JPS62226723A - Automatic equalizer - Google Patents

Abstract

PURPOSE:To reduce the converging time of an automatic equalizer by obtaining an initial value of a tap coefficient by operation from a training signal. CONSTITUTION:The initial value Of the tap coefficient is obtained in a short time to obtain the initial value of the tap coefficient by the calculation of a high speed Fourier transformation section 1 and an inverse high speed Fourier transformation section 4 or the like from the training signal 100. Since the training signal 100 used for deciding the initial value of the tap coefficient is inputted to a data RAM 7 as the initial value of the automatic equalizer, the start time of the automatic equalizer is reduced remarkably. Further, the training signal inputted to the data RAM 7 as the initial value is a binary signal whose position is clear and which is simple, while being outputted from a training pattern generating section 111 of a transmitter 11, then the decisisn in a decision section 134 is easy and the converging time of the tap coefficient of the tap coefficient RAM 6 is reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION [Purpose of the Invention (Field of Industrial Application) The present invention relates to an automatic equalizer that removes code amount interference caused by transmission path distortion and the like.

(Prior Art) Conventionally, as for the initial values of the tap coefficients of an automatic equalizer, there have been methods in which an initial value other than O is given only to the center tap, and O is given as an initial value to the other tap coefficients, or , a method is used in which O is given as an initial value to all tap coefficients.

However, in these methods, the initial value of the tap coefficient is significantly different from the value after convergence, so from the start of training,
The time required for the tap coefficients to converge becomes longer. Roughly speaking, one training word sent from the transmitting side via the transmission line and subjected to transmission line distortion cannot be used as the initial value of the signal to be equalized by the automatic equalizer, so ``m, etc. It requires time to input another signal as the initial value of the equalized signal, and a signal different from the training signal becomes the initial value of the signal to be equalized, so after inputting the initial value, it is easy for the determination unit to make an error in the determination. In this respect as well, there is a drawback that a long FR interval is required for the tap coefficients to converge.

(Problems to be Solved by the Invention) As mentioned above, in the conventional automatic equalizer system, the training signal for which the initial value of the tap coefficient is determined cannot be used as the initial value of the signal to be equalized, etc. This has the disadvantage that it takes a long time for the tap coefficients to converge.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide an automatic equalizer that eliminates the above-mentioned drawbacks and can shorten the convergence time of tap coefficients.

[Structure of the Invention] (Means for Solving the Problems) The automatic equalizer of the present invention includes a Fourier transform means for performing Fourier transform on a training signal input from a transmission path to calculate the frequency characteristics of the signal, and a transmission a comparison means for calculating an inverse characteristic of the transmission path distortion by comparing a frequency characteristic of a training signal that is not subjected to path distortion with a frequency characteristic obtained from the Fourier transform means; and a plurality of taps that performs Fourier transform on the inverse characteristic. Fourier transform means for calculating coefficients, tap coefficient holding means for holding the tap coefficients, signal holding means for holding training signals inputted from the transmission path in time series as initial values, and signal holding means for holding the signal 8. calculation means for multiplying the signal held in the tap coefficient holding means by the tap coefficient held in the tap coefficient holding means and adding the result; and a determination control means for feeding back to correct the tap coefficients of the tap coefficient holding means.

(Function) In the automatic equalizer of the present invention, the Fourier transform means,
By using the comparison means and the inverse Fourier transform means, the inverse characteristic of the transmission line distortion can be calculated at once from the training signal input from the transmission line, so the initial value of the tap coefficient can be obtained in a short time. . Therefore, the training signal input from the transmission path can be input to the signal holding means as the initial value of the signal to be equalized, and the binary undam signal can be input as the training signal to be input as the initial value of the signal to be equalized. When using this, the determination control means can easily make a highly accurate determination, so that the convergence time of the tap coefficients can be shortened.

