JP5120178B2 - Nonlinear system inverse characteristic identification device and method, power amplification device, and power amplifier predistorter - Google Patents

Description

  The present invention relates generally to nonlinear systems, such as nonlinear power amplifiers, and further to nonlinear power amplifiers that can be used in transmitters (eg, base stations and mobile stations) of wireless communication systems.

  As one of nonlinear systems, a power amplifier is an important component in many electronic devices, and can amplify a minute electric signal to meet the demand for remote transmission. Here, the energy to be amplified is supplied by a DC power source. That is, the power amplifier can convert DC energy into an AC signal.

According to the physical characteristics of the power amplifier, the characteristic graph indicating the voltage relationship between the input signal and the output signal in the power amplifier is divided into a linear region, a non-linear region, and a saturation region as the voltage of the input signal increases. FIG. 1 shows nonlinear input / output signal characteristics of the power amplifier PA. The region smaller amplitude of the input signal V _IN, the output V _OUT of the power amplifier PA is almost a linear amplification to the input signal V _IN, according to the amplitude of the input signal V _IN increases, the nonlinear characteristic of the power amplifier PA It gradually becomes clear and finally reaches saturation. This non-linearity is expressed by the fact that the spectrum of the amplified signal spreads out of the band due to intermodulation in the frequency domain, and distortion occurs in the band. FIG. 2 schematically shows the spread of the spectrum by a non-linear power amplifier.

  Ideally, the power amplifier is desired to perform only the function of linear amplification, i.e., the output signal is simply a linear amplification of the input signal, so that the power amplifier operates in the linear region. In this case, if the efficiency of converting a DC signal into an AC signal in the power amplifier becomes very low, a large amount of energy is wasted, and it may be considered to add an extra heat dissipation device.

  Therefore, on the one hand, to increase the efficiency of the power amplifier, and on the other hand, the signals in many modern communication systems have a large dynamic range (ratio of average power to peak power), so power amplifiers are often in the nonlinear region. It is made to work. This causes distortion of the output signal (in the frequency domain, distortion occurs in the band and spectrum spread occurs outside the band). In the field of power amplifiers, the distortion of a nonlinear power amplifier in which the current output signal is affected only by the influence of the current input signal is sometimes referred to as a memoryless non-linear characteristic of the power amplifier.

As one of the methods for improving the non-linear characteristics of the power amplifier, a baseband predistortion technique has attracted attention. FIG. 3 shows a schematic diagram of input / output signal characteristics of the power amplifier predistorter for canceling out the nonlinear characteristics shown in FIG. The baseband predistortion technology simulates the inverse characteristics of the power amplifier PA using the predistorter PD, applies predistortion before inputting the original input signal _VIN to the power amplifier, and improves the nonlinear characteristics of the power amplifier PA. Compensation is performed to obtain an amplified output signal V _OUT without distortion on the output side of the power amplifier. The characteristic of the input / output signal of the power amplifier having the predistortion processing function represents an excellent linear characteristic until it approaches the saturation region. FIG. 4 shows a schematic diagram of input / output signal characteristics of a power amplifier including a predistorter.

  Many of the predistorters that perform well are for single tone or narrow bandwidth signals. That is, most of them are used for memoryless nonlinear compensation of the power amplifier. However, as the signal bandwidth increases, the power amplifier may exhibit some memory effect (frequency selectivity). That is, the current output signal is not only related to the current input signal, but may also be related to the previous input signal. FIG. 5 shows a schematic diagram of input / output signal characteristics of a nonlinear power amplifier having a memory effect. The memory effect of the power amplifier represents an asymmetric spectrum near the carrier at its output. FIG. 6 shows a schematic diagram of a spectrum of a signal amplified by a non-linear power amplifier having a memory effect. That is, even if the spectrum of the carrier (desired signal) is completely symmetric, the pseudo spectrum due to distortion is asymmetric with respect to the center carrier.

  Therefore, in order to accurately simulate the inverse characteristics of the power amplifier having the memory effect described above, it is necessary to model the inverse characteristics of the power amplifier and estimate the parameters with a more complicated configuration and method. Predistortion is realized. Currently, for example, a memory polynomial model and a two-dimensional lookup table method are known as techniques for simultaneously compensating for both the memoryless nonlinear characteristic and the memory effect of a power amplifier.

