JPH0136284B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a data mode convergence method, particularly when data transmission is performed using a multi-point method, when the equalization ability of an automatic equalizer is lost, that is, when a divergence state occurs. This paper relates to a data mode convergence method that enables high-speed convergence.

A modulation/demodulation device (modem) is essential when transmitting digital data handled by a terminal device of a data processing device, a facsimile machine, etc. using an existing analog audio line.

In data transmission for voice band lines, when the data to be transferred is transmitted at high speeds such as 9600 bits/s,
The data to be transmitted is divided into 4-bit units, and as shown in Figure 1, this is modulated to one of the 16 eye reference points and sent out.The receiving side receives the signal from these 16 points. The eye reference point is identified and restored. In this case, when the modem sends data to other terminal devices on the line, it sends out a special pattern called a training signal prior to actual data transmission. On the receiving side, this special pattern is used to control the gain of the automatic gain adjustment circuit in the receiving section, regenerate timing, and preset the automatic equalizer.
It performs the initial sequence necessary for carrier regeneration, etc.

On the other hand, when the automatic equalizer loses its equalization ability (divergence) during the equalization process while receiving data by an automatic equalizer, etc., it requests the transmitter to send out a training signal again and initializes again. A common method is to achieve convergence using a sequence. However, as shown in Figure 2, when several terminal devices are connected in parallel to one transmission line (called a multi-point system), each time the automatic equalizer of the modem receiver of one terminal diverges, this error occurs. If training is repeated, the data processing efficiency of the entire system will decrease. Therefore, it is desired that the modem of the terminal that has diverged can be able to converge again while receiving data.

This will be explained in more detail as follows.

For modems that comply with CCITT Recommendation V29,
Since the coordinate arrangement of the signal, that is, the eye reference point, is not symmetrical as shown in FIG. I can't determine the position of the tap. Therefore, to start training, all tarips are grown from a constant value (eg 0) using local reference signals.

After this process, the position and size of the central tap are determined. These tap coefficients are generally variable values because they are adaptively modified in response to line fluctuations. However, if the equalization ability of the automatic equalizer is lost due to the addition of some kind of disturbance (large in-bulk noise interruption, large phase shear, etc.), each tap coefficient that was secured at the time of convergence becomes random. value and cannot be restored again.

Therefore, in order to improve this problem, the present inventor, applicant, etc. first tried to match the average power P _E of the equalizer output signal with the average power P _R of the reference signal held by the receiver. The center tap of the transversal type automatic equalizer is fixed at a constant value, and the equalizer output is determined in order to modify other tap coefficients using data with little interference between data. Data mode convergence method (Special application)
No. 54-155342, Japanese Patent Application Laid-open No. 56-78243).

The reconvergence method already proposed will be explained below with reference to FIG.

Incoming data must be equalized to one of the eye reference points (signal points) shown in Figure 1.If the line noise is Gaussian noise, each equalized data The signal points are uniformly distributed within a certain circle. Furthermore, since the probability that these signal points are transmitted is the same, if equalization is performed normally, the average power value Pd of the equalized data will be uniquely determined by the predetermined signal arrangement. It corresponds to a certain determined power value P ₀ .

However, 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 power value Px does not match Po. Therefore, the power average value Px of this diverging data is
One of the conditions for convergence is to match Po.
However, simply applying this condition does not guarantee convergence of the equalizer. This is because it is insufficient to correct vector data that also includes phase using power alone, and correction for phase is also required. This uses a statistical correction method (maximum slope method) that uses the squared error of the equalized data as an evaluation function, but when it diverges, it is difficult to evaluate the error for all data because there is a lot of erroneous data. Not effective.

Therefore, it is conceivable to use an error that is relatively correct as the second condition. In other words, in the signal arrangement shown in Fig. 1, each point in the dotted line inner area A has a shorter signal interval than each point in the dotted line outer area B, so there is a high probability that it will be judged incorrectly, so it is used for correction. This cannot be expected as much information, and therefore each point in the external region B is desirable data for correction.

