JP5303809B2 - Distortion compensation device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To compensate distortion effectively by a distortion compensating apparatus for compensating distortion occurring in an amplifier for amplifying an input signal by a predistortion system. <P>SOLUTION: A coefficient storage unit stores a distortion compensation coefficient for giving predistortion to an input signal corresponding to each address to the address. An address detection unit 6 detects the addresses of M bits, namely 0 to (M-1)-th bits corresponding to the input signal. An address generation unit 7 generates an address, where the values of 0 to (M-N-2)-th bits are rounded down, and the value of the (M-N-1)-th bit is rounded up when it is 1 and is rounded down when it is 0. An error mapping unit 9 maps data of the error between the input signal and an output signal of the amplifier to an address generated for the input signal for averaging for each address, and generates error information based on the result. Update control units 10, 11 update storage content of the coefficient storage unit based on the generated error information. <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to a distortion compensation apparatus, and more particularly to a distortion compensation apparatus that effectively performs distortion compensation.

  For example, power amplifiers are required to have high power efficiency and low nonlinear distortion, which are contradictory characteristics. In general, the input / output characteristics of a power amplifier are linear in a region where the input level is low, but become nonlinear when a certain input level is exceeded, and the output power eventually saturates. Usually, in order to improve the power efficiency of the power amplifier, since it is used at an operating point close to the saturation point, nonlinear distortion occurs due to the nonlinearity of the amplifier.

  In the W-CDMA (Wideband-Code Division Multiple Access) method and the OFDMA (Orthogonal Frequency Division Multiple Access) method, the ratio between the average power and the maximum power is large, and therefore it is necessary to be linear in a wide input power range. As one technique for compensating for non-linear distortion, a pre-distortion (PD) system is used. The PD system is a system that compensates for nonlinear characteristics by giving an inverse characteristic of the nonlinear characteristics to an input signal of an amplifier in advance. In particular, a digital pre-distortion (DPD) method that realizes a PD method by digital signal processing is currently the mainstream because it consumes little power.

FIG. 13 shows a configuration example of a power amplifier with a DPD function.
The power amplifier with a DPD function of this example includes a DPD unit 101, a D / A (Digital to Analog) converter 102, an up converter 103, a power amplifier (PA) 104, a directional coupler 105, a down converter 106, An A / D (Analog to Digital) converter 107 and a DPD control unit 108 are provided.

An example of the operation performed in the power amplifier with a DPD function of this example will be shown.
A digital complex signal composed of an in-phase component (I-phase component) and a quadrature component (Q-phase component) is input to the DPD unit 101 and the DPD control unit 108 as input signals. In this example, the input signal is a baseband signal.
The DPD unit 101 is disposed in a digital unit that is upstream from the power amplifier 104, and performs predistortion processing (processing that gives a reverse characteristic of the nonlinear characteristic of the power amplifier 104) on the input signal to perform a D / A converter 102. Output to.

The D / A converter 102 converts the signal input from the DPD unit 101 that is a digital signal into an analog signal and outputs the analog signal to the up-converter 103.
The up-converter 103 converts the signal input from the D / A converter 102 into a radio frequency signal and outputs the signal to the power amplifier 104.
The power amplifier 104 amplifies (power amplifies) the signal input from the up-converter 103 and outputs the amplified signal. At this time, nonlinear distortion occurs in the power amplifier 104, but since the inverse characteristics of the nonlinear characteristic are given to the signal in advance by the DPD unit 101, the output signal from the power amplifier 104 is a signal with compensated distortion.

Directional coupler 105 distributes a part of the output signal from power amplifier 104 and outputs it to down converter 106.
The down converter 106 converts the signal input from the directional coupler 105 from a radio frequency signal to a signal having an original frequency (in this example, a frequency before the frequency conversion by the up converter 103 and a baseband frequency). The frequency is converted and output to the A / D converter 107.
The A / D converter 107 converts the signal input from the down converter 106 that is an analog signal into a digital signal, and outputs the digital signal to the DPD control unit 108 as a feedback signal.

  The DPD control unit 108 follows a temperature change, an aging change, or the like based on an input signal or a signal (feedback signal) input from the A / D converter 107 so that good linearity can always be obtained. In addition, the DPD unit 101 (contents of predistortion processing performed by the DPD unit 101) is adaptively controlled.

FIG. 14 shows a configuration example of the DPD unit 101.
The DPD unit 101 of this example includes an amplitude detection unit 111, a distortion compensation table 112 made up of a lookup table (LUT: Look Up Table) made up of a memory, and a PD execution unit 113 made up of a complex multiplier. ing.

An example of the operation performed in the DPD unit 101 of this example is shown.
A digital input signal (baseband signal in this example) composed of an I-phase component and a Q-phase component is input to the amplitude detection unit 111 and the PD execution unit 113.
The amplitude detector 111 calculates the amplitude of the input signal and outputs the amplitude value to the distortion compensation table 112.

  The distortion compensation table 112 uses the amplitude value as a reference argument, and associates the amplitude value with a value corresponding to the inverse characteristic of the nonlinear characteristic of the amplifier that is the distortion compensation target (in this example, the power amplifier 104). Store (memorize). In general, there are AM / AM characteristics (amplitude) and AM / PM characteristics (phase) using the amplitude of an input signal as an index as inverse characteristics of the nonlinear characteristics of an amplifier that is a distortion compensation target.

The distortion compensation table 112 outputs a distortion compensation coefficient value (complex number in this example) corresponding to the amplitude value input from the amplitude detection unit 111 to the PD execution unit 113.
The PD execution unit 113 compensates the amplitude and phase of the input signal according to the reference result of the distortion compensation table 112. For example, the PD execution unit 113 performs complex multiplication of the input signal and the distortion compensation coefficient signal (value) from the distortion compensation table 112, and outputs the resultant signal to the D / A converter 102.

One method for adaptively controlling the distortion compensation table 112 is to use waveform comparison.
FIG. 15 shows a configuration example of the DPD control unit 108 using waveform comparison.
The DPD control unit 108 of this example includes a memory 121, a quadrature detection unit 122, a memory 123, a delay adjustment unit 124, an error detection unit 125, an error mapping unit 126, an error interpolation unit 127, and an LUT update unit 128.

An example of the operation performed in the DPD control unit 108 of this example will be shown. Note that t represents time (for example, time in sample units).
An input signal and a feedback signal (feedback signal) from the A / D converter 107 are input to the DPD control unit 108.
The quadrature detection unit 122 outputs the feedback signal to the memory 123 as a signal composed of an I-phase component and a Q-phase component by performing quadrature detection on the feedback signal.
The memory 123 stores the signal input from the quadrature detection unit 122.
The memory 121 stores an input signal.

The delay adjustment unit 124 adjusts the timing by adjusting the delay so that the delay time between the input signal stored in the memory 121 and the feedback signal stored in the memory 123 becomes zero, and the delay is adjusted in this way. The transmission signal Tx (t) and the feedback signal Rx (t) corresponding to the input signal are input to the error detection unit 125.
Here, as a method for adjusting the delay, for example, one of the signals stored in the memory is shifted and adjusted with an accuracy of one sample time, or a digital filter having a delay coefficient is less than one sample. Various known techniques such as a method of adjusting the delay can be used.