  The memory polynomial model is a simplified version of the Volterra model. The Volterra model uses the Volterra series to create an accurate model for nonlinear systems, and simulates the inverse Volterra series model of the power amplifier to theoretically offset the memoryless nonlinear characteristics and memory effects of the power amplifier. can do. The Volterra model is effective in offsetting the memoryless nonlinear characteristics and memory effects of power amplifiers, but it uses complex equations and requires a large amount of mathematical calculations, so its physical realization It is difficult. Therefore, a memory polynomial model, which is a simplified Volterra model, has been submitted. FIG. 7 shows a schematic diagram of a predistortion addition process for an input signal of a nonlinear power amplifier having a memory effect, using a memory polynomial model according to the prior art. Here, the output of the memory polynomial model is displayed by the current input, the previous input, and a correlated polynomial function. The ability to cancel the nonlinearity of the power amplifier depends on the number of previous inputs used and the order of the polynomial. In order to accurately describe the nonlinear characteristics of the power amplifier, a polynomial including a high-order term is used. Therefore, it leads to a matrix with many conditions, and numerical calculation becomes difficult.

  In the two-dimensional look-up table method, the inverse characteristics of the power amplifier are created in the look-up table according to the relationship between the current output of the power amplifier and its current input and the previous input. The non-linear characteristic of the power amplifier is canceled by applying predistortion to the input signal using the weighting factor acquired from the lookup table. FIG. 8 is a schematic diagram showing a predistortion addition process for an input signal of a nonlinear power amplifier having a memory effect using a two-dimensional lookup table according to the conventional technique. However, in a two-dimensional lookup table, not only is it necessary to store a large amount of predistortion data in a large capacity memory, but also because the memory length is limited, it is often the case that the current input and the immediately preceding data are stored. Only by the relationship between the input and the output, predistortion is added to the input signal of the power amplifier to cancel the memory effect of the power amplifier. If the memory length is increased, the multi-dimensional look-up table becomes very complex and difficult to implement.

  The above technical problems also exist in other nonlinear systems similar to the nonlinear power amplifier.

  Therefore, we want to be able to effectively reduce the demand for storage devices and numerical calculations.

  In addition, by using digital technology to achieve predistortion in the baseband, the powerless amplifier compensates for the memoryless nonlinear characteristics and memory effect of the power amplifier, and effectively reduces the demand for storage devices and numerical calculations. We would also like to provide a reverse characteristic identification apparatus and method, a power amplifier predistorter, and a power amplifier reverse characteristic identification apparatus and power amplification apparatus using the predistorter.

に従って、オリジナル入力信号x(n)に対する逆フィルタリングを行って逆フィルタ信号x_Ｆ(n)を生成する逆フィルタステップと、設定された条件を満たすまで、以上の各ステップを繰り返すことを含み、ルックアップテーブル関数Q{・}とフィルタ関数F{・}とは、非線形システムの逆特性を表す。 According to an example of the present invention, there is provided a nonlinear system inverse characteristic identification method for identifying an inverse characteristic of a nonlinear system from an original input signal x (n) of the nonlinear system and a corresponding feedback output signal y _G (n). The nonlinear system inverse characteristic identification method includes a lookup table function Q {•} generation step for calculating a lookup table function Q {•} from the inverse filter signal x _F (n) and the feedback output signal y _G (n). Intermediate predistortion added signal z (n) generating step for generating intermediate predistortion added signal z (n) by feedback output signal y _G (n) and look-up table function Q {·}, and intermediate predistortion added signal Using the filter function F {•} creation step to create the filter function F {•} using z (n) and the original input signal x (n), and the parameters of the filter function F {•},

And performing an inverse filtering step on the original input signal x (n) to generate an inverse filter signal x _F (n), repeating the above steps until a set condition is satisfied, The uptable function Q {•} and the filter function F {•} represent the inverse characteristics of the nonlinear system.

