JPS6282827A - Method and device for removing inter-code interference due to feedback of decision - Google Patents

Abstract

PURPOSE:To simplify control and to reduce the size of hardware by forming plural sample holding circuits, subtractors, a selector, a pattern checking circuit, and adaptive filter, and so on. CONSTITUTION:A receiving signal inputted to an input terminal 1 is supplied to the subtractor 2 and a false inter-code interference generated by the adaptor filter 15 is subtracted from the receiving signal by the subtractor 15 to obtain a difference signal (a receiving signal including a residue inter-code interference). The difference signal is supplied to a deciding unit 3, a block consisting of cascade connection of the sample holding circuit 81-8i and a subtractor 9 and the output of the deciding unit 3 is supplied to an output terminal 4, a pattern checking circuit 11 and the adaptive filter 15. In this case, a value obtained by subtracting a sample value obtained before iTsec from the current sample value is obtained from the output of the subtractor 9 and the residual inter-code interference component is exactly extracted. Consequently, control can be simplified without requiring complex control and the size of the hardware can be reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an intersymbol interference removal method and apparatus using decision feedback used to remove intersymbol interference that occurs during waveform transmission.

(Prior Art) A decision feedback isoelectric device is known as a known technique for removing intersymbol interference that occurs during waveform transmission.

(IBEB TRANSACTION)
SON COMMUNICATJON8) Volume 32 No. 3,
1984.

Pages 258-266. ) Decision feedback type equalization a with an adaptive type (ada 1
By generating pseudo intersymbol interference corresponding to the received data sequence ζ using a filter, the intersymbol interference that is received while the waveform is being transmitted through the transmission path is suppressed. At this time.

Each coefficient of the adaptive filter is successively modified by correlating the residual intersymbol interference with the determination result of the received signal.

When modifying the coefficients in the decision feedback equalizer I,
Included in the difference signal obtained by subtracting pseudo intersymbol interference from the received signal containing intersymbol interference is t'! ? 6. If the residual intersymbol interference cannot be correctly horizontally compared, a problem arises in that adaptive operation becomes impossible. For example, when a binary code such as a biker's code is used as a transmission path code, due to the nature of the binary code, there is no section where the received signal level is zero, and only intersymbol interference is extracted independently. This will result in the above-mentioned problem. Therefore, a conventional technique for solving this problem will be first described.

FIG. 10 shows a conventional example of a 1-decision feedback type equalizer. Here, the enclosure shown in FIG. 1O is connected to the transmitting side via a transmission line. Here, for simplicity, explanation will be given assuming baseband transmission.

In FIG. 10, a received signal containing intersymbol interference is supplied to input terminal 1 from the transmission path, and the received signal is input to 'WL'n unit 2, and the received signal supplied to input terminal l by subtracter 2. A difference signal (=received signal including residual intersymbol interference, [residual intersymbol interference] = [intersymbol interference] - [pseudo intersymbol interference]) is obtained by subtracting pseudo intersymbol interference from 1. It is supplied to a subtracter 6. The result determined by the determiner 3 becomes a binary data series, which is supplied to the output terminal 4, the AGC 7, and the adder 17'' Eve filter 5I.The output of the adaptive filter 5 is supplied to the subtracter 2. vessel 6 and AGC
The closed loop circuit consisting of n, AC+C7 operates to accurately extract only the residual intersymbol interference in the difference signal input to the fgL calculator 6. This is achieved by multiplication to generate a received signal that does not contain residual intersymbol interference. The received signal generated by the AGC 7 is subtracted from the difference signal at the output of the attenuator 2 in a subtracter 6. The output of subtractor 6 is A
It is fed back to GC7 and controls the gain of AGC7. At the same time, the output of the 'DFL calculator 6 is supplied to the adaptive filter 5 and used for coefficient updating. Subtractor 2゜Judgment device 3
．． From the adaptive filter 5 to the closed loop circuit,
It operates to cancel intersymbol interference in the received signal applied to input terminal I. This is achieved by the adaptive e-filter 5 generating pseudo intersymbol interference. Therefore, the adaptive filter 5 will be explained in detail.

Figure 1) shows a detailed block diagram of the adaptive filter 5 in Figure 10.6 Input signals 106 and 107 in Figure 1) are the output signals of the determiner 3 in Figure 10, respectively. It corresponds to the binary data series and the output signal of the M calculator 6.

1) The output signal 108 in FIG. 1 corresponds to the output signal of the adaptive filter 5 in FIG.

Input signal 106, delay element 100□.

Multipliers 101o, 101□, ...-101R-1
and coefficient generator 302o, 102, -=...-102R
-0 can be supplied. Delay element 100 giving a delay of 7 seconds
1. I(HJ2..., 100N/R-□Po, are connected in this order, and each can be realized by flip-frog. Here, H and R are positive integers, and R2 is It is a divisor of N. Also, the data period of the input signal 106 is HT seconds.The delay element 100m (m=I +2
．． . . . N/R-1), and a multiplier +01. .

