JP2002344418A - Prefix type nonlinear distortion compensator - Google Patents

Abstract

(57) [Problem] To provide a front nonlinear distortion compensator capable of performing wideband distortion compensation with a simple configuration. SOLUTION: A prefix type nonlinear distortion compensator includes a nonlinear characteristic compensator 8 for converting a baseband signal x (t) into a baseband signal y (t) by using an inverse characteristic h (P) of an amplifier 16. , Output from the amplifier 16 and the down converter 1
8, receives the baseband signal z ′ (t) that has passed through the A / D converter 20 and the quadrature detector 22 and normalized by the gain q of the amplifier 16, and receives the baseband signals y (t) and z ′ (t ), The input / output characteristic g (P) of the amplifier 16
And a non-linear characteristic estimator 32 that calculates the inverse characteristic h (P) of the non-linear characteristic and supplies it to the non-linear characteristic compensator 8.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a front nonlinear distortion compensator, and more particularly to a front nonlinear distortion compensator in which an inverse characteristic of input / output characteristics of an amplifier is approximated by a complex polynomial or a spline function.

[0002]

2. Description of the Related Art Conventionally, in order to realize nonlinear distortion compensation of a communication system, a front-end nonlinear distortion compensator using various digital circuits has been created. Representative examples of a pre-type nonlinear distortion compensator using a digital circuit include those using a look-up table and those that estimate an inverse function of an amplifier.

A pre-type nonlinear distortion compensator using a look-up table divides the characteristics of an amplifier into small sections with respect to the amplitude value of an input signal, stores a representative value of the section in a table, And a compensation value in a table corresponding to the multiplication result to obtain an output signal.

On the other hand, as a pre-type nonlinear distortion compensator for estimating an inverse function of an amplifier, pth-order-predi is used.
There is something called a poster. This prefix type nonlinear distortion compensator approximates an inverse function of a signal converter with a complex polynomial. The complex polynomial is derived from the most general Volterra series as a form of expressing a nonlinear function, and distortion compensation is performed.

[0005]

However, in a front-end nonlinear distortion compensator using a look-up table, since the values in the table take discrete values, floor noise increases due to discontinuity of the values. In order to eliminate floor noise, it is necessary to increase the number of data. However, there is a problem in that increasing the number of data increases the memory capacity.

[0006] pth-order-predistor
In ter, it is necessary to increase the order of the complex polynomial in order to increase the calculation accuracy. For this reason, there is a problem that it takes a long calculation time and the solution does not easily converge.

Further, for a wideband signal, a look-up table, pth-order-predisto,
In both cases, it is necessary to compensate for the non-linearity of the amplitude and phase for each frequency in the signal band, so that there is a problem that it is difficult to implement hardware.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problem, and an object of the present invention is to provide a front nonlinear distortion compensator capable of performing wideband distortion compensation with a simple configuration.

Another object of the present invention is to provide a front type nonlinear distortion compensator which can keep the noise floor low.

[0010]

According to one aspect of the present invention, a front nonlinear distortion compensator approximates an inverse characteristic of a signal converter by a complex polynomial, converts an input signal based on the complex polynomial, and converts the input signal. A non-linear characteristic compensator that supplies the subsequent signal to the signal converter, and a complex polynomial based on the signal supplied to the non-linear characteristic compensator and the conversion output signal converted by the signal converter. And a non-linear characteristic estimator for estimating the least squares of each coefficient.

The inverse function of the input / output characteristic of the signal converter is approximated by a complex polynomial, and each coefficient of the polynomial is estimated by least squares from the input / output signal of the signal converter. Therefore, the input / output characteristics of the signal converter can be estimated with high accuracy. Further, since a lookup table is not required, the noise floor can be kept low with a simple configuration.

Preferably, the input signal is an orthogonal frequency division multiplexed signal, and the nonlinear characteristic estimator discretely Fourier-transforms the signal supplied to the signal converter and the converted output signal converted by the signal converter. A discrete Fourier transform circuit, connected to the discrete Fourier transform circuit and the non-linear characteristic compensator, and based on a signal supplied to the Fourier-transformed signal converter and a transformed output signal, a minimum for estimating each coefficient of the complex polynomial by least squares. And a square estimator.

