JPS583413A - Impulse responsive automatic equalizer - Google Patents

Abstract

PURPOSE:To realize a reduction of both the arithmetic quantity required for training and the hardware quantity, by compensating the tap of the 1st filter with the correlation between the output and input series of the 2nd filter. CONSTITUTION:An impulse response detected at an impulse detecting part 1 is supplied to a coefficient symmetrical FIR2 which serves as a transversal filter having the symmetrical tap coefficients centering on the center tap in the training mode. Then the output series of the FIR2 is applied to a fixed equalizer 3 as a tap coefficient. The correlation is obtained by a correlator 4 between the output series of the FIR2 and the correlation values series of the equalizer 3 of its own. The obtained correlation value series is fed to a tap coefficient compensating part 5. At the part 5 the tap coefficient of the FIR2 is compensated with use of the output correlation value series of the correlator 4 so that the correlation value of orders excepting the zero order is set at zero. In such way, the training is over.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention provides impulse training,
The present invention relates to an impulse response automatic equalizer that sets a tap coefficient sequence in a transpanal filter.

Automatic equalizers are used in modems, etc. in data transmission lines, and they estimate distortion in the transmission path from the received signal and automatically set the characteristics to compensate for this distortion, thereby improving the quality of the received signal. It is used for the purpose of compensating for waveform distortion. Generally used as an automatic equalizer is one that uses a transversal filter and automatically adjusts its tap coefficient sequence to obtain desired compensation characteristics.

'In such an automatic equalizer, before the original data transmission, an operation (training) is performed to set the equalization i characteristic by sending a specific signal via the transmission path in advance.
is necessary. There are two methods for training automatic equalizers: one using pseudo-random code sequences and the other using impulses. , the latter method is better.

Furthermore, a method for training an automatic equalizer using impulses involves directly solving the determinant created by the cross-correlation between the impulse response sequence of the transmission line and the error signal.
There is a method to calculate the tap coefficient sequence, and a method in which the impulse response sequence is Fourier transformed, replaced with the frequency domain, flattened, and then inverse Fourier transformed to determine the tap coefficient sequence, but either method requires a small amount of calculation. Since the number of circuits is enormous, it is inevitable that the circuit scale will become large.

In response to this, the same applicant filed a patent application in 1983-12
No. 6487 is a method in which the value of the input series is directly given as the initial value of the tap coefficient, and then the tap coefficient is corrected by the method of least squares so that the output chi and pulse response become 0 except for the median value. has been proposed.This method can perform automatic equalization using relatively simple means.
Although it is actually possible to run on a normal eight-domain scale, if you try to widen the tail of the impulse response, the amount of calculations will inevitably increase, resulting in an increase in the eight-domain size.

The present invention attempts to eliminate such drawbacks of the conventional technology, and its purpose is to reduce the amount of calculations required for training and reduce the amount of hardware. The object of the present invention is to provide a new type of automatic equalizer.

The impulse response automatic equalizer of the present invention includes a first variable-tap transparsal filter that is symmetrical about the center tap, and a second variable-tap transparsal filter that uses the output sequence of the filter as its tap sequence. The taps of the first filter are corrected based on the correlation between the output series of the second transversal filter and the input series, and the number of taps of the first filter is shortened and symmetrical. By doing so, the amount of calculation for correcting the tap coefficients is reduced and hardware is saved. Furthermore, since the number of taps is short, convergence is quick, and performance as an automatic equalizer can be improved.

Hereinafter, the present invention will be described in detail with reference to Examples.

In the impulse response automatic equalizer of the present invention, the transpar fl&' filter is divided into two parts. .

The first transparsal filter, Hayashi 1, is a transparsal filter with symmetrical tap coefficients, and its output series is (y, 'm, )'-mHt =' t yo
* jl , * 9” #Yv-IF, -YM 'j
Then, this output series is expressed by the following equation! The output sequence yn is obtained by solving the equation.

Yn =, #-Lc, Xn-1' = self. Ci (x
n-i ”xn+i) (11 However, in the above formula, Ci is the tap coefficient of the first or transversal filter, (Xn
) is the received impulse response, taking a value of 9o as an example. In
In the transpersal filter 2, the neutral series (y,)
The value of 7 is given as a tap series. Therefore, the output series (Zn) is the autocorrelation series of the output series (yn).

Here, if the transversal fill of voice 1 and the coefficient (C,) of the evening are corrected so as to minimize the sum of squares other than the median value of the output sequence (4), then This means that the desired specific settings have been made. Now, if the coefficient after correction is C"01, and the coefficient before correction is COX", the following relationship holds true.

However, in equation 2), 2- is the output series 2n0. wealthy family
It shows the hand value. FIG. 74 shows the configuration of an embodiment of the automatic impulse response equalizer of the present invention. In the figure, 1 is an impulse detection section, 2 is a coefficient pair, symmetric FIR, S is a fixed equalizer, 4 is a correlator, 5 is a tap coefficient correction section, 0.6 is a buffer, 7 is a judgment section, 8 is, This is the part where the error occurred.