101). 4-1..., +01J+R-□ and coefficient generation J 02 .

102, +1, . . . , 102J+R-1. However, j=mXR. In the multiplier 101, 1
+R to 01..., 101 k+N-R (m=Q
, l, . . . , R-1), each coefficient which is the output of the coefficient generators 102, 102, . Calculator 103. I is input and added. The outputs of the eight adders 103o, H)3□..., 103R-□ serve as input contacts of the switch 1040. switch 104
It is a multi-contact switch with a period of flT seconds, and includes adders 103o, 103□, ...-1)03R.
- Select and output the output of □ every T/R second in this order.

It is assumed that the output signal is 108. Pseudo intersymbol interference occurs in the output signal 108, and pseudo intersymbol interference occurs every i'/R, seconds.1. Ru. a is an interpolation constant (interpolation factor)
In order to eliminate intersymbol interference within a required signal band, n is an integer of 2 or more. On the other hand, switch 1
Switch 105 operates in synchronization with H switch 104 and input/output are reversed. That is, switch 10
5 Performs the function of sequentially distributing H-jinkai-g-107 to 8 contact points every T/R seconds. Each contact output of switch 105 is fed to a coefficient generator present in the signal path input to the contact corresponding to switch 104 which operates synchronously. Next, the coefficient generation circuit will be explained in detail.

FIG. 12 shows the coefficient generator 1021 (!A'=0) in FIG.
．． 1. . . . , N-1). The input signal 200 in FIG. 12 is the input signal 106 in FIG.・・・・・・
..., corresponds to the output signal of J 00 N/R-1. Also.

The input signal 201 in FIG. 12 corresponds to the contact output of the switch 105 in FIG. 1). Furthermore, the 12th
The output signal 203 in FIG. 1) is the coefficient generator 10 in FIG.
21! It corresponds to the output of In Figure 12 1,
Input signals 200 and 201 are supplied to multiplier 204, and the multiplication result becomes one input of adder 2oG.

The output of the adder 205 is fed back through a delay element 206 of T seconds, and the update of the coefficients performed every second is multiplied by the input signals 200 and 20 supplied to the adder 204.
This is achieved by adding a correlation value of 1 to the coefficient value of one sample before. Output signal 203 is the coefficient.

(Problems to be Solved by the Invention) As described above, one of the subtracters 2 is applied to the pseudo intersymbol interference generated by the adder 1 deep filter 5 in FIG. becomes the input. In the yR calculator 2, a difference signal obtained by subtracting the pseudo intersymbol interference from the received signal supplied from the input terminal l (=received signal including residual intersymbol interference, however).

[Residual intersymbol interference] = [Intersymbol interference] - [Pseudo intersymbol interference]) is obtained and the non-determiner 3. It is supplied to a subtracter 6. −
The closed loop circuit consisting of the AGC 7 and the subtracter 6 operates to accurately extract only the residual intersymbol interference in the difference signal supplied to the subtracter 6. The Tagata signal from the determiner 30 supplied to the AGC 7 is multiplied by r and input to the subtracter 6. Here, r is a positive number. The signal supplied from the AGC 7 to the subtracter 6 is subjected to tfL calculation from the difference signal supplied to the subtracter 6.

It is fed back to the AGC 7 as a control signal. In AGC7,
Using the signal fed back from the 1ftS unit 6, r is corrected so that the output of the subtracter 6 is equal to the residual intersymbol interference. That is, the -AGC 7 pseudo-generates a signal other than the residual intersymbol interference included in the difference signal. The output of the subtracter 6 is supplied as an error to the adaptive filter 5,
Coefficient update C is used. Here, the adaptive
In order for the filter 5 to perform an adaptive operation, it is necessary to correctly supply residual intersymbol interference to the adder 17'' filter 5.However, in the difference signal that is the output signal of the subtractor 2, there is no residual intersymbol interference other than the residual intersymbol interference. Therefore, if it is assumed that the output signal of the subtracter 2 is directly supplied to the adaptive filter 5, the residual intersymbol interference cannot be accurately obtained.

The adaptive ability of the adaptive filter 5 will be lost. Therefore, conventionally, as shown in FIG. 10, a subtracter 6. With AGC7%: r+I]L, “1: Subtractor 2
A method has been used in which the adaptive operation of the adaptive filter 5 is guaranteed by subtracting signals other than pseudo residual intersymbol interference from the difference signal that is the output signal of the filter.