The error detection unit 125 compares the input transmission signal Tx (t) and the feedback signal Rx (t) to detect an amplitude and phase error for each sample, and an amplitude error (gain) Err _amp (t) And the phase error Err _phase (t) are output to the error mapping unit 126.
Here, for example, the amplitude error Err _amp (t) = | Rx (t) | / | Tx (t) | is expressed, and the phase error Err _phase (t) is viewed from the transmission signal Tx (t). A phase error expressed by a relative angle of the feedback signal Rx (t) is used.

The error mapping unit 126 receives the amplitude error Err _amp (t) and the phase error Err _phase (t) from the error detection unit 125, and also has an address of the distortion compensation table 112 according to the amplitude of the transmission signal Tx (t). Address (t) is input.
The error mapping unit 126 maps the amplitude error Err _amp (t) and the phase error Err _phase (t) input to the input address Address (t) to update the distortion compensation table 112 once. A plurality of mapping information to be used is output to the error interpolation unit 127 as error information Err _info (T) once every update period T of the distortion compensation table 112 as a group.

Here, such an address Address (t) may be notified from the DPD unit 101 to the DPD control unit 108, or may be detected by the DPD control unit 108 based on an input signal.
As an example of the configuration, the error mapping unit 126 of this example inputs the address obtained by the amplitude detection unit 111 (or a corresponding one) as the address Address (t). In this case, the delay adjustment unit 124 is used. Then, the input signal (transmission signal) is output as a transmission signal Tx (t) without performing signal processing using the reference signal, and the delay of the feedback signal is adjusted and output as the feedback signal Rx (t).
As another configuration example, a circuit having the same function (or similar function) as that of the amplitude detection unit 111 may be provided, and an address Address (t) may be obtained by inputting an input signal (transmission signal Tx (t)). Good.

The error interpolation unit 127 performs error averaging for removing the influence of noise of the feedback signal, and measures against collecting errors for all LUT addresses (addresses of the distortion compensation table 112) within a finite time. The error information Err _info (T) input from the error mapping unit 126 is interpolated to generate amplitude errors LUT_Err _amp (T) and phase errors LUT_Err _phase (T) of all addresses in the distortion compensation table 112. , Output to the LUT update unit 128.

Here, as an interpolation method, for example, a least square method is used.
In FIG. 16A, an address (input power) is shown on the horizontal axis, an amplitude error is shown on the vertical axis, and an example of least square approximation is shown. FIG. 16B shows an example of least square approximation, where the horizontal axis indicates an address (input power) and the vertical axis indicates a phase error.
In FIGS. 16A and 16B, white circles indicate error samples, and solid lines indicate approximate curves obtained by the method of least squares.

Based on the amplitude error LUT_Err _amp (T) and the phase error LUT_Err _phase (T), which are errors obtained by the error interpolation unit 127 and input from the error interpolation unit 127, the LUT update unit 128 is shown in (Expression 1). As described above, the amplitude information LUT _amp (t) for AM / AM compensation and the phase information LUT _phase (t) for AM / PM compensation are corrected, and these information are also stored in the distortion compensation table of the DPD unit 101. Set to 112.

[Equation 1]
LUT _amp ((n + 1) T)
= LUT _amp (nT) × μ {1 / LUT_Err _amp (nT)}
LUT _phase ((n + 1) T)
= LUT _phase (nT) + μ {−LUT_Err _phase (nT)}
・ ・ (Formula 1)

Here, T represents the update period of the distortion compensation table (LUT) 112 (for example, a value in sample units as in t), n represents an integer, μ does not have a sufficient average, and is sensitive to noise and the like. The value of 0.0 <μ ≦ 1.0 is used.
Note that it is desirable that the distortion compensation table 112 be stored in the form of complex numbers by performing, for example, polar coordinate-orthogonal coordinate conversion.

JP 2009-111958 A International Publication No. 2006/087864 Pamphlet

However, the conventional distortion compensation apparatus as described above has the following (Problem 1) to (Problem 3).
(Problem 1) In a power amplifier having complicated nonlinear characteristics, an approximation different from the ideal may be performed, so that sufficient performance of the DPD cannot be obtained.
FIG. 17 shows an example in which approximation different from the ideal is performed and the error cannot be approximated correctly. The horizontal axis indicates the address (input power), and the vertical axis indicates the amplitude error. White circles indicate error samples, and an example of an ideal approximation curve and an example of an approximation curve obtained by the least square method are shown.

(Problem 2) When μ in (Equation 1) is reduced in order to ensure the stability of distortion compensation, the amount of correction in one update decreases, and as a result, the initial convergence time of the nonlinear inverse characteristic And the time required for following increases.
(Problem 3) A preferred update method for the DPD that compensates for the memory effect is not specified.

The present invention has been made in view of such a conventional situation, and an object thereof is to provide a distortion compensation apparatus capable of effectively performing distortion compensation.
Specifically, the configuration (referred to as configuration example A1) that solves the above (problem 1), the configuration that solves the above (problem 2) (referred to as configuration example A2), and the above (problem 3). A configuration to be eliminated (referred to as configuration example A3) is realized.

[1] (Configuration Example A1) In order to achieve the above object, the present invention has the following configuration in a distortion compensation apparatus that compensates for distortion generated in an amplifier that amplifies an input signal by a predistortion method.
That is, the coefficient storage means stores a distortion compensation coefficient for predistorting the input signal corresponding to each address at each address. Address detecting means detects an address of 0th to (M−1) th M (M is an integer of 2 or more) bits corresponding to the input signal. The address generation means uses the set N (N is an integer smaller than M) and sets the bit values from 0 to (M−N−2) -th for the M-bit address detected by the address detection means. An address is generated by rounding down and rounding down the (M−N−1) -th bit value (0 is rounded down and 1 is rounded up). Error mapping means maps error data between the input signal and the output signal of the amplifier to the address generated by the address generation means for the input signal, averages each address, and error information based on the result Is generated. Update control means updates the stored contents of the coefficient storage means based on the error information generated by the error mapping means.

  Therefore, since error data is mapped and averaged to a representative address generated based on an address corresponding to an input signal, for example, in a power amplifier having a complicated nonlinear characteristic, Compared to the above, it is expected that approximation close to the ideal is performed and sufficient performance of the DPD is obtained, and distortion compensation can be effectively performed.

Here, in the coefficient storage means (for example, memory), as the correspondence between the address and the signal, for example, the correspondence between the signal level (for example, amplitude or power value) and the address can be used.
Further, as the distortion compensation coefficient, various values may be used depending on, for example, a configuration method for executing distortion compensation.

Various values may be used as the bit number M of the M-bit address and the bit number N of the higher N bits.
As the error, for example, an amplitude error or a phase error can be used.
The error data may be averaged in various modes. For example, a mode in which averaging is performed for each update period of the storage contents of the coefficient storage means can be used.

Further, as the error information, various information regarding the error may be used.
In addition, various modes may be used as a mode of updating the storage contents of the coefficient storage unit based on the error information. For example, a mode of updating so as to reduce the error indicated by the error information is used. Can do.

