficient of one symbol before in the first equalizer by a multiplication unit, and an equalizer. The waveform equalization is performed while updating the tap coefficient based on the region determination result. With this configuration, carrier offset compensation is possible even in TDD communication, and the range of offset frequencies that can be handled is widened to enable stable operation.

JP 2006-279470 A Japanese Patent Laid-Open No. 2004-007487

  In a digital power carrier, in a frequency offset correction circuit that estimates and corrects a frequency offset that occurs between transmission and reception, when the local crystal oscillator receives a transient temperature fluctuation, a transient fluctuation in the oscillation frequency of the local crystal oscillator In some cases, an error may occur from the estimated and corrected frequency offset without being able to follow other frequency fluctuation factors. This error is, for example, a maximum value of about 3 ppm from the estimated frequency offset value. If this becomes a residual phase error and the value of the feedback coefficient μ is small when the adaptive equalizer is in the tracking mode, the adaptive equalizer cannot correct this residual frequency offset. In this case, for example, as shown in FIG. 5, a phase gradient occurs in the output of the adaptive equalizer, and the error rate characteristic is deteriorated.

  Therefore, after the offset frequency is estimated and corrected by the frequency offset correction circuit, it is necessary to estimate and correct the residual phase error when a residual phase error occurs due to a transient temperature fluctuation or the like. In the configurations of Patent Documents 1 and 2, an equalizer for correcting the residual phase error as described above is set, and the tap coefficient of the direct equalizer is updated by an LMS (Least Mean Square) algorithm circuit. Since this is a method to be performed, another existing equalization circuit is required for the existing adaptive equalizer, and the circuit configuration is greatly complicated.

  Also, in order to follow the temperature change of the local crystal oscillator, the crystal oscillator may be changed to a temperature compensated crystal oscillator. In order to cope with this, it becomes necessary to design and manufacture a printed circuit board corresponding to a new mounting arrangement, and a new development cost is generated.

  The present invention has been made in view of the above circumstances, and uses an existing adaptive equalizer and a frequency offset correction circuit, without providing a new phase error detection circuit, and a temperature compensated crystal oscillator An object of the present invention is to provide a communication device and a frequency offset correction method for estimating and correcting a residual phase error caused by the temperature frequency characteristics of a local crystal oscillator without changing to.

A communication apparatus according to the present invention includes a frequency offset estimation / correction unit that calculates a frequency offset phase correction amount based on a complex autocorrelation of a symbol pattern transmitted from a transmission-side communication apparatus, and a transmission apparatus that transmits the frequency offset. An adaptive equalizer that compensates for degradation of transmission quality due to intersymbol interference of the received signal, and the adaptive equalizer equalizes the transmission symbol point estimated by equalization processing using a transversal filter a filter unit that outputs an output value, anda determination unit reference symbol point on the basis of the equalized output value output from the filter unit, the equalized output value output from the filter section correction Oh that calculates the complex cross-correlation between the output value indicating the reference symbol point output from the decision unit in accordance with the equalized output values, were calculated based on the complex cross-correlation the calculated A set amount, the by reflecting the frequency offset phase correction amount Frequency offset estimation and correction unit is calculated, the frequency offset caused by the temperature frequency characteristics of the local crystal oscillator could not be compensated by the frequency offset estimation and correction unit It is a communication device to be corrected.

In the communication apparatus according to claim 1, the frequency offset estimation / correction unit calculates a complex autocorrelation between two symbols of the symbol pattern transmitted from the transmission-side communication apparatus. the averaged phase rotation amount calculated from the complex autocorrelation and the calculated, based on the average phase rotation amount calculating the frequency offset phase correction amount, and the equalized output value output from the filter unit, It calculates the complex cross-correlation between the output value output from the determination unit, the residual phase error calculation unit for calculating a rotation angle of the frequency offset caused by the temperature frequency characteristics of the local crystal oscillator based on the complex-cross-correlation a, the correction offset which is calculated based on the rotation angle of the frequency offset caused by the temperature frequency characteristics of the local crystal oscillator in which the residual phase error calculator has calculated , Said that the frequency offset estimation and correction unit is added to the frequency offset phase correction amount calculated, is a communication device to reflect the complex cross-correlation in the frequency offset phase correction amount.