に従って、オリジナル入力信号x(n)に対する逆フィルタリングを行って逆フィルタ信号x_Ｆ(n)を生成する逆フィルタとを含む。 According to another example of the present invention, there is provided a non-linear system inverse characteristic identification device for identifying an inverse characteristic of a non-linear system from an original input signal x (n) of the non-linear system and a corresponding feedback output signal y _G (n). The non-linear system inverse characteristic identification device identifies a reverse characteristic of the non-linear system and generates a look-up table function Q {•} and a filter function F {•} representing the inverse characteristic of the non-linear system. And whether the lookup table function Q {•} and the filter function F {•} generated in the parameter calculator satisfy the setting condition, and the lookup table function Q {•} and the filter function F { And a convergence determination unit for controlling the parameter calculator so as to calculate repeatedly, the parameter calculator based on the inverse filter signal x _F (n) and the feedback output signal y _G (n). An intermediate predistortion by a lookup table generator for calculating a lookup table function Q {·}, a feedback output signal y _G (n) and a lookup table function Q {·} calculated by the lookup table generator Create filter function F {•} from intermediate predistorter that generates additional signal z (n), intermediate predistortion additional signal z (n) generated by intermediate predistorter, and original input signal x (n) Filter parameter calculator to be used and the filter function F {•} created by the filter parameter calculator,

And an inverse filter that performs inverse filtering on the original input signal x (n) to generate an inverse filter signal x _F (n).

  According to still another example of the present invention, a predistorter for a power amplifier is provided. The predistorter for the power amplifier takes an absolute value and performs quantization on an original input signal. To a lookup table address, and a one-dimensional lookup table that is arranged according to the lookup table function Q {·} and outputs a correction coefficient according to a lookup table address output from the address generator; , A multiplier that multiplies the original input signal by the correction coefficient output from the one-dimensional lookup table to obtain an intermediate predistorted signal, and an intermediate that is arranged according to the filter function F {·} and output from the multiplier Including a filter for performing filtering on the predistorted signal and generating a predistorted signal, and a look-up table The number and Q {·} and filter function F {·}, generated by the nonlinear system inverse characteristic identification methods described above, or is generated by the nonlinear system inverse characteristic identification device.

  According to another example of the present invention, a power amplifying device is provided, and the power amplifying device performs predistortion on an original input signal and outputs a predistorted input signal. A digital-to-analog converter that converts the predistorted input signal output from the predistorter module into an analog signal; an up-converter that up-converts the analog signal output from the digital-to-analog converter into an RF signal; A power amplifier that amplifies the RF signal output from the up-converter and outputs a power-amplified output signal, and the configuration of the predistorter module includes a predistorter of the power amplifier limited to the above Same as the configuration.

  According to another example of the present invention, a power amplifying device is provided, which is configured in the same manner as the power amplifier predistorter limited to the above, and predistorted with respect to the original input signal. The predistorter module that outputs the predistortion added input signal, the digital / analog converter that converts the predistortion input signal output from the predistorter module into an analog signal, and the output from the digital analog converter An up-converter that upconverts the analog signal thus converted into an RF signal, a power amplifier that amplifies the RF signal output from the up-converter, and outputs a power-amplified output signal, and two outputs of the power amplifier. Branches to a signal, outputs one of the signals as an output signal, and feeds the other signal to the attenuator A directional coupler for backing up, an attenuator for receiving a signal fed back from the directional coupler, attenuating the feedback signal, and a downconverter for downconverting the attenuated signal output from the attenuator to a baseband signal An analog-to-digital converter that converts the baseband signal output from the downconverter into a digital signal to be a feedback output signal, a delay device that delays the original input signal to generate a delayed input signal, and an analog・ Receiving the feedback output signal output from the digital converter and the delayed input signal output from the delay unit, and using the received feedback output signal and the delayed input signal, the lookup table function and the filter coefficient are online. Predistortion data updater .

〔ここで、F{・}、Q{・}、P{・}は、それぞれフィルタのフィルタ関数、ルックアップテーブルのルックアップテーブル関数と電力増幅器の増幅関数を表し、ただし、

ルックアップテーブル関数Q{・}は、次式

であり、
フィルタ係数は、次式

であり、
Lは、フィルタの記憶長さを表し、[|・|]は、絶対値取りと量子化とを表す〕
に従って誤差信号を算出すること、
及び、下記式：

〔ここで、

Uとμ_ｑの両方とも、収束までのステップ数である〕
により、ルックアップテーブル関数とフィルタ係数とをオンラインで更新すること、
を含む。 According to another example of the present invention, a method for online updating parameters of a predistorter module of a power amplifying device is provided, which method includes the following steps:
Compare the original input signal x (n) of the power amplifier and the feedback output signal, the following formula:

[Where F {•}, Q {•}, and P {•} represent the filter function of the filter, the lookup table function of the lookup table, and the amplification function of the power amplifier, respectively, where

The lookup table function Q {·} is given by

And
The filter coefficient is

And
L represents the storage length of the filter, and [| · |] represents absolute value acquisition and quantization)
Calculating an error signal according to
And the following formula:

〔here,

(Both U and μ _q are the number of steps to convergence)
To update lookup table functions and filter coefficients online,
including.