[2] (Configuration Example A2) In order to achieve the above object, the present invention has the following configuration in a distortion compensation device that compensates for distortion generated in an amplifier that amplifies an input signal by a predistortion method.
That is, the coefficient storage means stores a distortion compensation coefficient for predistorting the input signal corresponding to each address at each address. An address acquisition unit acquires an address corresponding to the input signal or an address based thereon. The error data selection means excludes error data out of the set range from the error data between the input signal and the output signal of the amplifier, and outputs other error data (error data within the set range). ) Is selected. The error mapping means maps error data between the input signal selected by the error data selection means and the output signal of the amplifier to the address obtained by the address obtaining means for the input signal, and for each address Averaging is performed, and error information based on the result is generated. Update control means updates the stored contents of the coefficient storage means based on the error information generated by the error mapping means.

  Therefore, with respect to the error data between the input signal and the output signal of the amplifier, the error data out of the set range is excluded and other error data (error data within the set range) is selected. Therefore, for example, it is expected that the initial convergence time of nonlinear inverse characteristics and the time required for tracking will be shortened while ensuring the stability of distortion compensation compared with the conventional case, and distortion compensation is effectively performed. It can be carried out.

Here, in the coefficient storage means (for example, memory), as the correspondence between the address and the signal, for example, the correspondence between the signal level (for example, amplitude or power value) and the address can be used.
Further, as the distortion compensation coefficient, various values may be used depending on, for example, a configuration method for executing distortion compensation.

Further, as the address acquisition means, for example, an address detection means for detecting an address corresponding to the input signal is used, or an address such as a representative is generated from the address detection means and the address detected by the address detection means. Address generation means can be used.
As the error, for example, an amplitude error or a phase error can be used.
Moreover, various ranges may be used as the range regarding the error. For example, both the upper limit and the lower limit may be used, or only one of the upper limit and the lower limit may be used. Moreover, as a range, for example, it may be expressed by “more than” or “less than”, or may be expressed by “greater than” or “less than”.
The error data may be averaged in various modes. For example, a mode in which averaging is performed for each update period of the storage contents of the coefficient storage means can be used.

Further, as the error information, various information regarding the error may be used.
In addition, various modes may be used as a mode of updating the storage contents of the coefficient storage unit based on the error information. For example, a mode of updating so as to reduce the error indicated by the error information is used. Can do.

[3] (Configuration Example A3) In order to achieve the above object, according to the present invention, a distortion compensation device that compensates for distortion due to the memory effect generated in an amplifier that amplifies an input signal by a predistortion method is configured as follows. It was.
That is, the memory effect coefficient storage means stores, for each address, a memory effect distortion compensation coefficient for applying predistortion to the input signal corresponding to each address. An address acquisition unit acquires an address corresponding to the input signal or an address based thereon. The error mapping means maps error data between the input signal and the output signal of the amplifier to the address acquired by the address acquisition means for the input signal, and the error mapping between the input signal and the output signal of the amplifier. Data obtained by inverting the sign (positive / negative sign) of data is mapped to the address acquired by the address acquisition means for the input signal that is past the input signal, averaged for each address, and error information based on the result Is generated. Update control means updates the stored contents of the coefficient storage means based on the error information generated by the error mapping means.

  Therefore, the error data between the input signal and the output signal of the amplifier is mapped to the address obtained for the input signal, and the data obtained by inverting the sign of the error data between the input signal and the output signal of the amplifier is Using a configuration in which the input signal is mapped to an address acquired for a past input signal, a configuration preferable for DPD that compensates for the memory effect (for example, updating of the storage contents of coefficient storage means based on error information) is realized. Distortion compensation can be performed effectively.

Here, in the coefficient storage means (for example, memory), as the correspondence between the address and the signal, for example, the correspondence between the signal level (for example, amplitude or power value) and the address can be used.
Further, as the distortion compensation coefficient, various values may be used depending on, for example, a configuration method for executing distortion compensation.

Further, as the address acquisition means, for example, an address detection means for detecting an address corresponding to the input signal is used, or an address such as a representative is generated from the address detection means and the address detected by the address detection means. Address generation means can be used.
As the error, for example, an amplitude error or a phase error can be used.
The error data may be averaged in various modes. For example, a mode in which averaging is performed for each update period of the storage contents of the coefficient storage means can be used.

Further, as the error information, various information regarding the error may be used.
In addition, various modes may be used as a mode of updating the storage contents of the coefficient storage unit based on the error information. For example, a mode of updating so as to reduce the error indicated by the error information is used. Can do.

  As described above, the distortion compensation apparatus according to the present invention can effectively perform distortion compensation.

It is a figure which shows the structural example of the DPD control part using the waveform comparison which concerns on 1st Example of this invention. It is a figure for demonstrating the process performed by an error representative point aggregation part. It is a figure which shows an example of the range of averaging of each representative point. (A), (b) is a figure which shows the example of the difference | error of the transmission signal Tx (t) and the feedback signal Rx (t). (A)-(f) is a figure which shows the example of data selection. It is a figure which shows the structural example of an error mapping part. (A), (b) is a figure which shows the example of an error. It is a figure which shows the structural example of the DPD control part using the waveform comparison which concerns on 2nd Example of this invention. (A)-(c) is a figure for demonstrating the error in case there exists a memory effect. It is a figure which shows the structural example which connected the memoryless PD part and the memory PD part in series. It is a figure which shows the structural example of the error mapping part used in order to compensate a memory effect. It is a figure which shows an example of the error mapping for memory DPD. It is a figure which shows the structural example of the power amplifier with a DPD function. It is a figure which shows the structural example of a DPD part. It is a figure which shows the structural example of the DPD control part using a waveform comparison. (A) is a figure which shows an example of an amplitude error and a least square approximation, (b) is a figure which shows an example of a phase error and a least square approximation. It is a figure which shows the example in case the approximation different from an ideal is performed and an error cannot be approximated correctly.

Embodiments according to the present invention will be described with reference to the drawings.
The overall configuration of the power amplifier with a DPD function that compensates for nonlinear distortion of the power amplifier by the digital predistortion method according to the present embodiment is the same as that shown in FIG. 13, for example.

  The components that solve the above (Problem 1) are mainly components related to the error representative point aggregating unit 7 in (Example 1) and (Example 2) (Configuration Example A1). ) Is mainly a component related to the data selection unit 8 in (Example 1) and (Example 2) (Configuration Example A2), and a component that solves the above (Problem 3) is mainly used. This is a configuration unit relating to the error mapping unit 61 (eg, using a delayed address signal for the input data of the averaging unit 74) in (Example 3) (Configuration Example A3).

  Regarding the above configuration examples A1, A2, and A3, the configuration example A3 can be applied to a memory effect type DPD, and for the configuration examples A1 and A2, for example, (configuration example A1 + configuration example A2), It is possible to combine them as appropriate, such as (Configuration Example A1 + Configuration Example A3), (Configuration Example A2 + Configuration Example A3), and (Configuration Example A1 + Configuration Example A2 + Configuration Example 3).

In addition, the invention grasped from the present embodiment includes a range in which a person engaged in the same industry having ordinary technical knowledge can easily guess.
For example, the DPD shown in the conventional example is an example, and may be applied to various DPDs.
As the distortion compensation table (LUT) and error values, not only polar coordinate system values (R, θ) but also orthogonal coordinate system values (I, Q) can be used. The exchange formula for these coordinate systems is I + jQ = R · exp (jθ).
Further, as an object to be interpolated in an integrated manner, an error is used in this embodiment, but a correction value may be used. Even in this case, processing is performed using not only the values of the polar coordinate system (1 / Err _amp (nT), −Err _phase (nT)) but also the values of the orthogonal coordinate system (I (nT), Q (nT)). May be performed.