  Further, in the communication device of the present invention, the residual phase error calculation unit 4 includes an equalized output value V (n) output from the filter unit and an output value d output from the determination unit as shown in FIG. A calculation unit 11 for calculating a complex cross-correlation value e (n) with (n), an averaging filter unit 12 for averaging the calculated values, and calculating an average from the averaged complex cross-correlation value er (n) The communication apparatus includes a phase rotation angle calculation unit 13 that calculates a residual phase error angle Δφer (n) and corrects the frequency offset of the signal.

  The frequency offset correction method of the present invention includes a frequency offset estimation / correction step for calculating a phase correction amount based on a complex autocorrelation of a symbol pattern transmitted from a transmission side communication device, and a transmission side communication device. An adaptive equalization step for compensating for transmission quality degradation of the transmitted signal, and the adaptive equalization step outputs an equalized output value indicating a transmission symbol point estimated by equalization processing using a tap coefficient. And a determination step for determining a reference symbol point based on the equalization output value output from the filter unit, and further, an equalization output value output at the filter step and an output at the determination step The complex cross-correlation with the output value is calculated, and the calculated complex cross-correlation is calculated in the frequency offset estimation / correction step. It is reflected in the amount, a correction method of frequency offset with a step of correcting a frequency offset of a signal transmitted from the communication device.

The present inventor has studied to solve the above-described problem. As a result, the output of the estimated transmission symbol V (n) output from the filter unit of the adaptive equalizer and its value are determined by the determiner as a reference symbol. The output determined at d (n) is branched, the complex cross-correlation between V (n) and d (n) is calculated, and the calculated values are averaged, so that no new phase error detection circuit is provided. It was found that the residual phase error vector er (n) = E [R + jI] can be calculated. E [.] Represents averaging.
Further, since the component of the residual phase error amount itself has a small value, the Atan (Arctangent) value due to the real part and imaginary part of er (n) is approximately 0.1 rad or less, and therefore the residual phase error angle Δφer ( n) ≈I can be approximated, and the Atan circuit can be omitted. Further, in the obtained residual phase error angle Δφer, the estimation error component and the noise component in the adaptive equalizer are removed by the averaging process, and only the average residual phase error angle component is extracted. This value is fed back and reflected to the frequency offset estimation / correction unit, and the residual phase error can be corrected by adding or subtracting from the frequency offset correction value estimated by the frequency offset estimation / correction unit. This has been found and the present invention has been achieved.

  According to the present invention, an existing adaptive equalizer and a frequency offset correction circuit are used to provide a local crystal oscillator without providing a new phase error detection circuit and without changing to a temperature compensated crystal oscillator. It is possible to provide a communication device and a frequency offset correction method for estimating and correcting a residual phase error caused by the temperature frequency characteristic of the.

It is a figure which shows the circuit structure in one Embodiment of the communication apparatus by this invention. It is a figure which shows the structure of the internal circuit of a residual phase error calculation part. FIG. 5 is a block diagram for explaining a circuit configuration of a frequency offset estimation / correction unit. It is a figure which shows the structural example of the PN code completion | finish determination circuit for offset estimation shown in FIG. It is a figure which shows an example of the constellation of the adaptive equalizer output in embodiment of this invention. It is a figure which shows an example of the constellation of the conventional adaptive equalizer output.

  FIG. 1 is a diagram showing a circuit configuration in an embodiment of a communication apparatus according to the present invention. A communication system to which a communication apparatus according to the present invention is applied includes a transmission-side communication apparatus that transmits a training pattern for compensating for an offset between communication apparatuses during training when starting communication, and a frequency received by receiving the training pattern. It is a communication system in which a receiving-side communication device that performs offset compensation is connected via a wired transmission path.

The communication device shown in FIG. 1 receives a symbol pattern based on a PN code transmitted from a communication device on the transmission side, and demodulates it by a demodulator (MODEM unit). The symbol pattern from the demodulator is input to the offset frequency estimation / correction unit 5, the frequency offset correction value is calculated, multiplied by the demodulated symbol pattern, and then the adaptive equalizer 1 for compensating for the deterioration of transmission quality. Is input.
The frequency offset estimation / correction unit 5 estimates and determines the frequency offset correction value based on the frequency offset estimation symbol pattern transmitted from the transmission side. When a transient temperature change occurs after the correction value is determined, a residual phase error of about 3 ppm at the maximum is generated from the determined frequency offset due to the temperature frequency characteristic of the local crystal oscillator. This residual phase error is corrected based on the output of the adaptive equalizer 1. Thus, the residual phase error can be corrected without providing a new phase error detection circuit.