Even for an input signal having a large dynamic range, such as an orthogonal frequency division multiplexed signal, a discrete Fourier transform is performed based on the input / output signal of the signal converter, and the least squares estimation is performed. The inverse function of the input / output characteristics can be estimated with high accuracy.

[0014] More preferably, the signal converter is connected to a first linear filter receiving the input signal, a first linear filter, receives an output of the first linear filter, and receives a nonlinear function approximated by a complex polynomial. Is modeled by a converter that converts the output of the first linear filter and a second linear filter that is connected to the converter and receives the output of the converter. The nonlinear characteristic compensator compares the inverse characteristic of the signal converter with the inverse characteristic of the second linear filter, the inverse characteristic of the converter, and the first characteristic.
And a function representing the inverse characteristic of the first and second linear filters based on the signal supplied to the first linear filter and the signal output from the second linear filter. The coefficients and the coefficients of the complex polynomial representing the inverse characteristic of the converter are estimated by least squares.

In order to cope with a wideband signal, a model in which a linear filter is arranged before and after a signal converter is adopted, and its inverse characteristic is estimated. For this reason, high-precision distortion compensation is possible even for a wideband signal.

According to another aspect of the present invention, there is provided a front nonlinear distortion compensator which approximates an inverse characteristic of a signal converter with a spline function, converts an input signal based on the spline function, and converts the converted signal into a signal. A non-linear characteristic compensator to be supplied to the converter, and a signal connected to the non-linear characteristic compensator, each coefficient of the spline function is least squared based on a signal supplied to the signal converter and a conversion output signal converted by the signal converter. And a non-linear characteristic estimator for estimating.

The inverse function of the input / output characteristics of the signal converter is approximated by a spline function, and each coefficient of the spline function is estimated from the input / output signal of the signal converter by least squares. For this reason, the order of the complex polynomial can be set low, and the amount of calculation can be reduced. Further, since a lookup table is not required, the noise floor can be kept low with a simple configuration.

[0018] Preferably, the signal converter is an amplifier.

[0019]

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

[0020] The First Embodiment baseband signal band signals S ₀ to be input to the [front standing nonlinear distortion compensator Operation Principle amplifier and y (t), band signal S _OUT output from the amplifier Let z (t) be the baseband signal of, and put the input / output characteristics of the amplifier as g (P), and between y (t), z (t) and g (P), Relationship is established. In general, an object of the preamble nonlinear distortion compensator is to obtain an input / output characteristic h (P) that satisfies the following equation (2) with respect to the input / output characteristic g (P).

If a front-end nonlinear distortion compensator satisfying the input / output characteristic h (P) can be realized, the following equation (3) is obtained.
And (4) are satisfied, and the input signal S _IN to the pre-type nonlinear distortion compensator is compared with the output signal S
_OUT can be linearized. Here, x (t) represents a baseband signal of the input signal S _IN to the front nonlinear distortion compensator.

[0022]

(Equation 1)

[Mathematic Model of Amplifier] "Toshio Nojima, Hideharu Okamoto:" Analysis of Amplifier Input / Output Nonlinearity between Traveling Waves by Complex Power Series Representation and Application to Distortion Compensation Method ", Transactions of the Institute of Electronics, Communication and Engineers, '81 / 12 , Vol.
64-B, no. As discussed in Section 12 above, the mathematical model of a nonlinear amplifier and its nonlinear distortion compensator has conventionally assumed that the signal of interest is a narrowband signal, that is, the frequency characteristics of the amplifier are sufficiently flat. Was. However, in recent wireless systems, the bandwidth of signals has been widened, and the characteristics of actual amplifiers and calculation results using mathematical models have become inconsistent.

