JPS631782B2 - - Google Patents

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to an automatic equalizer in a modem device used for data transmission, and in particular, to an automatic equalizer using a transversal filter. The present invention relates to a method for controlling automatic equalization of a transversal filter for convergence when it is lost, that is, when it becomes a divergent state.

[Conventional technology]

Automatic equalizers are essential for high-speed data transmission in order to compensate for line characteristics. Automatic equalizers are mainly composed of transversal filters with good phase characteristics to perform waveform equalization.
Generally, the tap coefficient of the filter cannot be uniquely determined because the line characteristics are unknown. In particular, the CCITT recommendations
For modems that use absolute reference, such as V29-compliant modems, even the center tap of the filter is undetermined, and these tap coefficients are usually determined by a special sequence called training.

Furthermore, the tap coefficient is adaptively modified to take a variable value in order to follow various fluctuations in the line.

However, due to some disturbance (such as large impulsive noise or instantaneous interruption), the equalized waveform output from the transversal filter may differ significantly from the reference signal waveform, and the error signal may become large. If this error signal does not converge even after a predetermined period of time has elapsed in this state, the tap coefficient of the transversal filter becomes uncontrollable, and the equalization ability of the automatic equalizer is lost. This state is generally called a divergence state, but when this divergence state occurs, each tap coefficient that was secured in the converged state becomes a random value, and the tap coefficients must be controlled to return to the converged state again. becomes impossible.

Therefore, if things continue as they are, training will have to be performed again to achieve convergence (determination of each tap coefficient).

[Problem that the invention seeks to solve]

Conventionally, when this type of modem is configured in a multipoint system where many remote stations are connected to one station using the same line and perform data communication alternately, if one station diverges, there is no response. Since no compensation had been made, the entire system had to be shut down and retrained.

This was a major drawback in making the system work efficiently.

The present invention provides a training method from data state that is useful as a method for dealing with this.

[Means for solving problems]

The present invention provides a transversal type automatic equalizer that normally controls the center tap and other taps based on judgment errors, and when the transversal type automatic equalizer enters a divergent state, the transversal type automatic equalizer Setting the coefficient of the center tap of the equalizer, and calculating the power of the signal of the equalizer output data in order to match the output power of the transversal type automatic equalizer to the average power of the reference signal possessed by the receiving section, The difference from the average power is accumulated, the gain of the phase control circuit installed in the output stage of the transversal type automatic equalizer is controlled by the accumulated value, and the output of the phase control circuit is adjusted to the output stage of the transversal type automatic equalizer. It is characterized by controlling the output by multiplying the output of the equalizer, and converging the equalizer using data of a signal arrangement with less interference between signals among the final equalized outputs.

The basic concept is that the average power (PE) of the equalizer output signal is equal to the average power (PR) of the reference signal at the receiver.
Automatic gain control (AGC) is performed on the equalization output, and in tap control, the center tap of the transversal automatic equalizer is fixed at a constant value, and other tap coefficients are modified. The equalizer output is determined using data with a small amount of interference between data. This feature is for reconvergence of the receiver equalizer.

Furthermore, when applied to the phase correction circuit postfix method, which is effective as a method for correcting phase distortion,
This difference from the average power is fed back to the output signal of the phase correction circuit to simplify calculations.

〔Example〕

Figure 1 shows what is called an automatic equalizer after a phase correction circuit, and the phase correction is performed using an existing method that can be expected to quickly track the phase because there is no delay from a filter or equalizer.

Phase correction circuit 1 is a PLL using a double integral type circuit.
The system extracts the phase error from the judgment error vector and controls the PLL.
Applying the algorithm of this method to this type of equalizer means that the output Y of the first half transversal section E is multiplied by the original phase correction circuit output AP to obtain the equalized output S, and at the same time, this equalization The output power is determined by the absolute value circuit 8, and the difference between this and the reference power obtained by integrating the reference signal R by the integrating circuit 5 is compared by the comparator 7, and the difference is accumulated by the accumulator 6. This controls the output level of the phase correction circuit 1,
The difference with the reference power is eliminated.

In other words, the output Y of the transversal section E in the first half
is multiplied by the amplitude-controlled signal Ap, resulting in S. Here, the output Ao of phase correction circuit 1 and
In the function with Ap, only the amplitude changes, and the direction of the vector remains the same.

The signal S obtained as the final equalized output is compared with the reference signal R, and the comparator 2 generates the difference e. One reason for the difference e is the transversal part E.
are fed back via line 11 to correct their respective tap coefficients. On the other hand, the amplitude information is normalized by the normalization circuit 3 using normalized data from the data generator 4, and only the phase Pe is extracted and input to the phase correction circuit 1.

Here, the received data has a predetermined signal arrangement (16 points in Figure 2) if the multi-level orthogonal amplitude modulation method is adopted, and must be equalized to one of them. If the line noise is Gaussian noise, then the equalized data will be uniformly distributed within a certain circle for each signal point.

In addition, since the probabilities that these signal points are transmitted are all the same, if equalization is performed normally, the average power value (Pd) of the equalized data will be based on the predetermined signal arrangement. It perfectly matches a certain uniquely determined power value (PR).