The symbol pattern input to the filter unit 2 of the adaptive equalizer 1 is equalized using the tap coefficient, the transmission symbol point (I, Q) is estimated, and the equalized output value (V (n) ) Is output. The equalized output value (V (n)) output from the filter unit 2 is input to the determiner 3.
The determiner 3 outputs an output value (d (n)) in which the equalized output value is determined as the reference symbol point (I, Q). In the adaptive equalizer 1, for example, the output value (d (n)) from the determiner 3 is used as a reference signal, the equalized output value (V (n)) from the filter unit 2, and the output from the determiner 3. An error from the value (d (n)) is calculated, and the tap coefficient of the filter unit 2 is updated based on the calculated error. Further, the output value from the determiner 3 is output to the decoder unit and subjected to decoding processing.

In the embodiment according to the present invention, the equalized output value (V (n)) of the estimated transmission symbol output from the filter unit 2 of the adaptive equalizer 1 described above and the value thereof are determined by the determiner 3 as a reference symbol. The output value (d (n)) determined in step (b) is branched, and the complex cross-correlation between V (n) and d (n) is calculated, so that a new phase error detection circuit is not provided, and The residual phase error vector e (n) = R + jI resulting from the temperature frequency characteristics of the local crystal oscillator is calculated without changing to the temperature compensated crystal oscillator.
The residual phase error calculation unit 4 shown in FIG. 2 uses the equalized output value (V (n)) from the filter unit 2 of the adaptive equalizer 1 and the output value (d (n)) from the determiner 3. In order to calculate the phase error Δe (n) due to the cross-correlation between V (n) and d (n) ^* , the following calculation is performed. The above ^* indicates a conjugate complex number.

e (n) = V (n) · d (n) ^*
= R (n) + jI (n)

  The obtained e (n) is a forgetting factor α by an average filter, and a vector er (n) of an average residual phase error is calculated as follows.

er (n) = αer (n−1) + (1−α) e (n)
As a result of this averaging, the Gaussian distribution becomes zero on average, so that the estimation error component and noise component in the adaptive equalizer are removed. Here, the forgetting factor α uses a value of about 0.999998.

  The Atan (Arctangent) value by er (n) can be regarded as almost the same as the imaginary part I (n) when the imaginary part I (n) is smaller than 0.1. If the angle is very small, the Atan circuit can be omitted, and the average residual frequency offset phase rotation angle Δφer (n) ≈I (n) can be obtained.

Phase rotation angle Δφer the calculated average residual frequency offset (n) is input to the frequency offset estimation and correction unit 5, the estimated frequency offset correction value in the frequency offset estimation and correction unit 5 (Δφ (k)) Is added or subtracted to correct the frequency offset amount of the signal output from the MODEM unit.

(Configuration of frequency offset estimation / correction unit)
FIG. 3 is a block diagram for explaining a circuit configuration of the frequency offset estimation / correction unit 5. Frequency offset estimation training pattern by 2 ⁿ code stream transmitted from the transmitting-side communication device is demodulated by the demodulator 22 operating as 4PSK demodulator symbol pattern by 2 ⁿ code sequence, the frequency offset estimation and correction unit Enter 5.

The complex autocorrelation calculation circuit (complex autocorrelation calculation unit) 24 of the frequency offset estimation / correction unit 5 uses the current symbol point r (n) and the (2 ⁿ / 2) symbol delay circuit 28 to delay 2 ⁿ / 2 symbols. R (n−2 ⁿ / 2) is sampled from the symbol pattern, and the autocorrelation between the two symbols is obtained. Here, since the same 4PSK symbol point is generated from the transmission side in a 2 ⁿ / 2 cycle by the PN code, r (n) and r (n−2 ⁿ / 2) are autocorrelations r ( n) · r ^* (n−2 ⁿ / 2) always indicates a value of 1 when there is no frequency offset because the phase rotation is 0 °.

Here, ^{assuming that} the transmission channel time-varying characteristics between r (n) and r (n−2 ⁿ / 2) do not change, the convolution amount due to the delay wave in the transmission line is the same, and the correlation of the delay wave Can also be 1. As a result, it is possible to remove the influence of the delayed wave, and the influence of noise is also removed by random averaging processing, and an autocorrelation value indicating only the frequency offset element is obtained.

  The 45 ° phase rotation circuit (phase rotation unit) 25 rotates the complex autocorrelation value output from the complex autocorrelation calculation circuit 24 by 45 °. As a result, the coordinates of the complex autocorrelation value are converted into the symbol arrangement position on the complex plane. Here, by rotating the 45 ° phase, the symbol point is in four quadrants ([++], [+ −], [−−], [− +]) of the complex plane composed of the real part and the imaginary part. It will always be placed in either.