Therefore, referring to FIG. 1, before and after the conventional amplifier model G, linear filters A (jω) and B
A mathematical model in which (jω) is inserted has been proposed. For example, "Kenichi HORIGUCHI et al.," A Disto
rtion Analysis Method for FET Amplifiers Using Nov
el Frequency-Dependent Complex Power Series Mode
l ", IEICE TRANS. ELECTRON., VOL.E32-C, NO.5 MAY 199
9 ", it is reported that the actual amplifier output when a wideband signal is input and the mathematical model shown in FIG. 1 match well. That is, to obtain the inverse characteristic h (P) of the amplifier, the linear filters A (jω) and B (jω)
ω) also needs to be compensated.

FIG. 2 shows an inverse system of the amplifier shown in FIG. Before and after the amplifier model H having the inverse characteristic of the amplifier model G, the filter 1 / B (jω) and the linear filter A having the inverse characteristic of the linear filter B (jω)
A filter 1 / A (jω) having the inverse characteristic of (jω) is provided.

However, if the signal band is sufficiently narrow compared to the band of the amplifier and the frequency characteristics can be considered to be infinitely constant, the linear filters A (jω) and B (jω) may be omitted. it can.

[Circuit Configuration of Prefix Nonlinear Distortion Compensator] FIG.
With reference to, the prefix type nonlinear distortion compensator according to the first embodiment of the present invention is for compensating for the nonlinear distortion of the amplifier 16. This pre-type nonlinear distortion compensator is a down-converter 2 for converting a radio frequency signal S _IN into an intermediate frequency signal.
And an A / D (Analog to Digital) converter 4 for sampling the intermediate frequency signal output from the down converter 2 and converting it into a digital signal, and an A / D converter 4, A quadrature detector 6 for converting the sampled intermediate frequency signal into a baseband signal x (t); and an amplifier 16 connected to the quadrature detector 6
Baseband signal x (t) using the inverse characteristic h (P) of
And a quadrature modulator 10 connected to the nonlinear characteristic compensator 8 for converting the converted baseband signal y (t) into an intermediate frequency signal. A D / A (Digital to Analog) converter 12 is connected to the quadrature modulator 10 and converts an intermediate frequency signal output from the quadrature modulator 10 into an analog signal.

The prefix type nonlinear distortion compensator further includes a D / A
An up-converter 14 connected to the converter 12 for converting an analog intermediate frequency signal into a radio frequency signal S ₀ and supplying it to the amplifier 16, and an amplifier 1 connected to the amplifier 16
6 which converts the output signal S _OUT output from the _C.6 into an intermediate frequency signal, and an A / A converter which is connected to the down converter 18 and converts the intermediate frequency signal into a digital signal.
The D / A converter 20 is connected to the D / A converter 20 and the A / D converter 20 to convert the digital signal into a baseband signal z (t).
6, a quadrature detector 22 that outputs a baseband signal z ′ (t) normalized by a gain q of 6, and the nonlinear characteristic compensator 8 and the quadrature detector 22 are connected to the input / output characteristic g of the amplifier 16.
The inverse characteristic h (P) of (P) is calculated, and the nonlinear characteristic compensator 8 is calculated.
And a non-linear characteristic estimator 32 provided to

It should be noted that the quadrature detector 6 is realized by analog,
The output of the down converter 2 is received by the quadrature detector 6, and the output of the quadrature detector 6 is converted to digital by the A / D converter 4,
The output may be supplied to the nonlinear characteristic compensator 8. Further, the quadrature modulator 10 is realized by analog, the output of the nonlinear characteristic compensator 8 is received by the D / A converter 12, the output of the D / A converter 12 is received by the quadrature modulator 10,
The output of the quadrature modulator 10 may be supplied to the up-converter 14. Further, the quadrature detector 22 is realized in analog form, the output of the down converter 18 is received by the quadrature detector 22, the output of the quadrature detector 22 is received by the A / D converter 20, and the output of the A / D converter 20 is A configuration for supplying to the nonlinear characteristic estimator 32 may be used.