However, if only external signal points are used in a divergent state, the polarity of the error will be biased and control will be unilateral. This depends on the position of the outside signal points within the determination area, but it is because the signal point arrangement in FIG. 1 has an overwhelmingly large number of outside points.
Therefore, the tap coefficient of the equalizer is modified in one direction, in this case in the direction of lowering the gain of the equalizer.
The equalizer gain must be compensated in some way. Therefore, at least the value of the center tap coefficient is fixed. Furthermore, an AGC circuit is placed at the equalizer output to satisfy the above condition, that is, matching of average power.

To implement this, as shown in FIG.
Input waveform equalization is performed by convolving C _k with input data Xi. Its general description is expressed as follows.

Y _k = _N 〓 ^n=-N Cn・Xk− _o C _p・Xk＋ _N 〓′ ^n=-N Cn・Xk− _o However, the latter stage Σ′Cn・Xk− _o is other cumulative value except n=0 shows.

The tap coefficient Ck takes various values depending on the line characteristics, but when it diverges, each tap coefficient Ck takes on an arbitrary value independently. Moreover, if the value of the coefficient Co of the center tap is variable, it will not converge again from this state.
Therefore, if a divergence of the equalizer is detected, the central tap coefficient is fixed at a certain value Ko, for example Ko=1.
This ensures the first basis for convergence of the equalizer. Next, a circuit that performs tap correction of the equalizer 10 by driving an outer point detection circuit 7 that extracts only the error related to the signal point in the outer region B shown in FIG. 1 from the divergent equalization output error e; That is, it is input to the tap coefficient calculation circuit 11. The tap coefficient calculation circuit 11 executes an algorithm (such as the maximum slope method using the least square error as the evaluation relation) for adaptively equalizing the equalizer 10. In the outer point detection circuit 7, the input data Xi is equalized by an automatic equalizer 10, subjected to phase adjustment by a phase adjuster 12, and transmitted to the determination unit 1. Based on the coordinate position, the 16 points shown in FIG. It is identified whether it corresponds to any of the signal points. This determining section 1 is provided with a signal determining surface in the form of a ROM, for example, in association with each signal point, and it is determined which signal point the input data Xi belongs to based on this signal determining surface. The determination output is sent to the outside and also transmitted to the outside point detection circuit 7, so that when the equalized output is determined to be a signal point in the outside area B (hereinafter referred to as outside signal point), This outer point detection circuit 7 is
It outputs 1 to the multiplier 5, and conversely outputs 0 to the multiplier 5 when it is determined that the signal is from the inner area A (hereinafter referred to as an inner signal point).

On the other hand, the output of the automatic equalizer 10 is transmitted to the equalized output power calculating section 2, and the magnitude of the vector of the output signal is calculated. At this time, the equalized output power calculating section 2 calculates the power of the output signal of the automatic equalizer 10 by (Real) ² +(Image) ² . Further, the original reference signal arrangement, that is, the unique average power of each of the 16 signal points, is determined by the reference power calculation unit 3, which calculates the power of the signal point by (Real) ² +
(Image) Calculate with ² . Then, the difference between the two powers is outputted by the adder 4. The value of this difference is
When the signal is an inside signal point, it is multiplied by 0 in the multiplier 5, and when it is an outside signal point, it is multiplied by 1. Therefore, only the difference at the outside signal point is integrated by the integrating circuit 8, and the limiter 9 is It acts as a control signal for the automatic equalizer 10 via the signal. Further, the equalization output and the judgment output are compared by the adder 6, and the phase adjustment is performed by the phase adjuster 12 according to the difference, but this is also performed using the outer signal points. Of course, the input data is digital AGC circuit 1
The automatic equalizer 1 is adjusted to a constant size by 3.
0.

In this way, in the circuit shown in FIG. 3, when the power of the final equalized output is large compared to the average power of the reference signal, a negative value is accumulated in the integrating circuit 8, and its output is also negative. , which is a constant
The value H that is multiplied by the equalized output because it is added to Ko
is decreased, the value of the equivalent output B becomes smaller,
The average power of this equivalent output can be brought close to the reference power.

However, although convergence is possible using the convergence method shown in FIG. 3, it has the disadvantage that it takes a long time to converge.