In the real part / imaginary part polarity detection circuit (real part / imaginary part polarity detection part) 29, the symbol point rotated by 45 ° is four quadrants ([++] (the real part and the imaginary part are [+]). ), [+-] (Real part is [+], imaginary part is [-]), [-] real part and imaginary part is [-]), [-+] (real part is [-], imaginary part It is detected whether the part is in [+])).
The transmission symbol deviation estimation circuit (transmission symbol deviation estimation unit) 30 calculates the phase deviation (phase rotation amount) according to the polarities (signs) of the real part and the imaginary part detected by the real part / imaginary part polarity detection circuit 29. presume. Here, when the real part and the imaginary part detected by the real part / imaginary part polarity detection circuit 29 are [++], the transmitted current symbol s (n) and the symbol s (n− ²ⁿ / 2) before are transmitted. ²ⁿ / 2) means that the same symbol has been transmitted, and the transmission symbol shift estimation circuit 30 can estimate the phase rotation amount to be 0 °.
Similarly, when the real part and the imaginary part are [− +], the phase rotation amount can be estimated to be 90 °, and when the real part and the imaginary part are [−−], the phase rotation amount is 180 °. When the real part and the imaginary part are [+ −], the phase rotation amount can be estimated to be 270 °.

The transmission symbol shift estimation circuit 30 outputs one of the values 1, j, −1, and −j according to the estimated amount of phase rotation. Here, when the symbol point detected by the real part / imaginary part polarity detection circuit 29 is in the [++] quadrant, that is, when the phase rotation is estimated to be 0 °, 1 is output.
When the symbol point is in the [− +] quadrant, that is, when the phase rotation is estimated to be 90 °, j is output, and when the symbol point is in the [−−] quadrant, that is, the phase rotation is 180 °. Is output when the symbol point is in the [+-] quadrant, that is, when the phase rotation is estimated at 270 °.

Since the autocorrelation of the transmitted symbol is represented by s (n) · s ^* (n−2 ⁿ / 2), 2 ⁿ / 2 symbols that have been subjected to 4PSK modulation are used as a frequency offset estimation training pattern by this method. In the case of transmission, the complex autocorrelation value rotated by 45 ° is always [++], and a value of 1 is obtained as the complex autocorrelation value. This theory will be described more specifically below.

The autocorrelation value calculated by the complex autocorrelation calculation circuit 24 is calculated in the presence of a frequency offset. When there is a frequency offset, for example, in order to determine that the autocorrelation value at intervals of 64 symbols (n = 7 in 2 ⁿ / 2 symbols) is always in the “++” quadrant, the phase is rotated by 45 °. The frequency offset must fall within the range of ± 45 °.

Here, since the error of the crystal oscillator is specified to be ± 30 ppm or less, no further error is shown. Since the maximum frequency used for communication in the communication system is 425 kHz, the frequency offset amount when there is an error of ± 30 ppm is about ± 13 Hz, and the rotation amount at this time is a symbol rate of 64 QAM of 32 kbps (31.25 μs). ) Then:
Rotation amount ≒ 360 x freq.offset x Ts
≒ 360 × 13Hz × 31.25 × 10-6
≒ ± 0.146 °

Thus, between 64 symbols, ± 0.146 × 64≈ ± 9.36 °, which does not exceed ± 45 °. Therefore, when the 2 ⁿ code string at the time of training is received, when the autocorrelation value is rotated by 45 °, it always enters the [++] quadrant of the complex plane, and the transmission symbol shift estimation circuit 30 Then, the phase rotation is estimated to be 0 °, and 1 is always output. That is, when the 2 ⁿ / 2 symbols 4PSK modulated as a training pattern for frequency offset estimation is transmitted, it is possible to always output 1 from the transmission symbol shift estimation circuit 30.
If there is a concern that the phase rotation will exceed ± 45 °, the transmission symbol shift estimation circuit 30 can ensure that the sampling symbol interval is as short as 32 symbols or 16 symbols. 1 can be output.

  The phase rotation amount calculation circuit (phase rotation amount calculation unit) 26 calculates the phase rotation amount due to the frequency offset by the following equation (1).