A first embodiment of the present invention will be described.
FIG. 1 shows a configuration example of a DPD control unit (corresponding to the DPD control unit 108 shown in FIG. 13) using waveform comparison according to an embodiment of the present invention.
The DPD control unit of this example includes a memory 1, a quadrature detection unit 2, a memory 3, a delay adjustment unit 4, an error detection unit 5, an amplitude detection unit 6, an error representative point aggregation unit 7, a data selection unit 8, and an error mapping unit 9. , An error interpolation unit 10 and an LUT update unit 11 are provided.

An example of the operation performed in the DPD control unit of this example will be shown. Note that t represents time (for example, time in sample units). Further, the flow of processing up to the error detection unit 5 is the same as that shown in FIG. 15, for example.
An input signal and a feedback signal (feedback signal) from an A / D converter (corresponding to the A / D converter 107 shown in FIG. 13) are input to the DPD control unit.
The memory 1 stores input signals.
The quadrature detection unit 2 outputs the feedback signal to the memory 3 as a signal composed of an I-phase component and a Q-phase component by performing quadrature detection on the feedback signal.
The memory 3 stores the signal input from the quadrature detection unit 2.

The delay adjusting unit 4 adjusts the timing by adjusting the delay so that the delay time between the input signal stored in the memory 1 and the feedback signal stored in the memory 3 becomes zero, and the delay is adjusted in this way. The transmission signal Tx (t) and the feedback signal Rx (t) corresponding to the input signal are input to the error detection unit 5.
Here, as a method for adjusting the delay, for example, one of the signals stored in the memory is shifted and adjusted with an accuracy of one sample time, or a digital filter having a delay coefficient is less than one sample. Various known techniques such as a method of adjusting the delay can be used.

The error detector 5 compares the input transmission signal Tx (t) with the feedback signal Rx (t), detects an error in amplitude and phase for each sample, and amplitude error (gain) Err _amp (t) And the phase error Err _phase (t) are output to the data selector 8.
Here, for example, the amplitude error Err _amp (t) = | Rx (t) | / | Tx (t) | is expressed, and the phase error Err _phase (t) is viewed from the transmission signal Tx (t). A phase error expressed by a relative angle of the feedback signal Rx (t) is used.
It is mapped as an error in the address of the distortion compensation table (corresponding to the distortion compensation table 112 shown in FIG. 14) corresponding to the amplitude of the transmission signal (input signal).

The amplitude detection unit 6 has the same function (or similar function) as the amplitude detection unit of the DPD unit (for example, one corresponding to the amplitude detection unit 111 of the DPD unit 101 shown in FIGS. 13 and 14). The address Address _M (t) corresponding to the amplitude of the input signal is obtained and output to the error representative point aggregation unit 7.
As described above, in this example, the DPD control unit detects the address Address _M (t) based on the input signal. However, as another configuration example, the DPD unit notifies the address Address _M (t) to the DPD control unit. A configuration may be used.

The error representative point aggregating unit 7 converts the address Address _M (t) input from the amplitude detection unit 6 into an address Address _N (t) and aggregates the errors into a small number of representative points, and sends the data to the data selecting unit 8. Output.
With reference to FIG. 2, processing performed by the error representative point aggregating unit 7 (processing for obtaining a representative point address Address _N (t)) will be described. M and N each represent a natural number, and M> N.
A distortion compensation table address Address _M (t) is configured by M bits (b _{0 to} b _M−1 ) shown in FIG. 2, and the number of addresses is 2 ^M. In this example, an error is mapped to an address Address _N (t) using the upper N bits (N = 3 in the example of FIG. 2) of the address Address _M (t). In this example, a preferred embodiment example, as b _M-N-1-th than all lower bits are bits 0, the b _M-N-1-th bit abandoned if 0 discard 1 Input City (0, If it is 1, it is moved up) and is mapped to a point of (2 ^N +1) considering the carry.

FIG. 3 shows an example of the average range of each representative point (address Address _N (t)) as an example in which errors are aggregated and averaged.
In this example, a case where M = 10 bits (distortion compensation table address Address _M (t) is 0 to 1023) and N = 3 (aggregation of nine representative points) is shown.
In FIG. 3, circles (◯) indicate error aggregation points (representative points). In this example, there are nine points from the first representative point to the ninth representative point, and the address of each representative point is [0]. 128, 256, 384, 512, 640, 768, 896, 1023].
The range of the address Address _M (t) to be aggregated (averaged in this example) at each representative point is [(0-63), (64-191), (192-319), (320-447). (448-575), (576-703), (704-831), (832-959), (960-1023)].
The errors mapped in this way are averaged at each representative point.
Thus, for example, conventionally addresses have used bits _{b M-1} ~b ₀ as Address (t) has, in this _example, are aggregated into the address with bits _{b M-1 ~b M-N} .

The data selection unit 8 inputs the amplitude error Err _amp (t) and the phase error Err _phase (t) from the error detection unit 5 and also inputs the address Address _N (t) from the error representative point aggregation unit 7. When outputting to the error mapping unit 9, data is selected, thereby shortening the time required for the initial convergence and tracking of the nonlinear inverse characteristic while ensuring the stability of distortion compensation.
Here, the reason why the distortion compensation is not stable is because of noise Noise (t) added to the feedback signal Rx (t), and therefore the power amplifier (for example, one corresponding to the power amplifier 104 shown in FIG. 13). This is because the error Error (t) due to the nonlinear characteristic cannot be measured accurately.

FIGS. 4A and 4B show examples of errors between the transmission signal Tx (t) and the feedback signal Rx (t). The horizontal axis represents the I-phase component, and the vertical axis represents the Q-phase component.
As shown in FIG. 4 (a), when the amplitude of the transmission signal Tx (t) is large, the influence of noise Noise (t) is relatively small. However, as shown in FIG. When the amplitude of the signal Tx (t) is small, the influence of noise Noise (t) becomes large.
In particular, when the amplitude of the transmission signal Tx (t) is around 0, the amplitude error Err _amp (t) becomes very large, and the phase error Err _phase (t) is also dominated by noise Noise (t).
When the contents of the distortion compensation table are adaptively updated using data greatly affected by the noise Noise (t) as described above, the contents of the distortion compensation table vibrate and sufficient distortion compensation performance cannot be obtained.

Thus, in this example, the data selection unit 8 selects the data as a condition that the amplitude error Err _amp (t) is outside the range of the threshold Th _amp or the threshold Th with the phase error Err _phase (t). If at least one of the conditions that the _phase is out of the range is satisfied, the sample data of the address Address _N (t), the transmission signal Tx (t), and the feedback signal Rx (t) is excluded. Thus, the error mapping unit 9 is not output.

As a result, in this example, only data in which both the amplitude error Err _amp (t) and the phase error Err _phase (t) are within the predetermined threshold range are used (for example, using the expression shown in (Expression 1)). B) to be used for the process of updating the distortion compensation table.
As another configuration example, it excluded if both amplitude error _{Err # 038} (t) and the phase error _{Err phase} (t) is outside the range of a predetermined threshold value, the amplitude error _{Err # 038} (t) and the phase error It is also possible to select one of Err _phase (t) if it is within a predetermined threshold range.

