JPS646577B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to waveform distortion removal from a nonlinear transmission path including a traveling wave tube amplifier (hereinafter abbreviated as TWT).

Digital communications in the microwave band, whether satellite or terrestrial, will be required to be operated using a higher-density transmission method from the perspective of effective use of the frequency band.

i.e. 1979 International Conference
on Communications (ICC'79) Conference Record, pages 48, 4, 1 to 48, 4, 6
“Characteristics of a
High Capacity 16QAM Digital Radio
System on a Multipath Fading Channel”
Also in 1979 National
Telecommunications Conference (NTC'79) Conference Record 35, 4, 1 - 35, 4,
Distortion Analysis of 64 on page 3
As you can see in ``QAM'', multilevel quadrature amplitude modulation (QAM) will be used.At this time, the problem is nonlinear distortion of the transmitting amplifier (TWT), and this distortion distorts the QAM signal. The nonlinear distortion of TWT is
Although they differ slightly depending on the TWT, they form one category. In other words, amplitude saturation characteristics (AM/AM conversion) and output phase rotation θ corresponding to input level x
(x) Characterized by characteristics (AM/PM conversion).

It is therefore generally possible to compensate for this type of distortion to a considerable extent with relatively simple circuits. As a known example of such compensation, there is, for example, "Study of TWT nonlinear compensation using automatic tracking complex synthesis predistortion" in document CS78-201 of the Communication Systems Study Group of the Institute of Electronics and Communication Engineers. However, if the nonlinear distortion is subjected to other waveform distortions, these simple circuits cannot compensate. Furthermore, since this known example was developed for SSB in the microwave band, it is not very suitable for digital transmission.

An object of the present invention is to provide a nonlinear equalizer that removes distortion of a digital signal passed through a nonlinear transmission path in which a medium that causes waveform distortion is present before and after a nonlinear circuit.

That is, in this invention, a transmission signal passes through a first medium that causes a first waveform distortion, passes through a nonlinear circuit that outputs (X) for an input X, and further passes through a second medium that causes a second waveform distortion. In the nonlinear transmission path that passes through the medium and reaches the receiving side nonlinear equalizer, a first equalizer that can equalize the waveform distortion of the second medium, an odd function input/output characteristic circuit and a variable complex coefficient circuit that input Y a nonlinear correction circuit that generates a complex composite distortion characteristic ^-1 (X) by passing through the input Y and adding it to the input Y; and a second medium that can equalize the waveform distortion of the first medium.
an equalizer, a discriminator that outputs an estimated value of a transmission code from the output of the second equalizer, and a distortion detector that detects waveform distortion from the input-output difference of the discriminator, and the distortion detector The variable complex coefficient circuit of the nonlinear correction circuit is controlled according to the product of the output of the distortion detector and the output of the discriminator in order to minimize distortion of the output of the nonlinear equalizer. It is a vessel.

According to this invention, the band-limited signal can be
It is possible to remove waveform distortion of a digital signal that passes through a normal nonlinear transmission path model, which is amplified by the TWT and received by a bandpass filter.

Next, the present invention will be explained in detail with reference to the drawings.

FIG. 1 is a block diagram of a base band model of a typical microwave band transmission line, which is composed of a transmitting side band limiting filter 10, a TWT 20, and a receiving side selection filter 11. At this time, receiver noise will be added to terminal 151.

Figure 2 is a diagram for explaining the communication method via satellite, and the signal from the ground station 30 is transmitted through a transmitting and receiving filter.
It reaches the ground station 32 via the TWT and a satellite 31. The base band model of this system is the third
As shown in the figure, the band limit filter 1 of the ground station 30
0, a transmission TWT 20, a reception selection filter 11 for the satellite 31, a transmission TWT 21, and a reception selection filter 12 for the ground station 31. As can be seen from this figure, the transmission path model for the multi-relay system is basically a repetition of the one shown in FIG. 1, so only the model shown in FIG. 1 will be explained below. Therefore, the present invention can be easily extended to multiple relay systems as well.