In the above equation (1), the denominator is the amount of phase rotation in the symbol interval (2 ⁿ / 2) estimated as transmitted on the transmission side, and the output output from the transmission symbol shift estimation circuit 30 Use the value. The numerator of the above formula (1) is the amount of phase rotation between symbols actually received via the transmission path.
Here, since the 2 ⁿ / 2 symbol pattern is transmitted as the frequency offset estimation training pattern as described above, 1 is output from the transmission symbol shift estimation circuit 30, so that the equation (1) The denominator is always 1, and the amount of phase rotation between 2 ⁿ / 2 symbols received via the transmission path is calculated as Δr. This phase rotation amount is an averaged phase rotation amount due to a frequency offset from which the cause of the delay wave and noise is removed.

The forgetting factor averaging circuit 27 uses the following equation (2) to average Δr (n) for each symbol.
R (n) = βR (n-1) + (1-β) ΔR (n) (2)
Here, β is a forgetting factor having a value of about 0.998.

In the ATA calculation circuit 32, the Atan (Arctangent) calculation is performed by the real part and the imaginary part of R (n) obtained by the forgetting factor averaging circuit 27, and the rotation angle of one symbol is obtained by dividing by 2 ⁿ / 2. Δθ (n) is obtained.

  The second complex multiplier 33 includes Δφ (n−1) delayed by the one symbol delay circuit 34, the average phase rotation angle Δθ (n) of one symbol, and the phase rotation angle Δφer (n of the average residual frequency offset. ) To calculate the phase correction amount Δφ (n) of the nth symbol of the demodulator, the process shown in the following equation (3) is performed.

e ^−j Δφ ⁽ⁿ⁾ = e ^{−j (} Δφ ^{(n−1) +} Δθ ^{(n) +} Δ ^{er (n))} (3)

The estimated phase correction amount Δφ (n) of the nth symbol is multiplied by the signal from the demodulator 22 by the first complex multiplier 35.
Thus, the frequency offset can be corrected by the frequency offset amount in which the residual phase error is corrected.

FIG. 4 is a diagram illustrating a configuration example of the offset estimation PN code end determination circuit 31 illustrated in FIG. 3. In the communication system using the communication apparatus of the present embodiment, the frequency offset training signal is transmitted from the transmission side, and then the code switched to the adaptive equalizer training signal is transmitted. After the training signal is transmitted, communication is started and a transmission signal based on random data is transmitted. For the frequency offset training signal, a code obtained by 4PSK-modulating 2 ⁷ code strings with 1 bit added to PN ⁷ (2 ⁷ -1) is used. The adaptive equalizer training signal uses a code obtained by 4PSK-modulating a code sequence of PN12 (2 ¹² -1).

  In the offset estimation PN code end determination circuit 31, the code conversion circuit 311 converts the output value (+1, j, −1, −j) from the transmission symbol shift estimation circuit 30 to 1 when the correlation is +1. When the inverse correlation is (−1), the value is converted to −1, the other values are converted to 0, and n symbols are input to the shift register 312. The shift register 312 inputs the tap output value to the adder 313. The adder 313 receives a value of 1, −1, or 0 corresponding to the number of taps T + 1 of the shift register 312, and the output of the adder 313 is the sum of the input ones. The maximum output value is T + 1. Therefore, in this case, since all the n symbols input to the shift register are 1 in the time domain in which the frequency offset estimation training pattern is transmitted, the output value of the adder 313 always indicates the n value. .

However, when the transmission apparatus on the transmission side shifts from the frequency offset estimation training pattern to the adaptive equalizer training pattern, the 2 ^m -1 code sequence does not always have a correlation of 1 at 2 ⁿ / 2 symbol intervals. As for the output from the transmission symbol deviation estimation circuit 30, the values +1, j, -1, and -j are output at random. Alternatively, a transmission symbol shift estimation circuit is set by setting a code that is opposite in sign to the first to m-th bit strings generated by the PN generator as the initial setting value of the shift register (m stage) of the PN generator. As for the output from 30, m / 2 symbols are continuously output as -1. In this case, +1, −1, and 0 are input to the shift register 312 randomly or continuously, and the output value from the adder 313 is not an n value.

  In the threshold determination circuit 314, the threshold value of (na) is set to the output value of the adder 313, so that when the value becomes (na) or less, the adaptive equalizer training pattern has been switched. Can be detected. When it is detected that the frequency offset estimation training pattern is switched to the adaptive equalizer training pattern, the threshold determination circuit 314 outputs a PN code switching detection signal. As a result, even if the frequency offset is compensated based on the training pattern transmitted from the transmitting side without generating the training pattern on the receiving side, the training pattern is switched from the transmitted training pattern. It will be possible to easily determine.