  According to the present invention, it is possible to avoid difficulty in numerical calculation due to the high condition number of the matrix.

The present invention relates to a nonlinear system, for example, an inverse characteristic identification device and method for a nonlinear power amplifier, a power amplifier predistorter, and an inverse characteristic identification device for a nonlinear system and a power amplifying apparatus including the predistorter, using digital technology Thus, by realizing predistortion in the baseband, the non-memory (memoryless) nonlinear characteristic and the memory effect of the nonlinear power amplifier are compensated.
Embodiments of the present invention will be described below with reference to the drawings. In the drawings, the same or corresponding technical features or components are displayed with the same or corresponding reference numerals.

  Identification of the inverse characteristic of the nonlinear system according to the present embodiment is performed based on a normal Hammerstein model. The Hammerstein model is a non-linear model constructed in consideration of the memory effect of a non-linear system, and is composed of a memoryless non-linear model and a linear time-varying system in series. FIG. 9 shows a schematic diagram of a configuration for constructing the inverse characteristic of the nonlinear system 94 using the Hammerstein model 90. As shown in FIG. 9, the Hammerstein model 90 includes an address generator 91, a look-up table (LUT) 92, a FIR (finite impulse response) filter 93, and a multiplier 95. The address generator 91 generates an address for searching the LUT 92 based on the input signal x (n), and selects a coefficient displayed in a complex form from the LUT 92 based on the address. The product obtained by multiplying the input signal x (n) and the complex coefficient by the multiplier 95 is digitally filtered by the FIR filter 93 to be the final predistorted signal u (n). Here, k represents the address of the LUT 92.

  The one-dimensional LUT is used to compensate for the memoryless nonlinear characteristics of the nonlinear system, and the filter is used to compensate for the memory effect of the nonlinear system. As a result, it is possible to avoid occupying a large amount of memory and using higher-order polynomials, and to reduce the amount of calculation. Here, we will consider how to establish a one-dimensional LUT and FIR filter parameters, and to better match the overall nonlinear characteristics of the nonlinear system by combining them. Therefore, this embodiment employs an “indirect learning method”. FIG. 10 shows an example of an indirect learning method for acquiring the inverse characteristic of the nonlinear system applied to this embodiment.

In general, the characteristics of nonlinear systems do not change frequently over time, so the learning algorithm and the predistortion model A identification module can use some input and output data sampling points (eg, n inputs). After data sampling points and n output data sampling points) are collected, data processing can be performed offline. After identification, new parameters are sent to the predistortion model A module of the nonlinear system so as to compensate for the overall nonlinearity of the nonlinear system. As shown in FIG. 10, in this process, when the data is updated off-line, the sampling points of the collected input and output data are used, and the error signal e (n) is minimized to reverse the characteristics of the nonlinear system. (Estimate the parameters of model A). In FIG. 10, x (n) is the input of the nonlinear system, y (n) is the output of the nonlinear system, G is the desired gain factor of the nonlinear system, and y _G (n) Is called the feedback output of the nonlinear system. The predistortion model A module can be realized by various programmable devices such as a field programmable gate array (FPGA), and the time-varying parameters of the nonlinear system can be easily updated in the field.

  FIG. 11 is a schematic diagram illustrating an example of an identification apparatus that acquires the inverse characteristics of the nonlinear system offline by the indirect learning method, that is, an example of the learning algorithm and the predistortion model A identification module illustrated in FIG. This apparatus is based on the indirect learning method shown in FIG. The data collection / storage module 301 in FIG. 3 sends the input / output data of the collected and stored nonlinear system to the identification module 300. The identification module 300 maps and outputs these data, and then the adder 306 The error signal e (n) is used as a basis for correcting the parameters of the identification module 300 by comparing with the input data of the nonlinear system stored in the collection storage module 301.