On the other hand, if normal equalization is not performed and the equalizer diverges, the equalized data can be any value up to the hardware limit, for example any value from the maximum value to the minimum value in terms of arithmetic accuracy. Therefore, the average value of power (Pd) does not match PR.

Therefore, it is necessary to make the average value PD of the power of this divergent data match PR.

However, simply applying this condition does not compensate for the convergence of the equalizer.

This is because it is insufficient to correct vector data including phase using power alone.
That is, some kind of correction is also required for the phase. For this purpose, a statistical correction method (maximum slope method) is used that uses the squared error of the equalized data as the evaluation index, but when the error diverges, it is impossible to evaluate the error for all the data because the data is too erroneous. It's not very effective because there are so many of them.

Therefore, it is possible to use an error that is relatively correct. In other words, in the signal arrangement shown in Figure 2, since the signal interval at the internal point A is shorter than that at the external point B, there is a high probability that the signal will be incorrectly determined, and this information cannot be expected to be used for correction. That's true. That is, the external B point is the desired data for correction.

The normalized data from the data generator 4 is therefore a value corresponding to the external point B of the signal constellation in FIG.

The following two methods can be considered for determining the external B point in the data generator 4.

(1) A method of dividing the coordinate space shown in Figure 2 into N×N square regions and checking whether the reference signal exists in the external region B.

(2) Divide the coordinate space shown in Figure 2 into several line segments radially, and further divide each line segment into two.
A method of checking whether a reference signal exists in the area of external B.

Then, a reference signal at the external point B is output using one of these methods.

Therefore, in the circuit 2 that performs a level comparison between the equalized output S and the reference signal R, it is determined that the variation amount (differential amount) of the judgment error e is equal to or higher than a predetermined level even after a predetermined time elapses. , it is detected that the transversal filter has entered a divergent state. In this case, the reference signal is a signal that randomly generates amplitude values and phases such as the 16 points in FIG. When the normalized data generating circuit 4 detects that the divergence state has occurred, it selects only the signal in the B area from the reference signal R and outputs it to the normalizing circuit 3, as described above. The normalization circuit 3 detects a phase difference Pe from the error signal with respect to the B-area signal based on the signal from the normalization data generation circuit 4, and outputs it to the phase correction circuit 1.

In FIG. 1, an integrator 5 is an integration circuit for averaging the determined level of the reference signal.

Time t obtained from this output and the absolute value circuit 8
A comparator 7 is provided, which compares the level of the equalized output at the circuit 7 with the level of the equalized output. When the equalized output level is greater than the average value of the reference signal, a negative value is accumulated in the accumulator 6. Since this is added to the value Co (usually 1) that determines the original gain of the phase correction circuit 1, the gain of the phase correction circuit 1 apparently decreases. In other words, the final output Ap of the phase correction circuit 1 at time t+1 becomes smaller, and the final output Ap of the phase correction circuit 1 at time t+1 becomes smaller.
Decrease S at 1. Conversely, when the equalized output level is small, a positive value is accumulated in the accumulator 6, and from the same consideration as above, Ap becomes large and S
increase. In this way, since the equalized output S becomes close to the average power of the reference signal, the pull-in becomes easy.
Furthermore, the tap coefficients of the transversal equalizer are fixed at arbitrary positions as a median value, or semi-fixed with a certain width, and this is one of the conditions for convergence of the equalizer.

〔Effect of the invention〕

As described above, according to the present invention, when transmitting data using an equalizer whose center tap position is undefined, when it is detected that the equalizer is in a divergent state, the equalized output is changed to the reference value. Since the tap coefficient of the transversal filter is controlled so as to match the average power of the signal, and the tap coefficient of the transversal filter is controlled by comparison with the reference signal at the outer signal position, the data can be adjusted without requiring the training signal again. This makes it possible for many remote stations to share the same transmission path and transmit data.
Effective in multi-point system.

[Brief explanation of the drawing]

Figure 1 is a block diagram of one embodiment of the method of the present invention.
FIG. 2 shows a signal distribution diagram for explaining the system of the present invention. In the figure, E is a transversal equalizer;
1 is a phase correction circuit, 2 is a comparator, 3 is a normalization circuit, 4 is a normal data generator, 5 is an integrator, 6 is,
Accumulator, 7 is comparator, 8 is absolute value circuit.

Claims (1)

[Claims] 1. In a transversal automatic equalizer that normally controls the center tap and other taps based on judgment errors, when the transversal automatic equalizer enters a divergent state, the transversal In addition to setting the coefficients of the center tap of the transversal automatic equalizer, the power of the equalizer output data signal is adjusted to match the output power of the transversal automatic equalizer to the average power of the reference signal of the receiver. calculate, accumulate the difference from the average power, control the gain of a phase control circuit arranged in the output stage of the transversal type automatic equalizer using the accumulated value, and control the output of the phase control circuit in the transversal type automatic equalizer. A data training method characterized by controlling the output by multiplying the output of an automatic equalizer, and converging the equalizer using data of a signal arrangement with less interference between signals among the final equalized outputs.