Next, nonlinear correction of the nonlinear amplifier TWT that outputs (x) in response to input x is performed by the following two methods. The first is to pass the input signal x through the inverse function circuit g(x) [ ^-1 (x)] of (x) in advance.
Pre-distortion method for supplying TWT, 2nd
is a derotation method in which the output of TWT is passed through g(x).

The latter method is a correction that can be performed on the receiving side, but this also requires that TWT and g(x) be directly connected, and cannot be applied to the model shown in FIG. 1.

In order to receive the _{signal from} the terminal 101 _of the model in FIG. Applying the inverse function circuit of Z=
It is necessary to correct (x) as g ^-1 (y _i ).

Figure 4 shows the estimator 40 of y _i mentioned above and g(x)
This figure shows the connection of 41 blocks.

Let us now consider the properties of the y _i estimator 40.

In FIG. 5, 50 is a random noise spectrum at the receiving end, and 51 is a signal spectrum at the receiving end, which is band-limited with a 50% roll-off with pulses having a period of T seconds.

FIG. 6 shows a noise spectrum 60 and a signal spectrum 61 that are changed and band-limited by the reception selection filter 11 on the reception side.

From this figure, the signal-to-noise power ratio of the output of the filter 11 has a certain constant value S/N.

Next, consider that the estimator 40 in FIG. 4 is configured with a sample value filter of T second samples. Then, the input spectrum to terminal 140 is S()+N()
By sampling, it takes the form 1/T _∞ 〓 ^n=-∞ S (+n1/T) + 1/T _∞ 〓 ^n=-∞ N (+n1/T) as shown in FIG.

By providing the estimator 40 having the inverse characteristic of the reception selection filter 11 (that is, the characteristic of equalizing the waveform distortion caused by the reception selection filter 11) for the above signal, the spectra 70 and 71 in FIG. The spectra are as shown in 80 and 81.

At this time, the signal-to-noise power ratio for the signal S( ) estimated as the input to the reception selection filter 11 is approximately equal to the S/N described above.

As a result, the estimator 40 estimates the signal at the input end of the reception selection filter 11 with accuracy according to the magnitude of reception input noise.

This is an effect that cannot be obtained when the estimator 40 is implemented as an analog filter to achieve the inverse characteristics of the reception selection filter 11.

Since the above T-second sample is a digital communication, it is a process that is normally performed for signal identification on the receiving side, so no special processing is introduced. Further, as a specific example of the sample value filter, a transversal type filter is generally used.

At this time, if the characteristics of the reception selection filter 11 are such that there is a zero within the transmission band, the filter 1
The inverse characteristic of 1 has infinity within the transmission band. In such a case, the estimator 40 uses y _p to y _i
The above inconvenience can be avoided by using a Kalman filter or the like that estimates . In this case, the transmission path parameters in the Kalman filter (equalization baseband frequency characteristics corresponding to distortion, etc.)
is controlled to match the actual value. It can also be said that the above-mentioned inverse characteristic is approximated so that the transversal type filter does not have the above-mentioned infinity, provided that the estimation accuracy is slightly lowered.

FIG. 9 is a diagram showing the configuration of the main part of this invention.
A first equalizer 40 that equalizes the waveform distortion of the reception selection filter, a nonlinear correction circuit 41 having an inverse characteristic (x) of the ^{characteristic} (x) of the nonlinear circuit existing on the transmission path, and a band limiter on the transmitting side. It consists of three second equalizers 42 that equalize the waveform distortion of the filter. At the output terminal 142, the transmission code sent by the transmitting side is obtained.

FIG. 10 shows a block diagram of a nonlinear correction circuit that has been commonly used in the past. In the figure, 1003
is a cube circuit, 1005 is a variable coefficient multiplier, 1004 is a variable phase shifter, and 1007 is a synthesizer. Block 1006 therefore constitutes a variable complex coefficient unit.

Suppose that for an input x, due to the nonlinear distortion of TWT, (x) = x + η x | x | ² is output (x). This is the actual
It is a good approximation of TWT nonlinear distortion. Here, η is a complex number such that η=α・e ^j 〓.