  The identification module 300 includes an address generator 302, a lookup table 303, a multiplier 304 and a dual filter 305. The address generator 302, lookup table 303, multiplier 304 and dual filter 305 here correspond to the address generator 91, lookup table 92, multiplier 95 and FIR filter 93 shown in FIG. A specific operation example will be described in the following sentence.

ここで、x(n)は、図１０に示す入力信号を表す。図１０に示す非線形システムの所望利得がGであり、非線形システムの出力がy(n)である場合、y_G(n)は、次式

で表示される。

は、同定信号（入力信号x(n)の推定値）を表し、以下のように表示される。

中間予歪付加信号z(n)は、以下のようである。

ここで、F{・}は、フィルタ３０５のフィルタ関数を表し、Q{・}は、ルックアップテーブル３０３のルックアップテーブル関数を表す。
[|・|]は、アドレス発生器３０２による絶対値取りと量子化とを表す。すなわち、

ｋは、ルックアップテーブルのアドレス（例えば０〜２５５）に対応する。 As can be seen from FIG. 11, the error signal e (n) is displayed as follows.

Here, x (n) represents the input signal shown in FIG. When the desired gain of the nonlinear system shown in FIG. 10 is G and the output of the nonlinear system is y (n), y _G (n) is given by

Is displayed.

Represents an identification signal (estimated value of the input signal x (n)) and is displayed as follows.

The intermediate predistortion added signal z (n) is as follows.

Here, F {·} represents the filter function of the filter 305, and Q {·} represents the lookup table function of the lookup table 303.
[| • |] represents absolute value acquisition and quantization by the address generator 302. That is,

k corresponds to an address (for example, 0 to 255) of the lookup table.

ここで、

である。 Therefore, the identification of the inverse characteristics of the nonlinear system will minimize the error signal e (n) by designing the filter 305 and the parameters of the lookup table 303. Since e (n) is time series, this identification is described as follows.

here,

It is.

  In order to effectively minimize the function E, F {·} and Q {·} are combined and optimized as an example of a suitable or optimal method. In the present embodiment, optimization calculation of F {•} and Q {•} is alternately performed to minimize E, and optimized F {•} and Q {•} (filter 305 and lookup table 303 ) And a calculation method that approaches the inverse characteristics of a nonlinear system.

First, the function Q {•} is analyzed and calculated. The contents of the lookup table are obtained by analyzing the gain of each sampling point. Here, the gain of each sampling point is displayed as follows.

のそれぞれは、一つの

に対応し、すなわち、一つのルックアップテーブルのアドレスｋに対応する。同じアドレスｋを有する

を１カテゴリに分類（１グループに分類）し、以下のように表示する。

That is, each

Each is one

That is, it corresponds to the address k of one lookup table. Have the same address k

Are classified into one category (classified into one group) and displayed as follows.

に対して、それぞれ、その絶対値と位相の平均を取ることで、折衷の利得特性を取得することができる。すなわち、

である。 For each group

On the other hand, by taking the average of the absolute value and the phase, it is possible to obtain a compromise gain characteristic. That is,

It is.

を取得することができる。
ここで、

である。
また、

は、Hadamard積を表し、上付き文字Tは、マトリックスの転置を表す。 And the lookup table function Q {•}

Can be obtained.
here,

It is.
Also,

Represents the Hadamard product, and the superscript T represents the transpose of the matrix.

  After obtaining the look-up table function Q {·} by Expression (12), the intermediate predistortion added signal z (n) can be calculated by Expression (4).

の絶対値と位相それぞれの平均値である。他のアルゴリズム、例えば、メジアン法、最小二乗法などを使用して当該折衷の利得特性を取得してもよい。ここでは、その具体的な説明を省略する。 When calculating the gain characteristics of eclectic as described above, what is selected for each group

Is the average value of each of the absolute value and the phase. The gain characteristics of the compromise may be obtained using other algorithms such as a median method and a least square method. Here, the specific description is omitted.

を取得する。
ここで、

であり、ここで、Nは、収集されたデータの長さであり、Lは、FIRフィルタの階数（メモリ深度）である。 Next, the filter function F {•} is analyzed and calculated. This can be obtained by the least square method. After obtaining the intermediate predistorted signal z (n), the filter coefficient is obtained by the least square method in order to obtain the optimum filter function F {•} and to approximate the input signal x (n).

