JPS6027457B2 - automatic equalizer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to the configuration of a transversal automatic equalizer in multilevel data transmission using orthogonal amplitude modulation.

In quadrature amplitude modulation, the pre-modulation part of the modulator and the post-demodulation part of the demodulator each have signals consisting of two sequences.

Therefore, when filtering is performed in this part, exactly the same operation must be performed on two series of signals.
However, when such a filter is created, in many cases, equipment errors occur, making it impossible to perform exactly the same operation, which causes distortion. This type of distortion cannot be removed by the conventional transversal automatic equalizer, which has two tap gains on each tap and allows these tap gains to act on the signal in a bridge manner. belongs to. The present invention is an automatic equalizer characterized in that by having four tap gains in each tap, it is possible to equalize distortion caused by device errors in a modem with a device having almost the same level of strength as a conventional automatic equalizer. It is a transformer.
The operating principle will be explained below.

In quadrature amplitude modulation, the modulation waveform &(t) is as shown in the following equation.

However, a; and bi are transmission data, r, (t) are ai
The impulse response of the basal band filter to r
2(t) is the impulse response of the baseband filter to bi, is the conversion angular frequency, and T is the data interval.

Received signal v when this signal is distorted through the communication channel
(t) is expressed by the following equation, assuming that the impulse response of the communication channel is P(t).

To simplify the formula, P (t) Nino rl (t-i) P (7) 7d7 of COS
'3' ql (t) Ni'rl (t-7) P (7) Sin's Id Ding {4} P2 (t) Ni'r2 (t-7) P (Ding) C
OS's 7d7'51q2(t)ni'r2(t-7)P(
7d7'6} of Sin, then the equation (2) can be rewritten as follows.

Therefore, the two detected outputs after demodulation are expressed by the following equations. An orthogonal relationship does not hold between the equation {8} and the two signals expressed on the equation side.

Therefore, the automation devices currently in use for orthogonal signals are not effective in such cases. Here, equalization is performed using a transversal automatic equalizer having a one-tap configuration and four types of tap gains.

The formula giving the input and output of the equalizer is shown below.

However, N is the number of taps, Cp・1, CO. ,,D
p・1, D〇. -(]=1, 2...N) is the tap gain.

The equalized output signals given by equation (1) and equation (11) are sampled, and estimated data is obtained by the determination circuit.

Therefore, in equations 00 and (11), it is only necessary to pay attention to the values at sample points at signal intervals T. P, (t), q2(t
), P2(i), ql(t), sentence(t), y(t), g
The sample value sequences for signals such as (t) and h(t) are respectively P.・n, q2・n, P2・n, q.・n, xn, blood,
If expressed as gn, hn, etc., gn and hn are expressed as follows from equations '8), (9}, ■ and (11). gn expressed by equations (12) and (13) and hn are completely equalized by an and b, respectively.
It must be n. As an example, Pl. n
Niql. n d P2. nniq2. n=0 (n≠-1)
Considering the case (14), gn
In order for =an and hn=bn to always hold true, the following equations must hold true.

The above equations are all 0 unless each tap gain is 1 = 0.
When 1=0, it is completely satisfied by giving the tap gain value by the solution of the four-dimensional linear system of equations,
Complete equalization is therefore possible. However, in the conventionally used complex operation type equalizer, (14)
If the formula holds, (15), (16), (17
) and the four equations corresponding to equation (18),
Since there are only two variables, the in-phase and quadrature tap gain values, there is generally no solution and sufficient equalization cannot be achieved.Of course, in actual cases, the distortion is calculated using equation (14). It's not something that satisfies me, but P...1, P2.1, q...,
, q, etc. have non-zero values over several sample points. However, even in this case, each tap has 4
If it is an automatic equalizer with one tap gain, by having a sufficient number of taps like a normal automatic equalizer,
Distortion can be brought close to 0. As a means for reducing distortion, there is, for example, an algorithm that minimizes the mean square error, which will be described later. For the reasons stated above, an equalizer having four tap gains per tap can effectively equalize the demodulated signals expressed by the following equations.

Also, although we are dealing with the shift in the filter of the modulator here, even if there is a shift in the filter after modulation of the demodulator, the input signal to the equalizer will be the same as in ■ and Equation '9}, and will be equivalent. effective. Next, a method for automatically modifying the tap gain of the equalizer will be described. The modification method uses the evaluation function as n=(gn-an)2N0(hn-bn)2
(19) is used, and a method is used in which the tap gains are successively determined by the steepest descent method that minimizes this.

Therefore, consider the differential of equation (19) with respect to the value of each tap gain, and make corrections according to this amount. The modification formula is expressed by the following formulas. Cp*1=Cpl-Qxn-, (gn
-an) (20) DP・, *=DP・“Qyn‐
, (gn-an) (21) CQ・1*=CO. “Q
xn-, (hn-bn) (22) Do・, *=DQ
·, -Qyn-, (hn-bn) (23) In these equations, the value of tap gain with * indicates the corrected tap gain value, and Q is a correction coefficient.

Although an and bn are transmission data, these correction methods can be easily realized by using estimated data or a reference signal prepared in advance on the receiving side. The principle of the present invention has been described above, and when the present invention is implemented as a device, the number of multiplications is the same for the same number of taps compared to the conventional automatic equalizer using postprime operations for quadrature amplitude modulation. It has a feature that it can be realized on almost the same scale, and that distortion due to device errors in the modulator and demodulator can be equalized with distortion in the line, so it has high practical value.