S _IN , S ₀ and S _OUT are given by the following equations (5)
(7) respectively. _{S IN = Re {x (t} ) exp (j2 (f RF t)}: Before standing input signal to the nonlinear distortion compensator _{... (5) S 0 = Re} {y (t) exp (j2 (f RF t )}: distortion compensated signal _{... (6) S OUT = Re} {z (t) exp (j2 (f RF t)}: the output signal from the amplifier 16 ... (7) x (t ), y (t) and z (t), respectively band signals S _iN, S ₀ and S _{O UT} of the complex envelope (baseband signal) is .j is .f _RF radio frequency representing the imaginary (Rad
io Frequency). t represents time.

The input / output characteristic G of the amplifier 16 normalized by the gain q is represented by the complex polynomial model (g ₀ ,
g ₁ ,. . . , G _p are complex numbers). The input / output characteristics of the amplifier 16 having nonlinear characteristics can be generally expressed by Volterra series expansion. The following three conditions "(1) The frequency characteristics of the amplifier 16 are uniform (flat). (2) Only distortion that cannot be removed by a filter near the signal frequency is considered. (3) In the baseband region. Equation (8) is a simplified version of the mathematical expression in which "replace with complex number expression."

[0032]

(Equation 2)

[0033] Put On the reverse characteristic calculation of the amplifier 16 following calculation sampling interval of the A / D conversion and D / A conversion and T _s. In addition, for the sake of simplification of description, the following description will be made on the assumption that T _s = 1, but this does not lose generality.

In the following description, the case where the signal is narrow band and the linear filters A (jω) and B (jω) can be ignored, and the case where the signal is wide band and the linear filters A (jω) and B (jω) cannot be ignored The case (FIGS. 1 and 2) will be described separately.

(1) Linear filters A (jω) and B
When (jω) can be ignored When the signal band is sufficiently narrow compared to the band of the amplifier 16, only the inverse characteristic y (t) = H {z ′ (t)} of Expression (8) can be obtained by calculation. I just need. Specifically, complex coefficients h ₀ , h ₁ ,. . . , H _p by calculation.

[0036]

(Equation 3)

In order to obtain the complex coefficient of the equation (9),
There are two types, a method of calculating in the time domain and a method of calculating in the frequency domain. Hereinafter, the solution of the complex coefficient by each method will be described.

A. Calculation in Time Domain A complex coefficient is obtained by the following processes (STEP1) to (STEP3).

(STEP 1) The baseband signals y (t) and z '(t) as input / output signals to / from the amplifier 16 are
(≧ p + 1) samples are obtained.

(STEP 2) Based on the acquired samples, the following simultaneous equations (unknown numbers: h ₀ , h ₁ ,.
h _p ).

[0041]

(Equation 4)

When this equation is expressed as a vector, the following equation (10) is obtained. V_y = V_ZV_h (10) where V_y, V_Z and Z_h are represented as follows.

[0043]

(Equation 5)

T represents transposition. Where A ^T is the matrix A
Means transposition. (STEP 3) From equation (10), V is calculated using the least squares method.
_H. Specifically, V_h is obtained from the following equation (11).

V_h = (V_Z ^H V_Z) ⁻¹ V_Z ^H V_y (11) where A ^H means the conjugate transpose of the matrix A.

B. Calculation in Frequency Domain The input signal to the amplifier 16 is OFDM (Orthogonal Frequ
In the case of an ency division multiplexing signal (orthogonal frequency division multiplex signal), the amplitude of the sampled signal fluctuates greatly. For this reason, when a signal having a sufficiently large dynamic range cannot be handled by the front nonlinear distortion compensator, an error occurs when V_h is obtained by calculation in the time domain. To solve this, a calculation for obtaining V_h in the frequency domain may be performed. The solution of V_h in the frequency domain is represented by the following (STEP 1)
This is shown in (STEP 3).

(STEP 1) The baseband signals y (t) and z '(t) as input / output signals to / from the amplifier 16 are
(The DFT size of the OFDM signal is preferred.)