Here, as a specific example, criteria for selecting data for amplitude error _{Err # 038} (t) (adopted) is expressed by Equation (2), and does not select the data for amplitude error _{Err # 038} (t) (rejected to) condition is expressed by equation (3), criteria for selecting data for phase error _{Err phase} (t) (adopted) is expressed by equation (4), the data for the phase error _{Err phase} (t) The condition that is not selected (not adopted) is expressed as (Equation 5).

[Equation 2]
Conditions of adoption data of amplitude error: | 10 log {Err _amp (t)} | <Th _amp
.. (Formula 2)
[Equation 3]
Conditions for non-acceptance data of amplitude error: | 10 log {Err _amp (t)} | ≧ Th _amp
.. (Formula 3)
[Equation 4]
Conditions for adoption data of phase error: | Err _phase (t) | <Th _phase
.. (Formula 4)
[Equation 5]
Condition of non-acceptance data of phase error: | Err _phase (t) | ≧ Th _phase
.. (Formula 5)

In the error detection unit 5 of the present example, when obtaining the amplitude error Err _amp (t) = | Rx (t) | / | Tx (t) |, the average of the transmission signal Tx (t) and the feedback signal Rx (t) Adjust the level so that the power is equal. (Expression 2) and (Expression 3) are examples in which the threshold is given in decibels, and real numbers may be used.
Further, in the error detection unit 5 of this example, the phase error Err _phase (t) is standardized so that Err _phase (t) = 0 when there is no phase distortion.

Note that the absolute value of the upper threshold and the absolute value of the lower threshold are not necessarily the same.
As a specific example, the absolute value of the upper threshold is different from the absolute value of the lower threshold. For example, when the threshold of the amplitude error Err _amp is given as a real number, data is selected (adopted) for the amplitude error Err _amp (t). The condition is expressed as (Equation 6), and the condition for not selecting (not _adopting ) data for the amplitude error Err _amp (t) is expressed as (Equation 7).

[Equation 6]
Conditions of adopted data of amplitude error: 0.25 ≦ Err _amp (t) ≦ 4.0
.. (Formula 6)
[Equation 7]
Conditions for non-acceptance data of amplitude error: Err _amp (t) <0.25 or 4.0 <Err _amp (t)
.. (Formula 7)

As an example of a method of determining the threshold Th _amp of the amplitude error Err _amp (t) and a method of determining the threshold Th _phase of the phase error Err _phase (t), a variance σ is obtained by taking statistics of error samples in advance, There is a method of using amplitude and phase such as 2σ and 3σ as threshold values.

5A to 5F show examples of data selection. In each case, the horizontal axis indicates an address (input power), and the vertical axis indicates an error (amplitude error or phase error). White circles indicate error samples. A curve representing the nonlinear characteristic (error) of the amplifier in the absence of noise or the like is also shown. Also, lines representing threshold values + Th _amp and -Th _amp of the amplitude error Err _amp (t) and lines representing threshold values + Th _phase and -Th _phase of the phase error Err _phase (t) are shown.

5A shows an amplitude error, FIG. 5B shows a phase error, and FIG. 5C shows an amplitude error Err _amp (t) with respect to FIG. 5A. Threshold values + Th _amp and -Th _amp are shown, and FIG. 5D shows threshold values + Th _phase and -Th _phase of the phase error Err _phase (t) with respect to FIG. 5B.

As another example, FIG. 5E shows a case where the threshold values + Th _amp and −Th _amp of the amplitude error Err _amp (t) are changed, and FIG. 5F shows the phase error Err _phase ( The case where the threshold value + Th _phase and -Th _phase of t) is changed is shown. As a specific example, the absolute value of the threshold value is decreased after the distortion compensation table (LUT) of the DPD unit is updated a predetermined number of times (for example, several times). In this case, for example, the originally desired nonlinear distortion error value becomes small and approaches zero. As a result, the absolute value of the threshold can be reduced, the accuracy can be further improved, and the distortion compensation table can be updated stably.

As described above, in FIGS. 5C to 5F, the error data in both the positive threshold line and the negative threshold line is used, and the error data outside the two lines is not used. Like that. As a result, particularly when the amplitude of the transmission signal Tx (t) is small, the effect of noise increases, so the effect is great.
Therefore, by removing data with large unnecessary signals such as noise, the contents of the distortion compensation table are stabilized, and the convergence time and the time required for tracking can be shortened.

The error mapping unit 9 is supplied with the amplitude error Err _amp (t) and the phase error Err _phase (t) and the address Address _N (t) from the data selection unit 8.
The error mapping unit 9 averages a plurality of mapping information for each address (each representative point) Address _N (t) once every update period T of the distortion compensation table (for example, a value in units of samples as in t). The error information Err _info (T) is output to the error interpolation unit 10. Here, the error mapping unit 9 maps, for example, the amplitude error Err _amp (t) and the phase error Err _phase (t) input to the input address Address _N (t), and performs a single distortion compensation table. A plurality of pieces of mapping information used for updating are grouped.

FIG. 6 shows a configuration example of the error mapping unit 9.
The error mapping unit 9 of this example includes an averaging unit 21.
The input data is a distortion compensation table address Address _N (t), an amplitude error Err _amp (t), and a phase error Err _phase (t). Input data is input in time series in units of samples.
The error mapping unit 9 has a function of obtaining an error map based on input data, and the averaging unit 21 averages a plurality of errors for the same address Address _N (t).
Here, the error map is, for example, an image like the error (white circle) shown in FIGS. 5A to 5F, but the error map is a virtual image, and it is not always necessary to generate the map. The combination (data set) that the error data of the address Address _N (t) is Err _amp (t) and Err _phase (t) may be grasped.

The error interpolation unit 10 performs interpolation of the error information Err _info (T) input from the error mapping unit 9, and, for example, amplitude errors LUT_Err _amp (T) and phase errors LUT_Err _phase (T) of all addresses in the distortion compensation table. Is output to the LUT update unit 11.
In this example, the error interpolation unit 10 uses the error information of the representative point address Address _N (t) to interpolate the error of the address between the representative points, so that the distortion compensation table (in the example of FIG. 3). 1024 address LUT).

Here, various methods may be used as the interpolation method. For example, various known techniques such as spline interpolation and Lagrangian interpolation can be used.
FIGS. 7A and 7B show examples of errors in this example.
FIG. 7A shows the amplitude error, and FIG. 7B shows the phase error. In each case, the horizontal axis indicates an address (input power), and the vertical axis indicates an error (amplitude error or phase error). White circles indicate error samples and examples of lines obtained by interpolation.
In this example, since interpolation is performed using a small number of points, for example, an approximate expression can be obtained without forcibly obtaining an approximate expression.