In this method, an AGC 7 uses a binary data sequence that is a determination result of a received signal to generate a received signal that does not include intersymbol interference, and a subtracter 6 subtracts it from the difference signal. A
The residual intersymbol interference component is obtained by the GC 7 and the subtracter 6, and the adaptive operation of the aggressive filter 5 is guaranteed. However, with the conventional control method, fl-AGC7
In addition, in order to obtain a sufficient degree of intersymbol interference suppression, complicated control is required to maintain the received signal, which does not include intersymbol interference, supplied from the AGC 7 to the subtracter 6 at a desired level. The disadvantage is that the size of the software becomes large.

An object of the present invention is to provide an intersymbol interference removal method and apparatus N using decision feedback that is simple and requires small hardware scale.

(Means for Solving the Problems) According to the present invention, there is provided a method for removing intersymbol interference occurring during waveform transmission using pseudo intersymbol interference generated by an adaptive filter.

After subtracting the pseudo intersymbol interference from the received signal containing intersymbol interference to obtain a difference signal, add or subtract the difference 1^ signal and a delayed signal obtained by delaying the difference signal to obtain the residual intersymbol interference. determining the interference and updating the coefficients of the adaptive filter using the residual intersymbol interference only when a specific pattern of the data sequence obtained by demodulating the difference signal is detected;
An intersymbol interference removal method using decision feedback is obtained, which is characterized in that upon updating, the coefficients of the aggressive filter used immediately before the update and the coefficients used before by the delay time of the delayed signal are simultaneously updated. .

Further, according to the present invention, when removing intersymbol interference that occurs during waveform transmission, the received signal is divided into a determiner that generates demodulated data, and receives the demodulated data and error signal supplied from the determiner. - an adaptive filter for generating an intersymbol interference replica according to A, a subtracter for obtaining the best result between a received signal containing the intersymbol interference and the intersymbol interference replica, and an output of the subtractor; a plurality of cascade-connected sample-and-hold circuits for sampling and holding; and an arithmetic unit for obtaining the difference or sum between the output of the subtracter and the output of the cascade-connected sample-and-hold circuits; A selector that selects either the output of the arithmetic unit or zero, and a pattern check circuit that receives the demodulated data and generates a signal that controls the selector, and uses the output of the selector as the error signal to control the aggressive Returning to the filter, the coefficients of the aggressive filter are updated by simultaneously updating the previously used coefficients and the coefficients used before the delay time of the cascaded sample-and-hold circuits. An intersymbol interference canceling device using decision feedback characterized by the following is obtained.

(Operation) Unlike the conventional method of multiplying the output of the 1-determiner by a constant to generate a received signal that does not contain residual intersymbol interference and subtracting it from the difference signal, the present invention focuses on the characteristics of the eye pattern of the received signal. The residual intersymbol interference was intercepted so that it could be extracted accurately with a certain probability.

That is, according to the characteristics of the eye pattern of the received signal of the transmission path code including the binary code system, the current sample value and 1T seconds (i
is a positive integer) The minimum probability that the previous sample values are approximately the same value or that the respective absolute values are approximately the same value with opposite polarity takes a certain positive value that is not zero.

Therefore, the difference signal (=received signal containing residual intersymbol interference)
By calculating the difference or sum of the current sample value and the sample value iT seconds ago for It becomes interference itself. Therefore, the ability to accurately detect residual intersymbol interference with a certain positive probability that it is not zero)
If the difference or sum thereof is used as an error signal, the adaptive operation of the adaptive φ filter is guaranteed. Furthermore, since the sum or difference between the current difference signal and the difference signal iT seconds ago is used as an error signal, the error signal depends on both the coefficient used immediately before and the coefficient used iT seconds ago. . Therefore, the convergence time of the coefficient can be shortened by simultaneously updating the coefficient used just before and the coefficient used iT seconds ago (Example) Next, referring to the drawings, the present invention will be explained. will be explained in detail.

FIG. 1 is a block diagram showing one embodiment of the present invention. In this figure, it is assumed that functional blocks given the same reference numbers as in FIG. 1O have the same functions as in FIG. 10. 1st
The difference between the figure and Fig. 10 is the sample and hold circuit 8□,
82. . . ., a block consisting of 8p (p=iXR) cascade connections, a subtracter 9, a selector 10. The pattern check circuit 1) is a part consisting of an adaptive filter 15.

The other configurations are exactly the same as in FIG. 10. However, for simplicity, it is assumed that the time required for sampling ζ of the sample-and-hold circuit can be ignored. Before explaining the details of this circuit, a brief description of the overall configuration will be given. The received signal input to the input terminal l is supplied to the subtracter 2. The difference signal obtained by subtracting the pseudo intersymbol interference generated by the adaptive filter 15 in the subtracter 2 (=received signal containing residual intersymbol interference) is 1 determiner 3, sample and hold circuit 8□, 88.・・・・・・、8. and a subtractor 9. The output of the determiner 3 is supplied to an output terminal 4, a pattern check circuit 1), and an adaptive filter 151. Adaptive filter 15. Subtractor 2. Sample 8...
,,.