(STEP 2) V_y, V_z ⁽⁰⁾ , V_
z ⁽¹⁾ ,. . . , V_z ^(N) by DFT (Discrete Fourier Transform). DFT is a linear operation. Therefore, equation (1)
Equation (12) is derived from 0).

V_ξ = V_ΦV_h (12) where V_ξ = DFT ｛V_y｝ V_ζ ^(l) = DFT ｛V_z ^(l) ｝, l = 0,..., P V_Φ = [V_ζ ⁽⁰⁾ , V_ζ ^{(1 ) ), ..., V_ζ (p)} ] (STEP3) from equation (12), by using the least squares method V
_H. Specifically, V_h is obtained from the following equation (13).

V_h = (V_Φ ^H V_Φ) ^-1 V_Φ ^H V_ξ (13) Here, A ^H means a conjugate transpose of the matrix A.

(2) Linear filters A (jω) and B
When (jω) cannot be ignored (when the input signal to the amplifier 16 has a wide band) The linear filters A (jω) and B (jω) shown in FIG.
Is represented by transfer functions shown in the following equations (14) and (15), respectively.

[0052]

(Equation 6)

The following (STEP0) to (STEP1)
According to the process of 2), the frequency characteristic A (j
ω) and B (jω), and the nonlinear characteristic G of the amplifier 16
(Y) and H (z) are sequentially calculated. The i-th estimated values are A ′ ⁽ⁱ⁾ (jω), B ′ ⁽ⁱ⁾ (jω), G
′ ^(I) (y) and H ′ ⁽ⁱ⁾ (z). In addition, in order to uniquely determine the estimated value, g ₀ = 1 and h ₀ = 1. This does not lose generality.

(STEP 0) The baseband signals y (t) and z ′ (t) as input / output signals to / from the amplifier 16 are
(In the case of an OFDM signal, the DFT size of the signal is desirable) Samples are acquired, and i = 0, A ⁽¹⁾ (jω _k ) = 1 (k
= 1,..., N), y (t) = x (t).

However, if each k-th frequency in the frequency domain is ω _k ,

[0056]

(Equation 7)

Is as follows. (STEP 1) By DFT, the baseband signal y
Calculate the frequency characteristics of (t) and z (t).

[0058]

(Equation 8)

[0059] (STEP2) estimates the frequency characteristics of the signal y ₁ of FIG. 1. Note that the frequency characteristic of the signal y _1, may be used as the calculation results of the later-described STEP 8.

V_ξ ₁ ⁽⁰⁾ = [ξ ₁ ⁽⁰⁾ (ω ₁ ), ξ ₁ ⁽⁰⁾ (ω ₂ ),..., Ξ ₁ ⁽⁰⁾ (ω _N )] = [A ⁽ⁱ⁾ (jω ₁ ) ξ ⁽⁰⁾ (ω ₁ ), A ⁽ⁱ⁾ (jω ₂ ) ξ ⁽⁰⁾ (ω ₂ ), ..., A ⁽ⁱ⁾ (jω _N ) ξ ⁽⁰⁾ (ω _N )] (STEP 3) The following calculation is performed. Where IDFT
Represents the inverse discrete Fourier transform.

[0061]

(Equation 9)

(STEP 4) The following relational expression is established between the signal y ₁ and the signal z.

[0063]

(Equation 10)

After all, the following equation holds.

[0065]

[Equation 11]

From the simultaneous equations, the unknowns b ₀ ′ ^{(i + 1)} , b ₁
′ ^{(I + 1)} , ..., b _n ′ ^{(i + 1)} , g ₁ ′ ^{(i + 1)} , ..., g _p ′ ^{(i + 1)} If the number of simultaneous equations N (the number of samples) is larger than the number of unknowns (n + p + 1), the simultaneous equations can be solved by the least square method or the like, and B ′ ^{(i + 1)} (j
ω) and G ′ ^{(i + 1)} (y).