In the figure, it is added to l[ after the data. The incoming signal sequence input from the transmission line, including the training signal containing the impulse signal, is input to the impulse detection section 1,
The impulse detection unit 1 then reproduces the impulse response of the transmission line as a waveform. This process, as shown in Japanese Patent Application No. 172091/1989 filed by the same applicant,
Impulse deprivation may be obtained by directly reproducing the waveform by calculating the sum or difference of the signals when the signal sequence is modulated by a periodic signal, or in general, impulses may be obtained by directly reproducing the waveform. If the included signal is a predetermined random sequence, it can be determined by correlating the # times signal with the random sequence. in this case,
For transmission formats such as quadrature amplitude modulation, the receiving distance is 1m.
The number will be treated as a complex number.

During training, switches 8W1.8W2.8W5 are all connected to the contact a side. Therefore, the impulse response sequence detected by the impulse detection section 1 is inputted to the coefficient symmetric type FIR2 via the switch SW1. The coefficient symmetric type FIR2 is a transparsal filter having tap coefficients that are symmetrical about the center tap, and its output series is directly applied to the fixed equalizer 3 as tap coefficients. As a result, the autocorrelation value series of the output series of the coefficient symmetric type FIR2 is obtained as the output of the fixed equalizer B.

The correlator 4 calculates the correlation between the output sequence of the coefficient symmetric FIR 2 and the autocorrelation value sequence of the fixed equalizer 3°, and inputs the obtained correlation value sequence to the tap coefficient correction unit 5. In the tap coefficient correction section 5, correlation! Using the output correlation value series of i4, the tap coefficients of the coefficient symmetric type FIR2 are corrected by K so that the correlation values other than the 0th order become 0. In other words, the correction of the tap coefficients in the coefficient symmetric type FIR2 focuses on the autocorrelation value of the input impulse and performs equalization so that the autocorrelation values other than the 00th order are set to 0. This method is similar to the method in Japanese Patent Application No. 126487/1983. The correction of the tap coefficients in the coefficient pair unisymmetric form FIR2 is performed as shown in equation (2) by taking advantage of its symmetry, so that the amount of required calculations is significantly reduced.

When the correction of the coefficients of the tap coefficient correction unit 5 for the coefficient symmetric type FIR2 converges (, the training is completed, and the switch 8W1.8W2.8W3 is connected to the contact point b411H.
c can be switched. As a result, the coefficient symmetric FIR2) output sequence is set as the tap sequence of the fixed equalizer 6. The data stored in the buffer 6 is transferred to the fixed equalizer 3
The equalized data is input to the determining section 7, where the sign is determined and an output is generated. The output of the determining section 7 is input to the error generating section 8, where it is compared with a reference signal and minute error components are extracted. The error component extracted by the error generating section 8 is fed back to the correlator 4 and correlated with the input data output from the buffer 6. In the tap coefficient correction unit 5, the characteristics of the automatic equalizer are corrected by adaptively correcting the tap coefficients of the coefficient symmetric FIR using the signal of the correlation value of the correlator 4 using the conventional least squares method or the like. Always keep it in optimal condition.

As explained above, in the impulse response automatic ticketing device of the present invention, procedures such as Fourier transform are not required to determine tap coefficients, and the transversal filter is divided into two, and the taps of the first filter are shorten the number,
Moreover, by making it symmetrical, the amount of calculation for correcting the tap coefficients is reduced, and hardware is saved. Furthermore, since the number of taps is short, convergence is quick, and therefore performance as an automatic equalizer can be improved.

[Brief explanation of the drawing]

The figure is a block diagram showing the configuration of an embodiment of the automatic impulse response equalizer of the present invention. 1... Impulse detection section, 2... Coefficient symmetrical FIR
16... Fixed equalizer, 4... Correlator, 5... Tap coefficient correction section, 6... Buffer, 7... Judgment section, 8
... Patent applicant for the part where the error occurred: Fujitsu Limited representative patent attorney Gobe Tamamushi, and 5 other people

Claims (1)

[Claims] In an automatic equalizer that automatically sets tap coefficient values in a transparsal filter to equalize transmission path distortion,
A first variable-tap transparsal filter that is symmetrical about the center tap, and a second variable-tap transparsal filter that is symmetrical with respect to the center tap, and a second variable tap transparsal filter that is symmetrical with respect to the center tap, and a second variable tap transparsal filter that is symmetrical with respect to the center tap, and a second variable tap transparsal filter that is symmetrical with respect to the center tap;
a variable tap transparsal filter, and when an impulse response reproduced from a training signal containing impulses from a transmission line is input to the first transparsal filter, the second transparsal filter 117)) due to the correlation between the output sequence and the input sequence of
An impulse response automatic equalizer characterized in that equalization is performed by correcting the tap coefficients of a seven-dimensional sparse filter.