- Hold circuit 80, 2. ．． 8. A block consisting of a cascade of subtractors 9. A closed loop circuit consisting of a selector IO realizes the adaptive operation of the adder 1 tape filter 15, and a pattern number check circuit 1) controls the closed loop 1 circuit so as to selectively update the coefficients to 0.

Regarding the configuration of the adaptive filter 15, see section 10.
Although the circuit configurations are basically the same as those explained in the figures, the circuit configurations in FIGS. 1) and 12 need to be slightly modified. This will be discussed later. Next, the output of subtractor 9 and the output of subtractor 2
The relationship with the residual intersymbol interference in the difference signal that is the output of is described in detail, but before that, the transmission code will be described.

Figure 2 (al, (bl) shows typical examples of binary codes; (a) shows the biphase code, and (b) shows the MSK (minimum shift keying) code H/l.
/ Indicates each figure. As shown in the second factor (alg), a pulse waveform with inverted polarity is assigned to the data of 10" and 1" in the biphase code.

In contrast, both pulses have the characteristic that the polarity is reversed at the center of 1 bit @ T seconds, and the positive and negative are balanced at the bit circle.

Figure 2 (b) ζ As shown, four types of pulse waveforms are prepared in the 1M5K code, that is, 10# mode with reversed polarity for each 10' and 1' data I).
! : '1'mode's two-wave type ) and Luss waveform '1) means. The mode transitions of these two relatives are indicated by the thick arrows in Figure 2 (g) and are determined by the mode in front of the current mode symbol. This MSK code is characterized by the fact that the polarity always inverts at the boundary 5'H of the exit symbol waveform. In addition, in the MSN code, fl for '1''
Although positive and negative are balanced within the symbol, '0''
There is no balance between positive and negative. However, it is clear from the direction of the thick arrow indicating the mode transition in Figure 2(b) that
As shown, if there is an even number of ``0'' in a continuous data series, the positive and negative values are balanced and can be almost ignored as DC components.The transmission line code shown in Figure 2 (a) is The signal is transmitted to the input terminal of FIG. 1 after being subjected to intersymbol interference.

Figure 3 shows an example of the received signal eye pattern when the transmission code shown in Figure 2 is adopted.
fl and (corresponding to bl, respectively, are the eye patterns of the bi-phase code and MSK code in Fig. 2.) As shown in the figure, the received signal eye pattern is rounded by removing high-frequency components. Originally, the received signal eye pattern contains intersymbol interference components, but to simplify the explanation, the eye pattern shown in the figure is ideal and does not contain intersymbol interference. shall be.

Now, the received signal A of the MSK code shown in FIG. 3(b) is
Pay attention to patterns. Depending on the characteristics of the received (i) eye pattern, the current sample value and the sample value before iT (i is a positive integer) are almost the same value, or their absolute values are almost the same with opposite polarities. So.

Is the sandal value from iT seconds ago (i is a positive integer) delayed by iT seconds and then subtracted from the current sandal value?

Alternatively, the received signal can be canceled by adding O. In Fig. 4 (at, (b) for i = 2 -
This shows how the received signal is canceled out, and the three waveforms are from the right: the symbol waveforms for the 1st rule, the T second cut, and the 2T sand. 1444 (a) In the case of the continuous pattern of waveforms shown in Fig. 4, the current symbol waveform and the symbol waveform 2T seconds ago are the same, and in the case of the continuous pattern of waveforms shown in Fig. The waveforms have the same absolute value.Also, other than the continuous pattern of waveforms shown in Figure 4, or the waveform pattern with the same absolute value and opposite polarity as the combination shown in the same figure, the current symbol Waveform and 2
T8) It is also clear that the absolute value of each symbol cannot be the same value ζ with the same or opposite polarity of the previous symbol waveform.

In the case of the waveform pattern shown in the fourth factor, it is easy to see that the current sample value and the sample value 2T seconds ago are the same value, or have opposite polarities and have the same absolute value. By subtracting the sample value delayed by 2T seconds from the undelayed sample value or adding it to the undelayed sample value m, the received signal component is canceled out and the output becomes zero. For the moment, we will discuss only the case of subtracting a sample value delayed by 2T seconds from a sample value that is not delayed, and we will discuss the case of adding a sample value delayed by 2T seconds to a sample value that is not delayed. will be mentioned later. Now, considering a non-ideal case, the received signal includes a residual intersymbol interference component. Considering the residual intersymbol interference component, since there is no correlation between the current residual intersymbol interference value and the residual intersymbol interference value 2T seconds ago, the residual intersymbol interference value 2T seconds ago is considered to be random noise. be able to. The amplitude distribution of the residual intersymbol interference value 2T seconds ago has positive and negative symmetry, and the amplitude d is Idl
The probability that δ becomes less than δ (however, δ less than 0) is not zero, but takes a certain positive value. Therefore, it can be seen that the probability that the output signal of the subtracter 9 contains accurate residual intersymbol interference takes a certain positive value that is not zero. Furthermore, the magnitude of residual intersymbol interference is generally sufficiently small relative to the received signal. Therefore, the waveform shown in FIG. 4 can be regarded as the received signal waveform, even if it is not ideal.