The LUT update unit 11 is expressed by (Equation 1) based on the amplitude error LUT_Err _amp (T) and the phase error LUT_Err _phase (T), which are errors obtained by the error interpolation unit 10 and input from the error interpolation unit 10. As described above, the amplitude information LUT _amp (t) for AM / AM compensation and the phase information LUT _phase (t) for AM / PM compensation are corrected, and these information are stored in the distortion compensation table of the DPD unit. Set. As a result, information on all addresses (1024 addresses in the example of FIG. 3) in the distortion compensation table is updated.
As the operation of the LUT update unit 11, for example, the same operation as that of the LUT update unit 128 shown in FIG. 15 can be used.

As described above, in the distortion compensation apparatus of the present example including the adaptive predistorter (adaptive predistorter), the DPD control unit includes the sample of the transmission signal Tx (t) and the sample of the feedback signal Rx (t). The error representative point aggregating unit 7 is provided for aggregating and averaging the errors in the representative points.
Further, in the distortion compensation apparatus of this example, the DPD control unit receives data for the purpose of not using the combination of the sample of the transmission signal Tx (t) and the sample of the feedback signal Rx (t) that are greatly affected by the error for adaptive control. The data selection part 8 which selects is provided.

Further, in the distortion compensation apparatus of this example, as a data selection method by the data selection unit 8, a method using a threshold value related to an amplitude error or a method using a threshold value related to a phase error is used.
Further, as another configuration example, it is possible to use a threshold value related to a correction value. For example, a method using an amplitude threshold value related to an error or a correction value or a method using a phase threshold value related to an error or a correction value. It is also possible to use it.

  Therefore, in the distortion compensation apparatus of this example, stable distortion compensation can be performed, and the initial convergence time and the time required for tracking can be shortened. Further, in the distortion compensation device of this example, sufficient distortion compensation performance can be obtained even in the case of complicated nonlinear characteristics.

As one configuration example (configuration example A1), in the predistortion distortion compensation amplification apparatus of this example, a waveform comparison type DPD that updates a distortion compensation table (LUT) using an error between an input signal (transmission signal) and a feedback signal is used. Has a function,
A distortion compensation table for storing a distortion compensation coefficient that gives predistortion to the input signal in association with an address corresponding to the input signal;
The error data between the input signal and the output signal of the amplifier (in this example, the power amplifier) is mapped to the address of the distortion compensation table corresponding to the input signal, and the distortion compensation table is calculated based on the error information obtained by mapping. A control unit for updating (in this example, a DPD control unit) is provided.
The error information is generated by an error mapping unit (in this example, the error mapping unit 9) that maps error data between the input signal and the output signal of the amplifier to an address corresponding to the input signal and averages at the address. And
As the address for mapping the error data, the 0th to MN−2nd bits are rounded down with respect to the distortion compensation table address composed of M bits (0th to (M−1) th bits) corresponding to the input signal. , M−N−1 is given by an address aggregating unit (in this example, an error representative point aggregating unit 7) that outputs an address obtained by rounding off the MN−1 bit.

In (Configuration Example A1), error data is collected at representative points before mapping. Specifically, the error data is aggregated to the representative points as a result by performing the above-described operation on the addresses mapped through the error representative point aggregating unit. In the first embodiment, representative points are taken at the stage of waveform comparison error, and interpolation is performed using the error.
Further, (Configuration Example A1) has an effect that data is aggregated only by operating addresses. For example, the error mapping unit has a function of averaging the same addresses as in the conventional case. Things can be used.
Note that the noise of the feedback signal, which is a problem in (Configuration Example A1), is, for example, general thermal noise.

As another configuration example (configuration example A2), in the predistortion distortion compensation amplification apparatus of this example, a waveform comparison type that updates a distortion compensation table (LUT) using an error between an input signal (transmission signal) and a feedback signal. Has the function of DPD,
A distortion compensation table for storing a distortion compensation coefficient that gives predistortion to the input signal in association with an address corresponding to the input signal;
The error data between the input signal and the output signal of the amplifier (in this example, the power amplifier) is mapped to the address of the distortion compensation table corresponding to the input signal, and the distortion compensation table is calculated based on the error information obtained by mapping. A control unit for updating (in this example, a DPD control unit) is provided.
The error data to be mapped is passed through an error data selection unit (in this example, the data selection unit 8) that excludes error data out of a predetermined error range from the error data of the input signal and the output signal of the amplifier. Given.

  In (Configuration Example A2), in order to reduce the influence of noise included in the acquired error, a threshold value is provided to select error data. In practice, as error data, for example, each of an amplitude error and a phase error is processed.

In the distortion compensation apparatus of this example, the coefficient storage unit is configured by the function of the distortion compensation table (LUT) of the DPD unit, and the address (Address _M (t)) detection unit is configured by the function of the amplitude detection unit 6. The function of the error representative point aggregating unit 7 constitutes an address (Address _N (t)) generating unit, the function of the data selecting unit 8 constitutes an error data selecting unit, and the error mapping unit 9 The error mapping means is configured by the above functions, and the update control means is configured by the functions of the error interpolation unit 10 and the LUT update unit 11.

A second embodiment of the present invention will be described.
FIG. 8 shows a configuration example of a DPD control unit (corresponding to the DPD control unit 108 shown in FIG. 13) using waveform comparison according to an embodiment of the present invention.
The DPD control unit of this example includes a memory 1, a quadrature detection unit 2, a memory 3, a delay adjustment unit 4, an error detection unit 5, an amplitude detection unit 6, an error representative point aggregation unit 7, a data selection unit 8, and an error mapping unit 9. , An LUT interpolation representative point updating unit 31 and an LUT interpolation unit 32 are provided.

Here, the configuration of the DPD control unit of this example includes, for example, a LUT interpolation representative point update unit 31 and a LUT interpolation unit 32 subsequent to the error mapping unit 9 as compared with the configuration shown in FIG. For convenience of explanation, similar processing units are denoted by the same reference numerals.
The DPD control unit of this example uses a method of updating the representative point of the distortion compensation table (LUT) and interpolating the distortion compensation table using the updated representative point information.

In this example, the error mapping unit 9 outputs error information Err _info (T) to the LUT interpolation representative point update unit 31.
Based on the error information Err _info (T) input from the error mapping unit 9, the LUT interpolation representative point update unit 31 updates information on only the representative point of the distortion compensation table using, for example, (Equation 1). The amplitude information LUT _amp (T) and the phase information LUT _phase (T) at the representative point are output to the LUT interpolation unit 32.

The LUT interpolation unit 32 uses the amplitude information LUT _amp (T) and the phase information LUT _phase (T) at the representative point input from the LUT interpolation representative point update unit 31 to perform interpolation (distortion compensation table information (this example)). Then, the amplitude information LUT _amp (t) for AM / AM compensation and the phase information LUT _phase (t) for AM / PM compensation are interpolated, and these pieces of information are set in the distortion compensation table of the DPD unit. .
Here, various methods may be used as the interpolation method. For example, various known techniques such as spline interpolation and Lagrangian interpolation can be used.

In the distortion compensation apparatus of this example, the coefficient storage unit is configured by the function of the distortion compensation table (LUT) of the DPD unit, and the address (Address _M (t)) detection unit is configured by the function of the amplitude detection unit 6. The function of the error representative point aggregating unit 7 constitutes an address (Address _N (t)) generating unit, the function of the data selecting unit 8 constitutes an error data selecting unit, and the error mapping unit 9 The error mapping means is configured by the above functions, and the update control means is configured by the functions of the LUT interpolation representative point updating unit 31 and the LUT interpolation unit 32.