Now, when the signal X is input to the TWT, the output will be (X) = X + η・X・|X| ² as stated above. This signal is expressed as g(x)=x+ξ・x|x| ² (ξ≡α・e ^j 〓; In FIG.
4, any value can be obtained. ) When passed through a nonlinear correction circuit with the following characteristics, its output C ₀ is C ₀ =X+η・X|X| ² +ξ｜|X+η・H|X| ²
(X+η・X・|X| ² )} X+η・X|X| ² +ξ・X|X| ² =X+(η
+ξ)・X・|X| ² Therefore, the error E for the original signal value X of C ₀ is
Subtracting the original value X from C ₀ gives E=C ₀ −X=(η+ξ)・X・|X| ² . From this formula, by setting ξ=-η, E
=0. Then, the control of ξ (dξ/dt) is performed as follows: dξ/dt=-∂|E| ² /∂ξ×(minimal coefficient), so that ∂|E| ² /∂ξ=0 at ξ=-η. Stabilize.

∂｜E｜ ² /∂ξ=∂｜E｜ ² /∂ξ _R +j∂｜E｜ ² /
∂ξ _I = 2 ⁽ ξ + _η ) · _| First, consider the quantity C ₁ shown above.

C ₁ ^{= E} × [complex conjugate of the discriminator output (= original value of
η)・|X| ^{2 ・}|X ^{| 2} so C ₁ = ¹ /| | ² /∂ξ has been found. It can be seen that the complex coefficient ξ of the nonlinear correction circuit can be controlled from E and the discrimination value in this way.

FIG. 11 is a block diagram showing an embodiment of the present invention, in which 40 and 42 are first and second equalizers;
41 is a nonlinear correction circuit, and block 900 constitutes a control circuit for the coefficient ξ of the nonlinear correction circuit, which embodies ξ=∫ ^t _-∞ −E・X ^* dt from the previous description. . First, 911 is a discriminator, 912 is a distortion detector that detects E, 904 is an imaginary part polarity inverter that inverts only the polarity of the imaginary part of the discrimination value in order to take the complex conjugate of the discrimination value, and 901 is E.
A multiplier that obtains the product of X ^* , 902 is an inverter, 9
03 is an integrator.

As explained above, the second equalizer 42 estimates the output signal of the nonlinear circuit, so
Based on this signal, ξ can be controlled so that the nonlinear distortion included here is minimized.

[Brief explanation of drawings]

Figure 1 is a block diagram of an equalized base band model of a nonlinear transmission path, Figure 2 is a diagram for explaining the satellite communication system, and Figure 3 is an equalized base band model of the system in Figure 2. 4 is a block diagram for estimating the output of the nonlinear circuit, FIG. 5 is a diagram showing the spectrum of the nonlinear circuit output, FIG. 6 is a diagram showing the spectrum at the output of the reception selection filter, and FIG. Figure 8 shows the spectrum after sampling the signal in Figure 6.
Figure 9 is a block diagram showing the configuration of the main part of the present invention, Figure 10 is a configuration diagram of a conventional nonlinear correction circuit, and Figure 11 is a diagram showing the spectrum after equalization with a variable equalizer. FIG. 1 is a block diagram showing one embodiment of the invention. In the figure, 40 is a first equalizer, 42 is a second equalizer, 41 is a nonlinear correction circuit, 911 is a discriminator, and 912 is a distortion detector.

Claims (1)

[Claims] 1 The transmission signal passes through a first medium that causes a first waveform distortion, passes through a nonlinear circuit that outputs (X) for input X, and then passes through a second medium that causes a second waveform distortion to the receiving side. In the nonlinear transmission path leading to the nonlinear equalizer, the input Y is passed through a first equalizer capable of equalizing waveform distortion of the second medium, an odd function input/output characteristic circuit, and a variable complex coefficient circuit, a nonlinear correction circuit that adds to the input Y to generate a complex composite distortion characteristic ^-1 (X); a second equalizer that can equalize the waveform distortion of the first medium; and the second equalizer. a discriminator that outputs an estimated value of the transmitted code from the output of the discriminator, and a distortion detector that detects waveform distortion from the input/output difference of the discriminator, and the nonlinear correction is performed to minimize distortion of the output of the distortion detector. A nonlinear equalizer, characterized in that a variable complex coefficient circuit of the circuit is controlled according to the product of the output of the distortion detector and the output of the discriminator.