Note that only the residual intersymbol interference can be accurately extracted by subtracting the sample value delayed by 2 samples from the undelayed Yang 7 value, because the received signal has the pattern shown in Figure 4 (al). Since the remaining intersymbol interference blocks cannot be accurately resolved, the Ada 7 table in Figure 1)
, Ilta 150 system # is not performed correctly. Therefore, in order to warn the adaptive filter 15 to stop (here, the continuous pattern of the received signal waveform is checked according to FIG. 4, and when it does not match the pattern in the same figure, Ic stops updating the coefficients of the adaptive filter 15. For the continuous waveform pattern shown in Figure 4, refer to Figure 2 fb), the data symbol is '000' or 'fl'.
Since this is the case of "J I 1", the pattern check detects a continuous pattern of "000" and 1) 1) in the data signal. In addition, in Fig. 4, the value obtained by subtracting the previous sample value of 2T seconds (data period is 1 second) from the current sample value was used, but it is also possible to detect a predetermined waveform pattern. It is also easy to use the adaptive filter 15 to accurately represent only the residual intersymbol interference with a certain probability. Next, the meaning of accurate extraction of residual intersymbol interference in the adaptive operation of the decision feedback isoelectric device will be explained with reference to FIG.

FIG. 1 In the embodiment shown, reference numerals 80, 82.
..., 8r) is a sunfur hold circuit, reference number 9 is a subtracter, reference number 10 is a selector, and number 1 is a sash
) is a pattern-check circuit, number 15 is an adaptive filter. Here, in order for the adaptive filter 15 to perform an adaptive operation.

The residual intersymbol interference component (=[intersymbol interference] - [pseudo intersymbol interference]) contained in the difference signal (=received signal containing residual intersymbol interference) that is the output of the subtracter 2 is accurately adaptive. The necessity for supplying the filter 15 has already been described. In FIG. 1, the sample hold circuit 8□,
The subtracter 9 and the block consisting of the cascade connection of 8□, ..., 8p are added for the purpose of satisfying this condition, and the output of the subtracter 9 is i T
The value obtained by subtracting the sample value from seconds ago appears.

As described in the explanation of FIG. 4, the residual intersymbol interference component in the difference signal that is the output of the subtractor 2 is subtracted by the combination of the received signal waveform pattern shown in FIG. 4(a). It is clear that the output of the device 9 is accurately extracted with a certain probability. On the other hand, when considering the residual inter-symbol interference component included in the output of the subtracter 9 using i;2 as an example, the current residual inter-symbol interference value and the residual inter-symbol interference value 2T seconds ago are uncorrelated. Therefore, the value of the residual intersymbol interference 2T seconds ago can be considered to be rank noise.The residual symbol d 2T seconds ago
The probability that Idl≦δ (however, 04δ) is not zero but takes a certain positive value. Therefore, it can be seen that the probability that the output signal of the subtractor 9 contains accurate residual intersymbol interference takes a certain positive value that is not zero. Therefore, if the output of the subtractor 9 is used to control the adaptive fist filter 15, the adaptive operation of the adaptive filter 15 is guaranteed. In addition, up to 1, refer to the sample hold directions 8□ and 8□ in Figure 1.

......, 8. A block consisting of a continuous connection of is 2T.
The case where a delay of seconds (i
A similar effect can be obtained even if the value is a positive integer).

Next, the operation of the pattern check circuit that controls the output signal of the subtracter 9 and the selector 10 will be explained.

The pattern check circuit detects the appearance of the waveform pattern of the received signal shown in FIG.

In other cases, the purpose is to stop updating the coefficients of the tag debug filter 15, and can be realized by the circuit shown in FIG. The input signal 51 of FIG. 5 and the data signal ζ of the determiner 3 of FIG. 1 are equal to the mode signal. In addition, in FIG. 1, 1 determiner 3 and pattern check circuit 1). and a judger adversarial sum circuit (XNOI(,) 55
Therefore, it is realized by taking the negative exclusive OR of the value delayed by 1) second with the current signal 53 using the XN0R55.
This is one input of 59. Similarly, the input signal 52
and 1) take the negative exclusive @reason of the second-delayed value and use the output as the other input of AND59. AND59 performs a logical product of the coincidence output of the Cheetai bar code and the coincidence output of the mode signal, and outputs the output signal 60. The output signal 60 is equal to the signal supplied to the selector 101 from the pattern check circuit 1) of FIG. Note that this is realized by connecting delay elements 53 and 56 tri-flip-flops that provide a delay of 17 seconds in i controlled columns.