To get.
here,

Where N is the length of the collected data and L is the rank (memory depth) of the FIR filter.

After calculating and obtaining the filter parameter W by equation (15), the filter function F {•} is displayed in the form of a matrix and a vector as follows.

を取得して、式（１）により誤差信号e(n)を算出することができる。誤差信号e(n)が設定された条件を満たす場合、入力信号x(n)の推定値

が要求を満たすと考えられる。これは、現在に使用されるルックアップテーブル関数Q{・}とフィルタ関数F{・}とが、非線形システムの特性によく適合し、図１０に示すプリディストーションモデルAのパラメータに適用することができることを意味する。 Then, the estimated value of the input signal x (n) is obtained by Equation (3) and Equation (18).

And the error signal e (n) can be calculated by equation (1). Estimated value of input signal x (n) when error signal e (n) satisfies the set condition

Is considered to meet the requirements. This is because the currently used look-up table function Q {•} and filter function F {•} are well suited to the characteristics of the nonlinear system and can be applied to the parameters of the predistortion model A shown in FIG. Means you can.

が要求を満すことができないため、さらにルックアップテーブル関数Q{・}とフィルタ関数F{・}とを修正する必要がある。この場合、フィルタ３０５を利用して、次式（１９）に従って入力信号x(n)に対する逆フィルタリングを行って、逆フィルタ信号

を得る。

On the other hand, if the error signal e (n) does not satisfy the set condition, the estimated value of the input signal x (n)

Cannot satisfy the request, it is necessary to further modify the lookup table function Q {•} and the filter function F {•}. In this case, the filter 305 is used to perform inverse filtering on the input signal x (n) according to the following equation (19) to obtain an inverse filter signal.

Get.

とともに、式（８）中のオリジナル入力信号x(n)の代わりに当該逆フィルタ信号

を用いて、式（４）、式（５）及び式（１２）〜式（１８）を利用して、改めてルックアップテーブル関数Q{・}とフィルタ関数F{・}とを算出する。そして、設定された条件を満たすまで、さらに、式（１）により誤差信号e(n)を算出する。 And the feedback output signal of the nonlinear system

And the inverse filter signal instead of the original input signal x (n) in equation (8).

Are used to calculate the look-up table function Q {•} and the filter function F {•} again using the equations (4), (5), and (12) to (18). Then, until the set condition is satisfied, the error signal e (n) is further calculated by the equation (1).

Here, the set condition is that after obtaining the error signal e (n), it is checked whether or not the normalized mean square error NMSE of the following equation (20) converges, or a threshold with a certain number of repetitions It can be determined whether or not it becomes larger.

  Of course, for those skilled in the art, the set condition is not limited to the normalized mean square error NMSE referred to above, but may be other judgment standards such as the mean square error MSE, the peak mean square error PMSE, etc. Should be able to understand.

  FIG. 12 shows a flowchart of the optimization process performed by the identification module 300 shown in FIG. 11, that is, how the input signal x (n) and the output signal y (n) of the nonlinear system are used. It shows whether the function Q {·} of the lookup table 303 describing the inverse characteristic of the nonlinear system and the function F {·} of the filter 305 are acquired.

とを取得する。 As shown in FIG. 12, first, in step S1201, the original input signal x (n) of the nonlinear system and the output signal that has been fed back and attenuated through the processing of the nonlinear system and processed by the identification module 300

And get.

Next, in step S1202, for each sampling point, an absolute value for the output signal y _G (n) is taken, a quantization operation is performed, and the output signal y _G (n) at each sampling point is handled by Equation (5). Get the address number k. In step S1203, the first normalized mean square error NMSE ₀ is set, and the iteration variable i is set to 0.

を取得し、式（９）により、各サンプリングポイントの利得を、対応の出力信号アドレスに応じていくつかのグループに分ける。次に、式（１０）〜式（１４）により、ルックアップテーブル関数Q{・}を取得する。 Next, in step S1204, grouping and averaging are performed according to equations (8) to (14) to obtain a lookup table function Q {·}. First, the gain at each sampling point is calculated by equation (8).

And the gain of each sampling point is divided into several groups according to the corresponding output signal address according to equation (9). Next, the look-up table function Q {·} is obtained from the equations (10) to (14).