Therefore, in order to improve these drawbacks, the present invention provides a data mode that enables high-speed data mode convergence by performing control based on the power difference and further based on the amount of intersymbol interference. The purpose is to provide a convergence method. To this end, in the data mode convergence method of the present invention, a control amount is added to the tap of the transversal automatic equalizer based on the difference between the output signal power of the transversal automatic equalizer and the reference signal power held on the receiving side. The tap coefficient is corrected by semi-fixedly applying a constant value to control the output signal power of the transversal type automatic equalizer to match the reference signal power, and the tap coefficient of the transversal type automatic equalizer is In a data mode convergence method in which a determination for modifying tap coefficients is made using data with a small amount of interference between data, an intersymbol interference signal is extracted from the transversal automatic equalizer, Based on this, the control amount of the transversal automatic equalizer is controlled based on the amount of intersymbol interference as well as the signal power level for the central tap.

An embodiment of the present invention will be described based on FIG.

In the figure, parts with the same reference numerals as others indicate the same parts. 10'
is a transversal type automatic equalizer, which includes a tap coefficient calculation circuit 11' for extracting the amount of intersymbol interference. This transversal type automatic equalizer 10' corresponds to the transversal type automatic equalizer 10 in FIG. 3, and the tap coefficient calculation circuit 11' corresponds to the tap coefficient calculation circuit 11.

14 is a multiplier, which is a tap coefficient calculation circuit 1;
1' is multiplied by a constant _r1 .

15 is a multiplier that adds the intersymbol interference signal multiplied by _r1 and the difference signal between the equalized output calculation section 2 and the reference power calculation section 3 output from the adder 4; It is. This is then integrated by an integrating circuit 8 and applied via a limiter 9 as a coefficient Co to the center tap of the automatic equalizer 10'.

Note that in the tap coefficient calculation circuit 11', there is a calculation formula for tap correction for each tap T, T, etc., as in the known transversal type automatic equalizer, and the tap coefficient is calculated using this calculation formula. Based on this calculation, each intersymbol interference amount can be obtained. Based on such well-known matters, the amount of intersymbol interference is calculated, and the calculation result is obtained as an intersymbol interference signal, which is converted into a constant by the multiplier 14.
It is multiplied by r ₁ and used as a control amount.

According to the present invention, in addition to control based on the power difference, which is the amplitude difference between the so-called equivalent output signal and the reference signal, it is also possible to perform control based on the amount of intersymbol interference for the central tap, which is the dominant tap. As a result, it is possible to converge the data mode extremely quickly even when the eye diverges.

[Brief explanation of drawings]

Figure 1 is an explanatory diagram of the eye reference point, Figure 2
The figure is an explanatory diagram of the multi-point system, and Figure 3 is a configuration diagram of the data mode convergence method using power difference.
FIG. 4 is a configuration diagram of an embodiment of the present invention. In the figure, 1 is a determination section, 2 is an equalized output power calculation section,
3 is a reference power calculation unit, 4 is an adder, 5 is a multiplier,
6 is an adder, 7 is an outer point detection circuit, 8 is an integration circuit, 9 is a limiter, 10 and 10' are transversal type automatic equalizers, 11 and 11' are tap coefficient calculation circuits thereof, and 12 is a phase adjuster. , 13 is a digital AGC circuit, 14 is a multiplier, and 15 is an adder.

Claims (1)

[Claims] 1. A control amount and a constant value based on the difference between the output signal power of the transversal automatic equalizer and the reference signal power held on the receiving side are semi-fixedly applied to the tap of the transversal automatic equalizer, and then tapped. The coefficients are corrected so that the output signal power of the transversal type automatic equalizer and the reference signal power match, and the judgment for correcting the tap coefficients of the transversal type automatic equalizer is performed using data exchange. In a data mode convergence method using data with a small amount of interference between symbols, an intersymbol interference signal is extracted from the transversal automatic equalizer, and the transversal automatic equalizer is controlled based on this. A data mode convergence method characterized in that the amount of data mode convergence is controlled based on the amount of intersymbol interference as well as the signal power level for the center tap.