(Example) An example of the present invention will be described below with reference to the drawings. 1st
The figure is a block diagram showing one embodiment of the automatic equalizer of the present invention. 1 is the demodulated signal of the received training signal (
A fast Fourier transform unit that performs fast Fourier transform on a training signal (hereinafter simply referred to as a training signal) and calculates its frequency characteristics;
2 is a memory that holds frequency characteristic data of a training signal that is not subjected to distortion; 3 is a memory that holds frequency characteristic data given from the fast Fourier transform section 1 by the frequency characteristic data given from memory 2, and the transmission is carried out for each frequency. 4 is an inverse fast Fourier transform unit that performs an inverse fast Fourier transform on the inverse characteristic to calculate N tap coefficients; 5 is an inverse fast Fourier transform unit that calculates N tap coefficients given from the inverse fast Fourier transform unit 4; 6 is a tap coefficient ram that holds N tap coefficients given from the tap coefficient cyclic unit; 7 is a tap coefficient ram that exchanges the given tap coefficients so that the largest among the tap coefficients becomes the center tap; 8 is a multiplier that multiplies the N signal values supplied from data ram 7 by the tap coefficient supplied from tap coefficient ram 6; An adder 134 that takes the sum of the multiplication results of the N multipliers 8 determines whether the corrected signal output from the adder 9 is a normal signal, and so that the error between the corrected signal and the normal signal is minimized. This is a determination unit that feeds back errors obtained in order to correct each tap coefficient of the tap coefficient ram 6.

FIG. 2 is a block diagram showing an example of a communication system including the automatic equalizer shown in FIG. 1. A transmitting device 11 and a receiving device 12 are connected by a transmission path 13. The transmitter 11 includes a training pattern generator 111 and a modulator 1.
12 and a D/A converter 113. The receiving device 12 also includes an A/D conversion section 131, a demodulation section 132, an automatic equalization section 133, and a determination section 134.

Next, the operation of this embodiment will be explained. First, during training to determine the tap coefficients of the automatic isolator, the training signal 100 input to the fast Fourier transform unit 1 and data ram 7 shown in FIG. 1 is input from the transmitter 11 shown in FIG. 2 as described below. Ru.

The transmitting device 11 in FIG. 2 is the receiving device of the other party! When training the 12 automatic equalizers, a training signal is sent out through the transmission line 13. The training pattern generation unit 111 of the transmitting device 11 has a built-in binary undam signal as a training pattern in which specific two points A and I out of the plurality of signal points shown in FIG. 3 are randomly conveyed. During training, this built-in training pattern is output to the modulation section 112. The training pattern is modulated by a modulation section 112, and further modulated by a D/Δ conversion section 113.
The signal is converted into an analog signal and sent to the receiving device 12 via the transmission line 13 as a training signal. Receiving device 12
The Δ/D converter 131 returns the received training signal to a digitalized modulated training pattern and outputs it to the demodulator 132. The demodulator 132 demodulates the input modulated training pattern and outputs the demodulated training pattern to the automatic equalizer 133 as a training signal 100. Returning to FIG. 1, the training signal @100 input as described above is input to the fast Fourier transform section 1 and the data ram 7. The fast Fourier transform unit 1 transforms the input training signal 100 from the time domain to the frequency domain and calculates the frequency characteristics of this training signal 100. In this case, since the training signal 100 is sent via the transmission line 13 in FIG.
The frequency characteristic of 100 includes distortion due to the transmission line 13. The frequency characteristic data of the training signal @100 obtained by the fast Fourier transform unit 1 is input to the comparator 3, and the frequency characteristic data of the undistorted training signal held in the memory 2 is also input to this comparator 3. Here, the frequency characteristic data from the fast Fourier transform section 1 is divided by the frequency characteristic data from the memory 2 for each of the N frequency regions, and as a result, Nflli
! The inverse characteristic of the transmission line distortion in each frequency region is obtained. These N inverse characteristics are subjected to inverse fast Fourier transform in the inverse fast Fourier transform section 4, resulting in N tap coefficients. These N tap coefficients are sent to the tap coefficient circulation unit 5, where the tap coefficients are exchanged so that the largest tap coefficient becomes the center tap, and the exchanged tap coefficients are held in the tap coefficient ram 6. Ru.