Embodiments of the present invention will be described below with reference to the drawings.

The figure is a block diagram showing the entire automatic equalizer of the present invention, and in the figure, sample values for each signal interval of the output of two mouse synchronous detectors are input from terminals 1 and 2. The signal input from terminal 1 is sent to delay element 3-1, which has the same delay time as the signal interval T.
and 3-2 are connected in series, and the signal input from terminal 2 is connected in series to delay elements 4-1 and 4-2. Output lines 31-1 and 3 are connected to connection lines connecting each delay element.
0-1-2, 31-1, 31-2, 32-1 and 32-2
are drawn out, and two variable attenuators act on each lead wire. That is, the lead line 30-
1, the tap gains of the tap gain circuits 17-1 and 17-2 are applied by the balancers 5-1 and 5-2,
Similarly, lead lines 30-2, 31-1, 31-2, 32
-1 and 32-2 tap gain circuits 17-3, 17
-4, 19-1, 19-2, 19-3, 19-4, 21
The tap gains of -1, 21-2, 21-3 and 21-4 are multipliers 5-3, 5-4, 6-1, 6-2, 6-3,
6-4, 7-1, 7-2, 7-3 and 7-4 act to form a variable attenuator. Among the outputs of these multipliers, multipliers 5-1, 5-3, 6-1, 6-3,
The outputs of 7-1 and 7-3 are summed by a summator 8--, which outputs an equalized signal to a line 10-1, and multipliers 5-2,
The outputs of 6-4, 6-2, 6-4, 7-2 and 7-4 are summed by a summator 8-2, and another equalized signal is output to line 10-2. A method for automatically modifying the tap gain in this equalizer will be described below. Railroad 10
-1 and 10-2 are equalized signals output to terminals 9-1 and 10-2.
1 and 9-2 and outputs an error signal to lines 11-1 and 11-2. The reference signals inputted from the terminals 9-1 and 9-2 may be estimated values of data obtained from the equalized signal, or the receiving side may have the same sequence as the transmission training sequence and output it in advance. The error signal output to the line 11-1 is multiplied by the signal drawn from the lead line 30-1 by the multiplier 13-1 and by the signal drawn from the lead line 30-2 by the multiplier 13-3. The error signal output to 11-2 is combined with the signal drawn out from the lead line 30-1 by the multiplier 13-2 and the signal drawn out from the lead line 30-1 by the multiplier 13-4.
The output signal of each multiplier is multiplied by the signals drawn from the multipliers 16-1, 16-3, 16.
-2 and 16-4, and the output is added to the current value in tap gain circuits 17-1, 17-3, 17-2 and 17-4 and stored. Similarly, the signals output to lines 11-1 and 11-2 are transmitted to multiplier 1.
4-1, 14-2, 14-3, 14-4, 15-1, 1
5-2, 15-3 and 15-4, the signal is multiplied by the signals drawn from the lead wires 31-1, 31-2, 32-1, 32-2, etc., and the attenuators 18-1, 18-2, 18
- 3, 18 - 4, 20 - 1, 20 - 2, 20 - 3 and 20 - 4 multiplied by a constant, tap gain circuit 19 - 1, 1
9-2, 19-3, 19-4, 21-1, 21-2, 2
The tap gain value is updated in steps 1-3 and 21-4. All the operations described here are performed once for each new signal coming in from terminals 1 and 2, and in this manner the automatic equalizer can be successively set to its optimum state. Although this embodiment describes an example of taps, the number of taps may generally be any number. Although the present invention uses a sampled input signal, it can also be realized by a method in which a continuous signal is generally input and the equalized output is sampled.

[Brief explanation of the drawing]

The figure is a block diagram showing an embodiment of the present invention, and 3-1,
3-2, 4-1 and 4-2 are delay elements, 5-1, 5-2, 5-3, 5-4, 6-1, 6-2, 6-3, 6-4
,7-1,7-2,7-3,7-4,13-1,13-
2,13-3,13-4,14-1,14-2,14-
3, 14-4, 15-1, 15-2, 15-3 and 1
5-4 is a multiplier, 17-1, 17-2, 17-3, 17
-4, 19-1, 19-2, 19-3, 19-4, 21
-1, 21-2, 21-3 and 21-4 are tap gain circuits, 16-1, 16-2, 16-3, 16-4, 1
8-1, 18-2, 18-3, 18-4, 20-1, 2
0-2, 20-13 and 20-4 are attenuators, and 8-1 and 8-2 are summators.

Claims (1)

[Claims] 1 In a demodulator for data transmission using orthogonal amplitude modulation, two delay lines input the two baseband signals demodulated by the two-axis synchronous detector, and a constant delay line drawn from the first delay line. For a set of multiple signals given different delays of integral multiples of time, and for a plurality of signals given different delays of integral multiples of a constant time drawn from the second delay line, a set of two different variable attenuators that act on each delayed signal, a summator for one of the two different variable attenuators that acts on each of the delayed signals, and a summator that calculates the sum of the other; For the automatic equalizer that obtains the two equalized signals respectively output from the above summator, the attenuation amounts of all the above variable attenuators are calculated based on the above two baseband signals, a separately prepared reference signal, and the corresponding one. It is equipped with means for correction by using two error signals obtained by the difference between the equalized signals of An automatic equalizer characterized by: