JPH0714169B2 - Frequency offset initialization method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Outline] In a line in which stations are connected by microwaves or the like, a slight difference between the oscillation frequencies of local oscillators of both stations appears as an offset of the carrier frequency of a transmission signal, so that data transmission is possible. The frequency offset amount of the carrier is predicted and initialized.

[Industrial application field]

The present invention relates to a carrier frequency offset prediction method for a data transmission device such as a modem that uses phase modulation such as the PSK method or the QAM method.

[Conventional technology]

In a line where the stations are connected by microwaves or the like, a slight difference between the oscillation frequencies of the local oscillators of both stations appears as an offset of the carrier frequency of the transmission signal.

In order to correct such a carrier frequency offset, a training signal is transmitted from one modem to transmit data between the modems to adapt the modem to the line characteristics.

On the other hand, it is required to shorten the training time in order to improve the efficiency of data transmission.

To measure the frequency offset, a carrier signal is sent as a training signal from the transmitting side, and it is calculated from the amount of phase rotation per unit time.

In addition, it is necessary to send impulses to the training signal for phase amplitude equalization (EQL).

When a carrier signal containing an impulse is received as a training signal, the impulse is subjected to distortion in the line and the skirt portion spreads on both sides, so the carrier signal is affected by the skirt portion.

When the training time is shortened, the proportion of the tail portion becomes relatively large, and the influence on the carrier signal becomes large.

The conventional frequency offset initialization method has drawbacks in that noise tolerance is weak and pull-in ability is weak in a shortened training time, and improvement of this point has been demanded.

FIG. 3 is a diagram for explaining a conventional frequency offset initialization method.

In the figure, T ₁ , T ₂ , and Tn are delay circuits, and E ₁ and E ₂ are arithmetic units. The same symbols represent the same objects throughout the drawings.

A training signal having a carrier signal and a single impulse is transmitted from a transmission side not shown in FIG.

On the receiving side, the received signal is converted into a digital signal by an A / D converter, and the processing described below is performed on the digital signal, but it will be described as analog signal processing.

On the receiving side, the carrier frequency of the transmitting side is almost the same as that of the carrier signal.
Sin wave and Cos wave are prepared, and the DC components of the alpart and the imaginary part are extracted by demodulating the received signal with both waves and used as the DC component vector.

The extracted DC component is an extremely low frequency signal having a frequency offset component of the line and a single impulse component signal.

This signal is processed by the circuit of FIG. 3 to determine the frequency offset amount.

The DC component is input to the delay circuit T ₁ delay circuit T ₁ → T ₂ → ·
---> Through Tn, the tap output of the delay circuit T _{1 and} the tap output of the delay circuit Tn are taken out as shown in the figure. Although these circuits have a dual structure of real part imaginary part, they are shown as a single structure in the figure.

Next, the arithmetic unit E ₁ multiplies the value obtained by taking the complex conjugate of one tap output and the value of the other tap output. By this processing, the phase difference θ between the two DC components of the delay circuits T _{2 to} Tn that are separated by time is calculated.

Since the signal of this phase difference θ depends on the magnitude of the DC component input to the circuit of FIG. 3, a suitable parameter as a correction value is given to the calculator E ₂ and multiplied by the phase difference θ to correct the position. The phase difference is calculated, and the arithmetic unit E ₂ divides the corrected phase difference by the delay times of the delay circuits T _{2 to} Tn to calculate the phase difference per unit time, that is, the frequency offset amount of the line.

The calculated frequency offset amount is not shown in the CAPC (ca
rrier automatic phase control) circuit, frequency offset initialization was performed to correct the received signal.

Since the operation of the CAPC circuit is not the subject of the present invention, its explanation is omitted, but the correction of the received signal by the CAPC circuit is disclosed in Japanese Patent Laid-Open No. 57-10.
It is described in Japanese Patent No. 4337.

[Problems to be solved by the invention]

In the above conventional method, the training signal sent from the transmitting side is a carrier signal containing a single impulse, and if the training time is shortened, the DC component obtained by processing the received training signal is before and after the impulse. It only exists in the foot part of.

Therefore, calculating the phase difference θ by capturing two points that are temporally separated by the delay circuits T ₁ , T ₂ ··· Tn in Fig. 3 is the sample value of the foot part before and after the transmitted single impulse. Since the above calculation is performed based on the above, there is a drawback that the line characteristics such as noise and phase characteristics have a great influence on the skirt portion, so that an error may occur in the prediction.

An object of the present invention is to provide a frequency offset initialization method which is less affected by line characteristics such as noise and phase characteristics.

[Means for solving problems]

The above problem is that in a modem of a modulation system in which signal components of mutually orthogonal axes are obtained from a received signal and the received signal is demodulated according to the frequency offset amount of the carrier, two impulses are transmitted from the transmission side at a predetermined time interval. At (T), the first impulse is input to the first transversal filter and the second impulse is input to the second transversal filter on the receiving side, and one of the outputs of the transversal filter is input. The frequency offset amount is calculated by taking the complex conjugate of the DC component of the impulse and multiplying it with the DC component of another one impulse to obtain the phase difference between the two impulses and dividing by the time interval (T). Then, the frequency offset amount is sent to the carrier automatic phase control circuit (CAPC) for frequency offset initialization. It is solved by the frequency offset yn Xia Rise method to symptoms.

[Action]

 The present invention is based on the following principle.

That is, when an impulse is sent to the line, the receiving side can obtain an impulse response signal corresponding to the line characteristics.

The pulse sequence forming the impulse response signal is the transfer characteristic itself of the line.

Therefore, if the line is approximated by a transversal filter, it becomes equivalent to a transversal filter having the same tap coefficient as this pulse sequence.

Therefore, if each transversal filter tap coefficient is X ₁ X ₂ X ₃ X ₄ X ₅ X ₆ X ₇ X ₈ X ₉ when an impulse with a value of 1.0 is passed through this transversal filter, each tap coefficient is Pulse sequence X ₁ * with distortion (waveform distortion, phase rotation) and frequency offset received according to
1 X ₂ * 1 X ₃ * 1 X ₄ * 1 X ₅ * 1 X ₆ * 1 X ₇ * 1 X ₈ * 1 X ₉ * 1
Will be received.

On the receiving side, prepare a transversal filter that uses the impulse response as the tap coefficient, and use it as the filter input.
When 1.0 is input, its output is the sum of the tap coefficients of the pseudo transversal filter, that is (X ₁ * 1 + X ₂ * 1 + X ₃
* 1 + X ₄ * 1 + X ₅ * 1 + X ₆ * 1 + X ₇ * 1 + X ₈ * 1 + X ₉ *
_{1 = X 1 + X 2 +} X 3 + X 4 + X 5 + X 6 + X 7 + X 8 + X 9) is obtained.

The sum of these tap coefficients is a DC velcro signal having a phase rotation component received by the line.

Because the cause of the phase rotation generated in the line is caused by each tap coefficient of the pseudo transversal filter, and it does not exist otherwise. Therefore DC vector signal of the sum _{(X 1 + X 2 + X} 3 + X 4 + X 5 + X 6 + X 7 + X 8 + X 9) of each tap coefficient is a component itself represents the phase rotation received by line.

Further, in the present invention, two impulses are sequentially transmitted from the transmitting side at a predetermined time interval T, and the receiving side takes the complex conjugate of the DC component of one of the two impulses to obtain the other one of the impulses. The frequency offset amount of the line is calculated by multiplying with the DC component to obtain the phase difference between the two impulses. The calculated frequency offset amount is sent to the CAPC circuit, and frequency offset initialization is performed.

As described above, according to the present invention, unlike the conventional method that uses the skirts before and after a single impulse, the calculation is performed based on the peak value of each impulse, so that the present invention has a large tolerance to fluctuations in line characteristics such as noise. There is an effect.

〔Example〕

FIG. 1 is a principle diagram of a frequency offset initialization method according to the present invention, and shows a signal position on a complex plane of a received signal.

In the figure, P _V is the vector of the first impulse P, and Q _V is the vector of the second impulse Q.

In the present invention, a first impulse P is first transmitted as a training signal from a transmission side (not shown), and subsequently, a second impulse Q is transmitted after a lapse of time T.

On the receiving side, the first received signal is the frequency-offset received signal.
The impulse P and the second impulse Q are converted into digital signals by an AD converter (not shown). Although digital processing will be performed below, description will be made as with analog signal processing.

On the receiving side, the carrier frequency of the transmitting side is almost the same as that of the carrier signal.
Sin wave and Cos wave are prepared, and the impulse components of the real part and the imaginary part are extracted by demodulating the received signal with both waves.