On the other hand, the training signal 10 input to the data ram 7
0 is held at N points in time series in the data column 7 as the initial value of the automatic equalizer. Next, each of the N multipliers 8 multiplies the signal value at each time point given from the data ram 7 by the tap coefficient given from the tap coefficient ram 6. The multiplication results of these multipliers 8 are summed by an adder 9, and as a result, a modified waveform of the training signal input to the data column 7 is output from the adder 9. The determination unit 134 determines the deviation of this corrected waveform from the original position shown in FIG. 3 as an error. Each tap coefficient in the tap coefficient column 6 is modified so that the error determined by the determination section 134 is minimized. Here, the time taken to obtain the tap coefficient with the minimum error of the determination unit 134 is the convergence of the tap coefficient Ill\\
It is called between. After the tap coefficients are held in the tap coefficient column 6 in this way, a normal signal is sent from the transmitter 11 in FIG. 2 to the receiver 12 via the transmission line 11, and the automatic equalizer 133 By repeating the operation, the adder 9 outputs a signal from which code amount interference caused by distortion in the transmission line 13 has been removed.
The signal is then supplied to the determination unit 134.

According to this embodiment, the initial values of the tap coefficients are obtained from the training signal @100 through calculations in the fast Fourier transform unit 1, the inverse fast Fourier transform unit 4, etc., so the initial values of the tap coefficients can be obtained in a short time. be able to. Since the training signal 100 used to roughly determine the initial values of the tap coefficients can be directly input to the data ram 7 as the initial values of the automatic equalizer, the operation start time of the automatic equalizer can be significantly shortened. I can do it. Further, since the training signal inputted to the data ram 7 as an initial value is a simple binary signal whose position is clear and output from the training pattern generation section 111 of the transmitting device 11, the determination section 1
34 is easy to determine, and the convergence time of the tap coefficients of the tap coefficient ram 6 can be shortened.

Effects of the Invention] As described above, the automatic equalizer of the present invention has the effect of shortening the convergence time of the automatic equalizer by calculating the initial value of the tap coefficient from the training signal.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the automatic equalizer of the present invention, FIG. 2 is a block diagram showing an example of a communication system including the automatic equalizer shown in FIG. 1, and FIG. 1 is a diagram illustrating selection of a binary null signal as a 1 to laning pattern. 1...Fast Fourier transform unit 2...Memory 3...Comparator 4...Inverse fast Fourier transform unit 5...Tap coefficient circulation unit 6...Tap coefficient ram 7...Data ram 8 ... Multiplier 9 ... Adder 134 ... Judgment Department Agent Patent Attorney Nori Chika
Ken Yudo Sanno -

Claims (1)

[Claims] Fourier transform means for performing Fourier transform on a training signal input from a transmission path to calculate the frequency characteristics of this signal; and a frequency characteristic of a training signal that is not subjected to transmission path distortion and a frequency characteristic obtained from the Fourier transform means. a comparing means for comparing and calculating an inverse characteristic of the transmission path distortion; an inverse Fourier transform means for inverse Fourier transforming the inverse characteristic to calculate a plurality of tap coefficients; and a tap coefficient holding means for holding the tap coefficients. , a signal holding means for holding a training signal input from the transmission path as an initial value in time series; and a signal holding means for multiplying the signal held in the signal holding means by a tap coefficient held in the tap coefficient holding means. It comprises a calculation means for adding the results, and a determination control means for determining the deviation of the output signal of the calculation means from the normal position and feeding back the error in order to correct the tap coefficient of the tap coefficient holding means. Features an automatic equalizer.