A third embodiment of the present invention will be described.
In this example, a method for updating a distortion compensation table (LUT) using waveform comparison in a DPD (memory DPD) that compensates for a memory effect will be described.
With reference to FIGS. 9A, 9 </ b> B, and 9 </ b> C, an error when there is a memory effect will be described.
FIG. 9A shows an example of a time waveform of a transmission signal, where the horizontal axis represents time and the vertical axis represents amplitude.
FIG. 9B shows an example of a time waveform of the amplitude error, where the horizontal axis represents time and the vertical axis represents the amplitude error.
FIG. 9C shows an example of amplitude error mapping (with memory effect), where the horizontal axis represents an address and the vertical axis represents an amplitude error.

When a transmission signal having a time waveform as shown in FIG. 9A is amplified and affected by the memory effect, an error detection unit (for example, one corresponding to the error detection unit 5 shown in FIG. 1) is used. The detected amplitude error is as shown in FIG. 9B, and when this error is mapped, as shown in FIG. 9C, the generated error may be different even if the amplitudes of the transmission signals are equal. The same phenomenon occurs with respect to the phase error.
Here, in the conventional typical DPD, the average AM / AM characteristic and the AM / PM characteristic can be compensated, but the memory effect cannot be compensated. It is provided separately.

FIG. 10 shows a configuration example in which a memoryless PD unit 41 that is a DPD unit that compensates for AM / AM characteristics and AM / PM characteristics and a memory PD unit 42 that is a DPD unit that compensates for the memory effect are connected in series. It is.
Here, as the memoryless PD unit 41, for example, a predistorter (DPD unit) as shown in FIG. 14 is used.
The memory PD unit 42 in this example includes an amplitude detection unit 51, a memory effect distortion compensation table 52 including a lookup table (LUT) including a memory, a delay unit 53, an adder 54, a multiplier 55, and an adder 56. It has.

An example of the operation performed in this example is shown.
To the memoryless PD unit 41, an IQ digital baseband signal of a complex vector is input as an input signal, for example, as shown in FIG. 13 and FIG.
The signal predistorted by the memoryless PD unit 41 is output to the memory PD unit 42.

In the memory PD unit 42, the signals (I-phase component and Q-phase component) input from the memoryless PD unit 41 are input to the amplitude detection unit 51, the multiplier 55, and the adder 56.
The amplitude detector 51 detects the amplitude of the input signal and outputs the detection result (amplitude value) to the memory effect distortion compensation table 52.

The memory effect distortion compensation table 52 uses the amplitude value as a reference argument, and the distortion generated by the memory effect for the amplitude value and the distortion compensation target amplifier (for example, one corresponding to the power amplifier 104 shown in FIG. 13). And a value for compensating for (a value of a distortion compensation coefficient related to the memory effect) are stored (stored) in association with each other.
The memory effect distortion compensation table 52 outputs the distortion compensation coefficient value (complex number in this example) of the address corresponding to the amplitude value (reference argument) input from the amplitude detection unit 51 to the delay unit 53 and the adder 54. To do.

The delay unit 53 delays the distortion compensation coefficient input from the memory effect distortion compensation table 52 and outputs it to the adder 54.
The adder 54 calculates a difference between the distortion compensation coefficient input from the memory effect distortion compensation table 52 and the distortion compensation coefficient input after being delayed from the delay unit 53, and outputs the result to the multiplier 55.
The multiplier 55 multiplies the input signal (output signal from the memoryless PD unit 41) and the signal input from the adder (distortion compensation coefficient difference), and outputs the result to the adder 56. Thereby, in this example, the reverse characteristic of the distortion generated by the memory effect is expressed in the form of a sum.
The adder 56 adds the input signal (output signal from the memoryless PD unit 41) and the signal (multiplication result) input from the multiplier 55, and uses the resulting signal as a predistorter output signal. Output. This signal is output to the D / A converter 102 in the configuration shown in FIG. 13, for example.

  In the signal output in this way, the AM / AM characteristics, AM / PM characteristics of the amplifier (in this example, the power amplifier), and characteristics for compensating for distortion caused by the memory effect are input to the amplifier in advance. The signal can be applied to the signal, and the distortion of the output signal from the amplifier can be compensated.

Next, an example of a method of applying a waveform comparison to the memory DPD (conventional technology is not considered to be disclosed) is shown.
FIG. 11 shows a configuration example of the error mapping unit 61 used for compensating the memory effect.
The error mapping unit 61 of this example is the same as the DPD control unit shown in FIG. 1, FIG. 8, or FIG. 15 as the configuration of the DPD control unit that updates the memory effect distortion compensation table 52 of the memory PD unit 42. When using such a configuration, it is used as an error mapping unit (the error mapping unit 9 in FIGS. 1 and 8 or the error mapping unit 126 in FIG. 15).
Note that, for each of the memoryless PD unit 41 and the memory PD unit 42, a control unit (in this example, a DPD control unit) is provided separately.

The error mapping unit 61 of this example includes a delay unit 71, a sign inverting unit 72, a code inverting unit 73, and an averaging unit 74.
An example of the operation performed by the error mapping unit 61 of this example will be shown.
The address address (t), the amplitude error Err _amp (t), and the phase error Err _phase (t) are input to the error mapping unit 61. The address Address (t) is input to the averaging unit 74 and the delay unit 71, the amplitude error Err _amp (t) is input to the averaging unit 74 and the sign inverting unit 72, and the phase error Err _phase (t) is input to the averaging unit. 74 and the sign inversion unit 73.

For example, in the configuration of FIGS. 1 and 8, the address selection (address) (Address) (Address _N (t), amplitude error Err _amp (t) The phase error Err _phase (t) is input. Further, for example, in the configuration of FIG. 15, the amplitude error Err _amp (t) and the phase error Err _phase (t) are input to the error mapping unit 61 from the error detection unit 125 which is the previous processing unit.

The delay unit 71 delays the input address Address (t) by τ (in this example, τ samples), and outputs the address (t−τ) to the averaging unit 74 as a result.
The sign inverting unit 72 inverts the sign of the input amplitude error Err _amp (t) and outputs the result -Err _amp (t) to the averaging unit 74.
The sign inverting unit 73 inverts the sign of the input phase error Err _phase (t) and outputs the result −Err _phase (t) to the averaging unit 74.

The averaging unit 74 includes an address Address (t), an amplitude error Err _amp (t), a phase error Err _phase (t), a delayed address Address (t−τ), an inverted amplitude error −Err _amp (t), and an inversion. The phase error −Err _phase (t) is input, and a plurality of errors for the same address are averaged.

Using such a function of the averaging unit 74, the error mapping unit 61 obtains a plurality of mapping information once for each update period T of the memory effect distortion compensation table (for example, a value in units of samples as with t). On average for each address Address (t), the error information Err _info (T) is processed as a next stage processing unit (for example, the error interpolation units 10 and 127 in FIGS. 1 and 15 or the LUT interpolation representative point update unit in FIG. 8). 31).
In this example, the case where the update period T of the distortion compensation table of the memoryless PD and the update period T of the memory effect distortion compensation table of the memory PD are the same is shown. Embodiments with different periods may be used.