The selector 10 receives a control signal from the pattern check circuit 1), selects the output signal of the subtracter 9 or zero according to the control signal, and supplies the selected signal to the adaptive filter 15. This is when the selector 10 supplies the output signal of the subtracter 9 to the adder 1y'' filter 15.The selector 10 and the pattern check circuit 1) select 1 only when residual intersymbol interference is accurately detected. The coefficients are updated.

The pseudo intersymbol interference generated by the adaptive filter 15 of FIG. , subtractor 2
Then, the input signal of input terminal l is the difference signal obtained by subtracting pseudo inter-symbol interference from the received signal (=received signal including residual inter-symbol interference, [residual inter-symbol interference] = [inter-symbol interference] - [pseudo-symbol interference]) interference]) is obtained, and the determiner 3. Delay element 8. A 0 selector 10 supplied to the subtracter 9 selects the output signal of the # subtracter 9 or zero according to the output signal of the pattern check circuit and supplies it to the adaptive filter 15. The result determined by the determiner 3 is supplied to the adaptive filter 15 and appears at the output terminal 4 at the same time. Coefficient correction is performed using the output signal of the adaptive filter j5t e-selector IO. Next, we will discuss the differences between the adaptive filter 15 shown in Fig. 1 and the aggressive filter 5 shown in the %IO diagram using Figs. 6, 7, 1), and 12. explain.

FIG. 6 is a block diagram of the adaptive filter 15 shown in FIG. 1, and corresponds to FIG. 1). In the block diagram of the adaptive filter shown in FIG.
02 j, 102 J + i...-102
j+R-1 (j=mXR, m=1.2;・-, N7Qt
-1), the output signals of not only the delay element 100m but also the delay element 100m+i are simultaneously supplied. Therefore, a total of (N/a-1+i) delay elements are required. By making this connection, l! It is possible to simultaneously update the coefficients of the filter used immediately before the update and the coefficients used before by the time delayed by the sample-and-hold circuit shown in Figure 1. In response to the change in the structure of the adaptive filter as shown in Figure 6,
The coefficient generator 102.2 is also modified as shown in FIG. FIG. 7 corresponds to FIG. 12, and in FIG. 12 as well, the signal 200 supplied to the multiplier 204 is 1 multiplier 204 and 2002, respectively. I have changed. As the output signal H2O01 of the delay element 100m, the output signal of the delay element 100m+i is set to 200. , n multipliers 204□, 20
4. supplied to.

0, which is multiplied by the signal 201 from the switch +05 in FIG.
It becomes an adder of 205 on 5. As explained above.

By changing the configuration of the aggressive filter and simultaneously updating the coefficients of the filter used before the update and the coefficients used previously by a time delayed by the sample-and-hold circuit of the first prisoner.

The convergence time of coefficients can be shortened.

In addition, in FIG. 1, the sample hold circuit 80.8
．． It was assumed that the time required for sampling 8p was negligible, but if this assumption did not hold, the number of sample-and-hold circuits would be ((ltT/(T-Rδ)
), )+1) It is good to have at least 1). This C# this.

δ is the time required for sampling by the sample and hold circuit,
May [X] be the largest integer not exceeding X.

It is always equal to the sample-and-hold period of the valley sample-and-hold circuit by T/about. Now, the phases of adjacent sample φ hold circuits are shifted by ('1'/)t-δ).

At this time, one sample-and-hold circuit holds the sample value for (T/1 mode δ) seconds minus the zero time required for sampling ζ. The arrangement is R,=4,δ=
For 'l'/32, it is sufficient to prepare 5 or more sample-and-hold circuits, and the sJ circuit has 5 sample-and-hold circuits connected in series.

The total hold time is 31T/32. 5I for this
l! This is the maximum hold time that can be achieved by connecting 1 sample-and-hold circuits in series. Total hold time T
To achieve this, it is a good idea to shift the sample phases of adjacent sample hold circuits by T15 in order.
Even if the sample phases of the four sample-and-hold circuits are sequentially shifted by 7T/32, and the remaining one is shifted by 4T/32 from the sample phase of the previous sample-and-hold circuit, the overall hold time can be reduced to T. It is. in this way,
The overall hold time can be reduced to T by appropriately adjusting the sample phases of adjacent sample-and-hold circuits. Similarly, for any δ smaller than +T/R, it is possible to obtain any hold time by connecting a sufficient number of sample hold circuits in series and selecting the sample phase appropriately. Therefore, in general, even if the time required for sampling I cannot be ignored, 'V
Any hold time that is an integer multiple of can be obtained.

FIG. 8 is a block diagram showing another embodiment of the present invention.