(STEP 5) The signal z ₁ (=
estimating a frequency characteristic of y ₂ / q).

[0068]

(Equation 12)

(STEP 6) The following calculation is performed.

[0070]

(Equation 13)

(STEP 7) When i ≠ 0, the following relational expression is established between the signal z ₁ and the signal y.

[0072]

[Equation 14]

The following equation is derived from the above equation.

[0074]

(Equation 15)

From this simultaneous equation, the unknown: a
^{_{0 '(i + 1),}} a 1' (i + 1),. . . , A _m ′ ^{(i + 1)} . If the number of simultaneous equations N (the number of samples) is larger than the number of unknowns (m), this can be solved by the least squares method or the like, and the estimated value of A ′ ^{(i + 1)} (jω) is obtained. Can be

When i = 0, the following equation is obtained.

[0077]

(Equation 16)

From this simultaneous equation, the unknowns: a ₀ ′ ⁽¹⁾ ,
a ₁ ′ ⁽¹⁾ ,. . . , A _m ′ ⁽¹⁾ , h ₁ ′ ⁽¹⁾ ,. . . , H _p
´ Calculate ⁽¹⁾ .

(STEP 8) When i ≠ 0, the frequency characteristic of the signal z ₂ (= y ₁ ) in FIG. 2 is estimated.

[0080]

[Equation 17]

(STEP 9) When i ≠ 0, the unknowns: h ₁ ^{(i + 1)} ,. . . , H _p ^{(i + 1)} .

[0082]

(Equation 18)

(STEP 10) New amplifier input x
(T) is acquired by N samples.

(STEP 11) After the input signal x (t) is subjected to nonlinear distortion compensation, it is converted into a real signal S ₀ and input to the amplifier 16. In addition, each numerical value can be represented as follows.

[0085]

[Equation 19]

(STEP 12) N (≧ p + 1) samples of the output z (t) of the corresponding amplifier 16 are obtained, and x (t)
If z is substantially equal to z (t), the calculation is terminated. Otherwise, the process returns to STEP1.

As described above, according to the present embodiment, the inverse function of the input / output characteristic of the amplifier is approximated by a complex polynomial, and each coefficient of the polynomial is estimated by the least squares from the input / output signal of the amplifier. Therefore, the input / output characteristics of the amplifier can be estimated with high accuracy. Further, since a lookup table is not required, the noise floor can be kept low with a simple configuration.

Further, even for an input signal having a large dynamic range such as an OFDM signal, the inverse Fourier transform is performed based on the input / output signal of the amplifier and the least-squares estimation is performed. Function can be estimated with high accuracy.

Further, in order to cope with a wideband signal,
A model in which a linear filter was placed before and after the amplifier was adopted, and its inverse characteristic was estimated. For this reason, high-precision distortion compensation is possible even for a wideband signal.

[Second Embodiment] Referring to FIG.
The front-end nonlinear distortion compensator according to the second embodiment
Wave number signal S_INDownconverter that converts
Connected to the down converter 2 and the down converter 2.
Sampling the intermediate frequency signal output from the barter 2
A / D converter 4 for converting to a digital signal, and A / D conversion
Hilbert transform is connected to the output of the A / D converter 4
Hilbert converter 42 for performing conversion and Hilbert converter 4
2 and treats the output of the Hilbert transform as an imaginary number
Imaginary number handling unit 44, A / D converter 4 and imaginary number handling
A / D converter 4 and imaginary number handling
The outputs of the unit 44 are added and the signal x_aAddition that outputs (t)
Connected to the adder 52 and the adder 52, and the inverse characteristic of the amplifier 16
h (P), the signal x _a(T) is the signal y_aChanged to (t)
The nonlinear characteristic compensator 8 to be replaced
And the converted signal y_aExtract the real part of (t)
A real part extractor 56 for outputting an inter-frequency signal;
Connected to the output unit 56 and output by the real part extractor 56
D / A converter 12 for converting a frequency signal into an analog signal
And