Here, the error mapping unit 61 of this example is different from the configuration of the error mapping unit 9 shown in FIG. 6, for example, based on the input to the error mapping unit 61, the address Address data address delayed by τ samples. The inverted amplitude error −Err _amp (t) and the inverted phase error −Err _phase (t) are mapped to (t−τ).
The delay amount τ given by the delay unit 71 is preferably equal to the delay amount given by the delay unit 53 of the memory PD unit 42, for example, depending on the characteristics of the memory effect.
Each of the sign inversion units 72 and 73 has a function of inverting the sign of the signal. For example, the signal is multiplied by −1. This sign inversion corresponds to, for example, taking a difference by the adder 54 of the memory PD unit 42.

FIG. 12 shows an example of error mapping for the memory DPD. The horizontal axis indicates an address, and the vertical axis indicates an amplitude error.
In this example, an example of mapping in the case of τ = 1 sample is shown. Although the data is handled equally, for the sake of explanation, the mapping point of Address (t) having no delay is indicated by a white circle (◯), and the mapping of Address (t−τ) having a delay is also shown. The dots are indicated by black circles (●).
Here, the mapping point (white circle) of Address (t) is equal to that shown in FIG.

Moreover, as a method for updating the memory effect distortion compensation table using such data, for example, the same method as the conventional method can be applied.
As a specific example, the memory effect distortion compensation table of the memory DPD can be converged by selecting a value of 1.0 or less as μ in (Expression 1) and repeating the update using (Expression 1). It is possible to compensate for non-linear distortion caused by the memory effect. In this manner, the memory effect distortion compensation table (LUT) of the memory DPD can be adaptively controlled using the waveform comparison.

As described above, in the distortion compensation apparatus of this example, in the DPD control unit that controls the memory effect distortion compensation table for compensating the memory effect, the amplitude error and phase error at a certain time are changed to the amplitude at the previous time. The corresponding memory effect distortion compensation table is used as an address error.
Thereby, in the distortion compensation apparatus of this example, waveform comparison can be applied to the memory DPD, and distortion generated by the memory effect can be compensated favorably.

As one configuration example (configuration example A3), in the predistortion distortion compensation amplification apparatus of this example that compensates for distortion due to the memory effect generated by an amplifier, a distortion compensation table is used by using an error between an input signal (transmission signal) and a feedback signal. (LUT) has the function of waveform comparison type DPD to update,
A memory effect distortion compensation table that stores a distortion compensation coefficient that predistorts the input signal in association with an address corresponding to the input signal; and
Error data between the input signal and the output signal of the amplifier (power amplifier in this example) is mapped to the address of the memory effect distortion compensation table corresponding to the input signal, and the memory effect is obtained based on the error information obtained by mapping. A control unit (in this example, a DPD control unit) that updates the distortion compensation table;
The error information is generated by an error mapping unit (in this example, error mapping unit 61) that maps error data between the input signal and the output signal of the amplifier to an address corresponding to the input signal and averages the data at the address. And
Here, as the error data mapped in the error mapping unit 61, error data between the input signal and the output signal of the amplifier is mapped to an address corresponding to the input signal at the corresponding time, and at the same time, Data obtained by inverting the sign of error data with respect to the output signal is mapped to an address corresponding to the input signal before the corresponding time.

  In (Configuration Example A3), when the waveform comparison DPD is applied to memory effect distortion compensation, the error mapping unit 61 maps error data to the same address in the past (for example, several samples before). Is called.

  In the distortion compensation apparatus of this example, in the examples of FIGS. 1, 8, and 15, the memory effect coefficient storage unit is configured by the function of the memory effect distortion compensation table (LUT) 52 of the DPD unit. The address detection means is configured by the function of the amplitude detection unit 6 (or the amplitude detection unit 111 of the DPD unit 101), and the address generation unit is configured by the function of the error representative point aggregation unit 7. The function of the unit 8 constitutes error data selecting means, and the function of the error mapping part 61 constitutes error mapping means. Further, in the examples of FIGS. 1 and 15, the update control means is configured by the functions of the error interpolation units 10 and 127 and the LUT update units 11 and 128, and in the example of FIG. 8, the LUT interpolation representative point update unit. The update control means is configured by the functions of 31 and the LUT interpolation unit 32.

Here, the configuration of the system and apparatus according to the present invention is not necessarily limited to the configuration described above, and various configurations may be used. The present invention can also be provided as, for example, a method or method for executing the processing according to the present invention, a program for realizing such a method or method, or a recording medium for recording the program. It is also possible to provide various systems and devices.
The application field of the present invention is not necessarily limited to the above-described fields, and the present invention can be applied to various fields.
In addition, as various processes performed in the system and apparatus according to the present invention, for example, the processor executes a control program stored in a ROM (Read Only Memory) in hardware resources including a processor and a memory. A controlled configuration may be used, and for example, each functional unit for executing the processing may be configured as an independent hardware circuit.
The present invention can also be understood as a computer-readable recording medium such as a floppy (registered trademark) disk or a CD (Compact Disc) -ROM storing the control program, and the program (itself). The processing according to the present invention can be performed by inputting the program from the recording medium to the computer and causing the processor to execute the program.

1, 3, 121, 123... Memory, 2, 122... Quadrature detection unit, 4, 124... Delay adjustment unit, 5, 125... Error detection unit, 6, 51, 111. ..Error representative point aggregating unit, 8 .... Data selection unit, 9, 61, 126..Error mapping unit, 10, 127..Error interpolation unit, 11, 128..LUT update unit, 21, 74..Average 31.. LUT interpolation representative point update unit 32.. LUT interpolation unit 41.. Memoryless PD unit 42.. Memory PD unit 52.. Memory effect distortion compensation table 53, 71. 54, 56 .. adder, 55 .. multiplier, 72, 73 .. sign inverting unit,
101 .. DPD section, 102 .. D / A converter, 103 .. Up converter (frequency conversion section), 104 .. Power amplifier, 105 .. Directional coupler, 106 .. Down converter (frequency conversion section) 107... A / D converter 108.. DPD control unit 112.. Distortion compensation table 113.. PD execution unit (for example, multiplier)

Claims (1)

In a distortion compensation device that compensates for distortion caused by a memory effect generated by an amplifier that amplifies an input signal using a predistortion method,
Coefficient storage means for storing a distortion compensation coefficient for giving predistortion to an input signal corresponding to each address at each address;
Address detecting means for detecting an address of 0th to (M−1) th M (M is an integer of 2 or more) bits corresponding to an input signal;
Using the set N (N is an integer smaller than M), the bit values from 0 to (M−N−2) -th are rounded down for the M-bit address detected by the address detecting means, and (M− N-1) address generating means for generating an address obtained by rounding off the 0th bit value;
Error data selection means for selecting error data within a set range for error data between the input signal and the output signal of the amplifier;
The error data between the input signal selected by the error data selection means and the output signal of the amplifier is mapped to the address generated by the address generation means for the input signal and selected by the error data selection means. The data obtained by inverting the sign of the error data between the input signal and the output signal of the amplifier is mapped to the address generated by the address generating means for the input signal in the past of the input signal , and the average for each address And error mapping means for generating error information based on the result,
Update control means for updating the storage content of the coefficient storage means based on the error information generated by the error mapping means;
A distortion compensation apparatus comprising:

Images