In the figure, it is assumed that functional blocks given the same reference numbers as in FIG. 1 have the same functions as in FIG. The difference between FIG. 8 and FIG. 1 is that the subtracter 9 in FIG. 1 is replaced with an adder 12, and the other parts are exactly the same. Therefore, regarding the difference signal that is the output of subtractor 2 in FIG. 8, the sum of the current difference signal value and the difference signal value i 141 seconds ago appears at the output of the adder, It will be used to control the adaptive filter 15.

! To explain the case where =2 as an example, the waveform pattern shown in FIG. 4(b), that is, the data signal is '010'' +0
1", the residual intersymbol interference component in the difference signal output from the subtracter 2 can be accurately expressed at the output of the adder 12 for the same reason as explained in FIG. 4 (al). It is clear that the adder 1
If the adaptive filter 15 is controlled using the output of 2, the adaptive operation of the adaptive filter 15 is guaranteed. The waveform pattern shown in FIG. 4(a) can be detected by the pattern check circuit shown in FIG. 5, but in order to detect the waveform pattern shown in FIG. 4(b),
X to detect coincidence of mode signals in the circuit shown in the figure
Exclusive OR may be used instead of NOR, 58 to detect mismatch of mode signals. Incidentally, the sample hold circuit 83 in FIG. ......, we have explained the case where a delay of 2T seconds is given to a block consisting of 8p cascade connections as an example, but as mentioned in the explanation of Fig. 4, the delay + Li is 1) seconds. (i is a positive integer) also has the same effect.

The present invention has been explained above in detail based on examples, but
When MSK code is adopted, the pulse waveforms for 20# and 1)# are different, and '? ! The number of grooves in the adapter dip filter 15 is slightly different from the case shown in FIG. 6 due to the two reasons of having rho mode and 1 mode. That is, ′
In response to the different pulse waveforms of 0# and ``''j#, it is necessary to prepare two types of tap coefficients and update them individually, and the coefficients are distinguished by the mode signal received from the 1 determiner 3. Furthermore, in the explanation up to now, sample and hold circuits 80, 83, . . .
....

Assuming that the delay amount of a block consisting of 8, cascaded connections is 2T seconds or 1) seconds (i is a positive integer), -C is assumed.

Needless to say, n in the vicinity of 1) seconds is practically sufficient.

The present invention has been explained in detail using the 1M8 code as an example, but as a transmission line code, for example, the code shown in FIG. 2(a)
The biphase code shown in can be used. What is different when using a bi-phase code than when using an MSK code is that the received signal has an eye turn. Therefore, the pattern-checking method is also unique to biphase codes.

Referring to FIG. 3 1al, at the point where the symbol waveform switches to the received signal eye pattern of the biphase code, '
It takes two values: 4 and less than zero. Therefore, it is necessary to check a pattern different from the received signal eye pattern of the MSK code, which always exceeds the value of O at the point where the symbol waveform changes. FIG. 9 shows a pattern φ check waveform corresponding to the biphase code. In the same figure (al is when controlling the adaptive filter 15 by the value obtained by subtracting the sample value 2T seconds ago from the current sample value,
FIG. 4B corresponds to the case where the adaptive filter 15 is controlled by the value obtained by adding the sample value 2T seconds ago to the current sample value. In FIG. 3, 1al and (b), five consecutive waveforms, in order from the right, are symbol waveforms of T seconds after the present, the present, T seconds ago, 2T seconds ago, and 3T seconds ago. Since it is impossible to know in advance the symbol waveform T seconds from now, wait until the symbol waveform T seconds from now is determined -) 1. Update the coefficients. In the case of bi-phase codes, the increased symbol waveform differs depending on the symbol waveform of one symbol before and after 1, so the value obtained by subtracting the sample value 2T seconds ago from the current sample value, or the value obtained by subtracting the sample value 2T seconds ago from the current sample value, When controlling the adaptive filter 15 by the value obtained by adding the first sample, five symbol waveforms are used, two symbols each before and after the current symbol waveform, and two symbols before and after the symbol waveform 2T seconds ago, for a total of 6 symbols. , when using the symbol 2T ago, the previous l symbol of the current symbol waveform and 2T
The l symbol after the symbol waveform of seconds ago is common, so the total is 5
The continuous pattern of the received signal waveform over symbol t must be checked. Figure 9(a) deni-101
The same figure shows the continuous pattern of '01'' and 'ooooo' (in bJ, the continuous pattern of '01)00' and '1)00t'. In the case of bi-phase code, the pattern check circuit shown in Fig. The five continuous waveforms shown in are detected, and only when this waveform is detected.''
If a logic circuit that outputs 1' is constructed, it can be used simultaneously with the pattern check circuit for the MSK code shown in FIG. However, in the case of a biphase code, the input signal to the pattern check circuit is only a data signal.