When the PN code switching detection signal from the threshold determination circuit 314 is input to the signal output control circuit 36 shown in FIG. 3, the signal output control circuit 36 releases the signal output suppression to the adaptive equalizer so far. The signal is output to the adaptive equalizer. As a result, an adaptive equalizer training pattern is output to the adaptive equalizer, and training in the adaptive equalizer is started.
In addition, the PN code switching detection signal from the threshold value determination circuit 314 in the offset estimation PN code end determination circuit 31 is also output to the ATRAN calculation circuit 32 at the same time, ends the averaging process after the elapse of time t symbols, and rotates one symbol. The angle is fixed at Δθ (t−M + 1) to the previous value. M is the number of registers of the shift register 312.
After completion of the above training, the demodulator is switched from the 4PSK demodulator to 2 ⁿ QAM demodulation, and the data signal sequence is demodulated.

  FIG. 5 is a diagram illustrating an example of the output of the adaptive equalizer in the embodiment. With the above configuration, a very simple calculation circuit is added to the existing adaptive equalizer and the frequency offset estimation / correction unit without adding a new adaptive equalizer or LMS algorithm for correcting the residual frequency offset. Thus, the output of the adaptive equalizer as shown in FIG. 5 is obtained. Here, as compared with the applied equalizer output in which the residual phase error is not corrected as shown in FIG. 6, the residual phase error that cannot be corrected by the adaptive equalizer can be estimated and corrected.

DESCRIPTION OF SYMBOLS 1 ... Adaptive equalizer, 2 ... Filter part, 3 ... Determinator, 4 ... Residual phase error calculation part, 5 ... Frequency offset estimation / correction | amendment part, 11 ... Complex cross correlation calculation part, 12 ... Forgetting factor average filter part , 13 ... phase rotation angle calculation unit, 21 ... adder, 22 ... demodulator, 24 ... complex autocorrelation calculation circuit, 25 ... 45 ° phase rotation circuit (phase rotation unit), 26 ... phase rotation amount calculation circuit (phase rotation) 27) forgetting coefficient averaging circuit, 28 ... symbol delay circuit, 29 ... imaginary part polarity detection circuit, 30 ... transmission symbol shift estimation circuit, 31 ... PN code end determination circuit for offset estimation, 32 ... ATAN Calculation circuit 33... Second complex multiplier 34... 1 symbol delay circuit 35... First complex multiplier 36... Signal output control circuit 311. ... threshold Judgment circuit.

Claims (2)

A frequency offset estimation / correction unit that calculates a frequency offset phase correction amount based on a complex autocorrelation of a symbol pattern transmitted from a transmission-side communication device, and intersymbol interference of a signal transmitted from the transmission-side communication device. An adaptive equalizer that compensates for degradation of transmission quality,
Said adaptive equalizer, reference based on the filter section for outputting an equalized output values indicating the transmission symbol point estimated by equalization processing using the transversal filter, the equalized output value output from the filter section A determiner for determining a symbol point;
Calculating said equalized output value output from the filter unit, the complex cross-correlation between the output value indicating the reference symbol point output from the decision unit in accordance with the equalized output values, the complex of the calculated The correction offset amount calculated based on the cross-correlation is reflected in the frequency offset phase correction amount calculated by the frequency offset estimation / correction unit, and the temperature of the local crystal oscillator that could not be compensated by the frequency offset estimation / correction unit A communication apparatus that corrects a frequency offset caused by frequency characteristics. 2. The communication device according to claim 1, wherein the frequency offset estimation / correction unit calculates a complex autocorrelation between two symbols of a symbol pattern transmitted from the transmission-side communication device, and calculates the complex autocorrelation from the calculated complex autocorrelation. An averaged phase rotation amount is calculated, and the frequency offset phase correction amount is calculated based on the averaged phase rotation amount,
Said equalized output value output from the filter unit, calculates the complex cross-correlation between the output value output from the determination unit, due to the temperature-frequency characteristic of the local crystal oscillator based on the complex-cross-correlation A residual phase error calculation unit for calculating the rotation angle of the frequency offset;
The frequency offset phase correction amount of the compensation offset amount calculated based on the rotation angle of the frequency offset caused by the temperature frequency characteristic, the frequency offset estimation and correction unit has calculated the local crystal oscillator in which the residual phase error calculator has calculated To add the complex cross-correlation to the frequency offset phase correction amount.

Images