The prefix type nonlinear distortion compensator further includes a D / A
An up-converter 14 connected to the converter 12 for converting an analog intermediate frequency signal into a radio frequency signal S ₀ and supplying it to the amplifier 16, and an amplifier 1 connected to the amplifier 16
6 which converts the output signal S _OUT output from the _C.6 into an intermediate frequency signal, and an A / A converter which is connected to the down converter 18 and converts the intermediate frequency signal into a digital signal.
D / D converter 20 and A / D converter 20
A Hilbert converter 48 for performing a Hilbert transform on the output of the converter 20, an imaginary number handling unit 46 connected to the Hilbert converter 48, and handling the output of the Hilbert converter 48 as an imaginary number; an A / D converter 20 and an imaginary number handling unit The signal z _a, which is connected to the A / D converter 20 and the outputs of the A / D converter 20 and the imaginary number handling unit 46, is output from the A / D converter 20.
A signal z ′ obtained by normalizing (t) with the gain q of the amplifier 16
_a (t) is output, and the nonlinear characteristic compensator 8
And a non-linear characteristic estimator 32 connected to the adder 54 for calculating an inverse characteristic h (P) of the input / output characteristic g (P) of the amplifier 16 and providing the calculated characteristic to the non-linear characteristic compensator 8.

S _IN , S _0, and S _OUT are _calculated by the above equation (5).
To (7). Therefore, the signal x
_{a (t), y a (} t) and z _a '(t), the following equation (1
6) to (18). These are simply the shift of the intermediate frequency from DC to the intermediate frequency. In the algorithm described in the first embodiment, x
(T), y (t) and z ′ (t) are each represented by x
be replaced with _{a (t), y a (} t) and z _a '(t),
The inverse characteristic h (P) of the amplifier 16 can be estimated.

[0093] _{x a (t) = x (} t) exp (j2πf IF t) ... (16) y a (t) = y (t) exp (j2πf IF t) ... (17) z a '(t) = _{z'(t) exp (j2πf IF} t) ... (18) [ embodiment 3] in embodiment 1, by approximating the inverse characteristic of the amplifier 16 in complex polynomial, in this embodiment, approximated by a spline function I do. Each coefficient of the spline function is obtained by performing least-squares estimation using the input / output signals of the amplifier 16.

In the following calculation, A / D conversion and D / A
Let T be the sampling interval for conversion. Note that the following description is made on the assumption that T = 1 for simplification of notation, but this does not impair generality.

In the following description, it is assumed that the signal has a narrow band and is linear.
When filters A (jω) and B (jω) can be ignored
Will be described. Signal band compared to the band of amplifier 16
When the region is sufficiently narrow, the inverse characteristic y of the above equation (8) is obtained.
Only (t) = H {z ′ (t)} can be obtained by calculation.
No. Specifically, the spline relation expressed by the following equation (19)
The complex coefficient h of a number_{i, l}(L = 0, ..., p, i =
0,. . . , M) are calculated. Where the complex
The number of numbers is (p + 1) (m + 1). Note that the expression
In (19), p is the order of the spline function,
μ_i(I = 1,..., M) represents a contact point. Also,
From the definition of the pline function, the equation (19) and
And its n-order derivatives (n = 1, 2, ..., p-1) are continuous
It is. Therefore, there are mp constraints. This
Considering these conditions, the unknown coefficient h _{i, l}Number of
Becomes (p + 1) (m + 1) -mp = (p + m + 1)
Can be reduced to After that, the same as in the first embodiment
Thus, a simultaneous equation is made from the input and output signals of the amplifier 16,
By solving the simultaneous equations by the least squares method, the complex
Number h_{i, j}Ask for.

[0096]

(Equation 20)

As described above, according to the present embodiment, the inverse function of the input / output characteristic of the amplifier is approximated by the spline function, and the complex coefficient of the spline function is estimated by the least squares from the input / output signal of the amplifier. Therefore, the input / output characteristics of the amplifier can be estimated with high accuracy. Further, since a lookup table is not required, the noise floor can be kept low with a simple configuration. Furthermore, since the degree p of the complex polynomial can be set low, the amount of calculation can be reduced.