  After calculating and obtaining the look-up table function Q {·}, in step S1205, the intermediate predistortion added signal z (n) is calculated by Equation (4). Thereafter, in step S1206, 1 is added to the repetition variable i. That is, i = i + 1.

を算出して、式（２０）により正規化平均二乗誤差NMSE_iを算出する。 In step S1207, the filter is designed to obtain the filter function F {·}, and the normalized mean square error NMSE _i is calculated. First, a filter matrix Z _M is created by Expression (17), and a filter function F {·} is acquired by Expression (15). Then, the estimated value of the input signal x (n) is obtained by Equation (3) and Equation (18).

And the normalized mean square error NMSE _i is calculated according to equation (20).

After obtaining the normalized mean square error NMSE _i, in step S1208, normalized mean square error NMSE _i whether converge, or to determine whether greater than a certain threshold number of repetitions. If so, the process proceeds to step S1209 to output the current look-up table function Q {•} and filter function F {•} to the field programmable gate array FPGA so as to realize a predistorter in hardware. .

を取得する。 If the normalized mean square error NMSE _i does not converge and the number of iterations is not greater than a certain threshold, the process proceeds to step S1210, and a filter is used to determine the input signal x (n). Do an inverse filter

To get.

を用い、式（４）、式（５）及び式（１２）〜式（１８）を利用して、改めてルックアップテーブル関数Q{・}とフィルタ関数F{・}とを算出する。続いて、正規化平均二乗誤差NMSE_iが収束し又は繰り返し数がある閾値より大きくなるまで、式（２０）により、正規化平均二乗誤差NMSE_iを算出する。 Then, the process returns to step S1204 and repeats steps S1204 to S1208. Along with the feedback output signal y _G (n) of the nonlinear system, the inverse filter signal instead of the original input signal x (n) in equation (8)

Are used to calculate the look-up table function Q {•} and the filter function F {•} again using the equations (4), (5), and (12) to (18). Subsequently, the normalized mean square error NMSE _i is calculated by Expression (20) until the normalized mean square error NMSE _i converges or becomes larger than a certain threshold.

  Hereinafter, the configuration of a suitable non-linear system inverse characteristic identification device for realizing the identification method will be described in detail. FIG. 13 is a block diagram of the nonlinear system inverse characteristic identification device according to this embodiment.

The data storage device 810 shown in FIG. 13 stores the original input signal x (n) and the attenuated feedback output signal y _G (n), which are transmitted to the parameter calculator 800. Is done. The parameter calculator 800 includes an address generator 801, a lookup table generator 802, a filter parameter calculator 803, an intermediate predistorter 804, a filter 805, an inverse filter 806, a selector 807, and the like.

が入力され、式（８）〜式（１４）によりルックアップテーブル関数Q{・}を生成し、生成されたルックアップテーブル関数Q{・}は中間プリディストータ８０４に出力される。 The feedback output signal y _G (n) is input to the parameter calculator 800 and then converted into an address k by the address generator 801 according to the equation (5). The converted address k is input to the lookup table generator 802 together with the feedback output signal y _G (n). The lookup table generator 802 also includes a signal x (n) from the selector 807 or an inverse filter signal.

Is generated, and the lookup table function Q {•} is generated by the equations (8) to (14), and the generated lookup table function Q {•} is output to the intermediate predistorter 804.

The intermediate predistorter 804 receives the look-up table function Q {·} and the feedback output signal y _G (n) from the look-up table generator 802, and receives the intermediate predistortion added signal z ( n) is generated, and the generated intermediate predistortion added signal z (n) is output to the filter parameter calculator 803 and the filter 805, respectively.

  The filter parameter calculator 803 receives the intermediate predistortion added signal z (n) and the original input signal x (n) from the intermediate predistorter 804, and uses the filter function F {・} Is generated. The generated filter function F {·} is output to the filter 805 and the inverse filter 806, respectively.

を生成し、生成されたオリジナル入力信号の推定値

をNMSE算出器８０８に出力する。 The filter 805 processes the intermediate predistortion added signal z (n) generated by the intermediate predistorter 804 according to the expression (3) by the filter function F {·} generated by the filter parameter calculator 803 to obtain the original Input signal estimate

And an estimate of the generated original input signal

Is output to the NMSE calculator 808.