Next, if the demodulator output is separated into a first impulse component and a second impulse component, and the first impulse component is input to point p in FIG. 2 and the second impulse component is input to q, At the output of the integrator 1P and the output of the integrator 1Q, a DC component having the phase rotation received by the line is obtained.

This is the DC component α of the first impulse P in FIG.
Is the DC component β of the impulse Q of and is the vectors P _V and Q _V on the phase plane.

Here, since the first impulse P and the second impulse Q have a difference by the time T, if the phase difference θ between the vector P _V and the vector Q _{V is} divided by this time T, the phase difference per unit time, that is, the line The frequency offset amount of is obtained.

FIG. 2 is a diagram showing an embodiment of the frequency offset initialization method according to the present invention.

In the figure, 1P and 1Q are integrators, 2 to 6 and 9 are multipliers, and 7
Is an adder, and 8 is a reciprocal ROM for generating a reciprocal.

The received first impulse P and second impulse Q are input to points p and q, respectively, through the AD converter and demodulator (not shown) as described above.

The output of each tap of the delay circuits T ₁ , T _2, ...
And enter in 1Q.

The output of each tap is integrated by integrators 1P and 1Q to obtain a DC component centered on the peak value of the impulse.

As described above, these circuits have a dual configuration of a real part and an imaginary part.

Therefore, this DC component is the vector P _V and the vector Q _{V in} Fig. 1.
Equivalent to.

Next, the vector obtained by taking the complex conjugate of the output vector Q _V of the integrator 1Q and the output vector P _V of the integrator 1P are matched in time and multiplied in the multiplier 2 to output the output vector (x + jy ) Is obtained.

The following processing is performed to normalize the amplitude of this output vector to a point having a radius of 1.

The output vector of the multiplier 2 is separated into a real part and an imaginary part, multiplied by a constant by the multipliers 3 and 5, and squared by the multipliers 4 and 6, respectively.

As a result, the outputs of the multipliers 4 and 6 are scalar quantities (x ² , y ² ).
become. Further, the adder 7 takes the sum of these scalar quantities,
The reciprocal of the square root of its output (x ² + y ² ). Is obtained by the reciprocal ROM8 to obtain a predetermined unit amount.

This unit amount and the output vector (x + jy) of the multiplier 2 are multiplied by the multiplier 9 and the amplitude-normalized phase difference θ vector is obtained to the point of radius 1. Ask for.

Further, the output of the multiplier 9 is divided by the time difference T between the two impulses by a circuit (not shown) to calculate the phase difference per unit time, that is, the frequency offset amount, which is sent to a CAPC circuit (not shown) for frequency offset initialization. Then, the received signal is corrected.

〔The invention's effect〕

As described above, the present invention is calculated based on the integral value of a signal centered on the peak values of impulses P and Q, and the conventional method uses the value of the carrier signal at the foot portion of the impulse. Compared with, the influence on the rotation characteristic is small, and the accuracy of the calculated frequency offset amount is high.

[Brief description of drawings]

FIG. 1 is a principle diagram of a frequency offset initialization method according to the present invention. FIG. 2 is a diagram showing an embodiment of the frequency offset initialization method according to the present invention. FIG. 3 is a diagram for explaining a conventional frequency offset initialization. In the figure, T ₁ , T ₂ , and Tn are delay circuits, P _V is the vector of the first impulse P, Q _V is the vector of the second impulse Q,
1P and 1Q are integrators, 2 to 6 and 9 are multipliers, 7 is an adder, 8 is a reciprocal ROM, and E ₁ and E ₂ are arithmetic units.

Claims (1)

[Claims] 1. In a modem of a modulation system for obtaining signal components of mutually orthogonal axes from a received signal and demodulating the received signal according to a frequency offset amount of a carrier, two impulses are transmitted from a transmitting side for a predetermined time. The first impulse is input to the first transversal filter, the second impulse is input to the second transversal filter, and the first impulse is input to the second transversal filter at the receiving side. The frequency offset amount is obtained by taking the complex conjugate of the DC component of one impulse and multiplying it by the DC component of another one impulse to obtain the phase difference between the two impulses and dividing by the time interval (T). The frequency offset amount is calculated and sent to the carrier automatic phase control circuit (CAPC) for frequency offset initialization. Frequency offset initiative Xia Rise method in which the butterflies.