The embodiments disclosed this time are to be considered in all respects as illustrative and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

[0099]

As described above, it is possible to provide a front nonlinear distortion compensator having a low noise floor and a simple structure. It is also possible to compensate for a wideband signal.

[Brief description of the drawings]

FIG. 1 is a diagram showing a mathematical model of an amplifier having a wide frequency characteristic.

FIG. 2 is a diagram showing an inverse system of the amplifier shown in FIG. 1;

FIG. 3 is a block diagram illustrating a hardware configuration of a prefix nonlinear distortion compensator.

FIG. 4 is a block diagram showing a hardware configuration of a prefix nonlinear distortion compensator.

[Explanation of symbols]

2 Down converter, 4,20 A / D converter, 6
Quadrature detector, 8 Nonlinear characteristic compensator, 10 Quadrature modulator, 12 D / A converter, 14 Upconverter, 1
6 amplifier, 18 downconverter, 22 quadrature detector, 32 nonlinear characteristic estimator, 42,48 Hilbert transformer, 44,46 imaginary number handling unit, 52,54 adder.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Renmei Son 3--14-1, Hiyoshi, Kohoku-ku, Yokohama-shi, Kanagawa Prefecture Inside Keio University (72) Inventor Akira Sano 3--14 Hiyoshi, Kohoku-ku, Yokohama-shi, Kanagawa Prefecture No.1 Keio University F term (reference) 5K022 DD01 DD23 DD24 5K060 BB07 CC04 CC11 FF06 HH01 HH14 JJ16 KK06

Claims (5)

[Claims] A nonlinear characteristic compensator that approximates an inverse characteristic of a signal converter with a complex polynomial, converts an input signal based on the complex polynomial, and supplies a converted signal to a signal converter; A nonlinear characteristic estimator connected to a characteristic compensator and configured to estimate each coefficient of the complex polynomial in a least-square manner based on a signal supplied to the signal converter and a conversion output signal converted by the signal converter. , Prefix type nonlinear distortion compensator. 2. The method according to claim 1, wherein the input signal is an orthogonal frequency division multiplexed signal, and the non-linear characteristic estimator separates a signal supplied to the signal converter and a converted output signal converted by the signal converter, respectively. A discrete Fourier transform circuit that performs Fourier transform, and is connected to the discrete Fourier transform circuit and the nonlinear characteristic compensator, and based on the signal supplied to the Fourier-transformed signal converter and the converted output signal, the complex polynomial 2. The non-linear distortion compensator according to claim 1, further comprising a least-squares estimator that estimates each coefficient by least-squares. 3. A signal converter comprising: a first linear filter receiving an input signal; a nonlinear function connected to the first linear filter, receiving an output of the first linear filter, and approximated by a complex polynomial. A converter for converting the output of the first linear filter using: a second connected to the converter and receiving the output of the converter
Wherein the nonlinear characteristic compensator converts the inverse characteristic of the signal converter into the inverse characteristic of the second linear filter, the inverse characteristic of the converter, and the inverse characteristic of the first linear filter. Based on the signal supplied to the first linear filter and the signal output from the second linear filter, a coefficient of a function representing an inverse characteristic of the first and second linear filters, and 2. The prefix nonlinear distortion compensator according to claim 1, wherein least-squares estimation is performed on a coefficient of a complex polynomial representing an inverse characteristic of the converter. 3. 4. A nonlinear characteristic compensator that approximates an inverse characteristic of the signal converter with a spline function, converts an input signal based on the spline function, and supplies a converted signal to a signal converter. A nonlinear characteristic estimator connected to the characteristic compensator and configured to estimate each coefficient of the spline function by least square based on a signal supplied to the signal converter and a conversion output signal converted by the signal converter. , Prefix type nonlinear distortion compensator. 5. The front nonlinear distortion compensator according to claim 1, wherein said signal converter is an amplifier.