In the above explanation, when the automatic filter 15 is controlled by the value obtained by subtracting the sample value 2T seconds ago from the current sample value, or by the value obtained by adding the sample value 2T seconds ago to the current sample value However, when using the sum or difference between the current bubble value and the sample value iT seconds ago.

The adaptive operation of the adder 1-tiff filter 15 is controlled by detecting waveform patterns of one symbol waveform before and after the two symbol waveforms focused on the pattern check circuit, and six symbol waveforms. Considering transmission path codes other than Kowara's codes in the same way, the received signal pattern corresponding to FIG. 4 is detected and the updating of the coefficients of the adaptive filter 5 is controlled.

It is clear that the residual intersymbol interference can be extracted accurately with a certain probability.

(Effects of the Invention) As described in detail above, according to the present invention, the difference or sum of the current value and the value iT seconds ago (where i is a positive integer and T is the data period) is calculated for the difference signal. As a result, the residual intersymbol interference component contained in the received signal can be accurately extracted with a certain positive value that is not zero. Therefore, using the above difference or sum, detect a continuous pattern of received signal shaping that roughly extracts the residual intersymbol interference component, and selectively update the coefficients to control the adder-e filter. According to
Adaptive behavior is guaranteed. Also, from the present invention E, i'
By combining a subtractor or an adder with a block consisting of a #1 series connection of several sample-and-hold circuits that provide a delay of 1' seconds, the above-mentioned adaptive operation can be guaranteed, so complex control is not necessary. It is possible to provide a method and device for removing intersymbol interference that is simple and requires small hardware scale and that does not require decision feedback. Furthermore, in rare cases, the present invention simultaneously updates the currently used coefficient and the coefficient that was used iT seconds ago, so that the coefficient convergence time can be shortened.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention;
)(at, (bl is a diagram explaining the transmission line code, FIG. 3(at, (tl is a diagram showing the eye pattern corresponding to the transmission line code in FIG. 2, FIG. 4(at, (bl is MSK
FIG. 5 is a diagram showing a received signal waveform pattern from which residual intersymbol interference is accurately extracted in a code, and FIG. 5 is a diagram showing the configuration of a pattern check circuit for an MSN code. FIG. 8 is a block diagram showing another embodiment of the present invention. FIGS. 9(a) and (bli) (If-Az code - received signal waveform in which residual intersymbol interference is accurately extracted) Pattern-i@10 is a block diagram showing a conventional example of intersymbol interference removal using decision feedback. FIG. 2 is a detailed block diagram of a coefficient generation circuit of FIG. In the figure, 1... Input terminal, 2...
...Subtractor, 3...Determiner, 4
......Output terminal, 8□, 82. ...
..., 8p sample/hold circuit, 9...
...Subtractor, 10...Selector 1)
1) Pattern check circuit, 15
...each represents an adaptive filter. Also, in Figure 8 1, I...
...Input terminal, 2...Subtractor,
3・・・・・・・・・Judgment device. 4...Output terminal, 8,8.・・・・・・
..., 8i... Sample-hold circuit, 10...- Selector, 1)... Pattern check circuit, ! 2...Adder. 15...adaptive filter, son+4
represent.

Claims (2)

[Claims] (1) In a method of removing intersymbol interference that occurs during waveform transmission using pseudo intersymbol interference generated by an adaptive filter, a difference signal is obtained by subtracting the pseudo intersymbol interference from a received signal that includes intersymbol interference. After obtaining the difference signal, the residual intersymbol interference is obtained by adding or subtracting the difference signal and a delayed signal obtained by delaying the difference signal, and when a specific pattern of the data sequence obtained by demodulating the difference signal is detected. only,
The coefficients of the adaptive filter are updated using the residual intersymbol interference, and in updating, the coefficients of the adaptive filter used immediately before the update and the coefficients used before the delay time of the delayed signal are simultaneously updated. An intersymbol interference removal method using decision feedback characterized by updating. (2) When removing intersymbol interference that occurs during waveform transmission, a determiner that receives the received signal and generates demodulated data, and an adaptive intersymbol interference that receives the demodulated data and error signal supplied from the determiner. an adaptive filter for generating a replica; a subtracter for obtaining a difference between the received signal containing the intersymbol interference and the intersymbol interference replica; and a subtracter for sampling and holding the output of the subtractor. a plurality of cascade-connected sample-and-hold circuits; an arithmetic unit for obtaining the difference or sum between the output of the subtracter and the output of the cascade-connected sample-and-hold circuits; a pattern check circuit that receives the demodulated data and generates a signal for controlling the selector; the output of the selector is fed back to the adaptive filter as the error signal; The coefficients of the adaptive filter are updated by simultaneously updating the coefficients used immediately before the update and the coefficients used before the delay time of the cascaded sample-and-hold circuits. Interference remover.