とオリジナル入力信号x(n)とを受信し、式（２０）により正規化平均二乗誤差NMSEを算出して、収束判断器８１１に出力する。 The NMSE calculator 808 estimates the original input signal generated by the filter 805.

And the original input signal x (n) are calculated, the normalized mean square error NMSE is calculated by the equation (20), and is output to the convergence determination unit 811.

をセレクタ８０７に出力する。 The inverse filter 806 installed symmetrically with the filter 805 also receives the filter function F {•} from the filter parameter calculator 803, and uses the filter function F {•} to perform an original input signal x (n). Inverse filter signal with inverse filtering

Is output to the selector 807.

を選択し、必要に応じてその両者の何れかをルックアップテーブル生成器８０２に出力する。具体的には、セレクタ８０７は、通常、第一回目にルックアップテーブル関数Q{・}とフィルタ関数F{・}とを算出する場合、オリジナル入力信号x(n)をルックアップテーブル生成器８０２に出力するが、その後の繰り返し算出において逆フィルタ８０６によって生成された逆フィルタ信号

を選択してルックアップテーブル生成器８０２に出力する。 The selector 807 receives the original input signal x (n) and the inverse filter signal.

And outputs either of them to the lookup table generator 802 as necessary. Specifically, the selector 807 normally calculates the lookup table function Q {•} and the filter function F {•} for the first time using the original input signal x (n) as the lookup table generator 802. , But the inverse filter signal generated by the inverse filter 806 in the subsequent iterative calculation

Is output to the lookup table generator 802.

  The convergence determination unit 811 records the normalized mean square error NMSE and the number of repetition steps, and uses the normalized average square error NMSE or the number of repetition steps as a determination condition. When the determination condition is not satisfied, that is, when the normalized mean square error NMSE does not converge or the number of repetition steps does not exceed a predetermined threshold, the convergence determination unit 811 performs a parameter so that the above process is repeated. The calculator 800 is controlled. When the determination condition is satisfied, that is, when the normalized mean square error NMSE converges or the number of iteration steps exceeds a predetermined threshold, the convergence determination unit 811 performs the latest calculation and obtains the filter function F {· } And the parameter calculator 800 are controlled so as to output the lookup table function Q {•}.

  In the following, taking a nonlinear power amplifier as an example, how to use the nonlinear system inverse characteristic identification method and identification device as described above to compensate for the memoryless nonlinear characteristic and memory effect of the power amplifier, the linear amplification region is Whether to enlarge will be described in detail. That is, how to perform predistortion on a nonlinear power amplifier by using the above-described nonlinear system inverse characteristic identification method and identification device as described above will be described.

  First, in order to identify the inverse characteristics of the nonlinear power amplifier, it is necessary to collect input / output data of the nonlinear power amplifier. FIG. 14 shows a block diagram of an apparatus for collecting power amplifier input / output data. Here, the power amplifier is an example of a power amplifier used for a transmitter of a base station and a mobile station in a communication system, but the power amplifier can be applied to other cases as well.

As shown in FIG. 14, a device for collecting input / output data of a power amplifier includes a digital / analog converter 101 that converts a digital input signal x (n) into an analog signal (baseband signal), and a digital / analog conversion. Up-converter 102 that up-converts the baseband signal output from device 101 to an RF signal, power amplifier 103 that amplifies the RF signal output from up-converter 102, and the output of power amplifier 103 is branched into two signals. One signal is output as an output signal y (n), the other signal is fed back to the attenuator 105, and the output of the power amplifier 103 fed back from the directional coupler 104 is output. An attenuator 105 that receives and attenuates the feedback signal, and an attenuation signal output from the attenuator 105 Down converter 106 for computing downconversion to baseband signal, a digital signal a baseband signal which is obtained by coating downconverting by a down converter 106, i.e. feedback output signal y _G analog-to-digital converter 107 for converting the (n), A data acquisition and storage module 100 that collects and stores the original input signal x (n) and the feedback output signal y _G (n) is included. The original input signal x (n) and the feedback output signal y _G (n) stored in the data acquisition and storage module 100 are displayed as vectors as follows.

After collecting the original input signal x (n) and feedback output signal y _G (n) of the power amplifier using the apparatus shown in FIG. 14, the nonlinear system inverse characteristic identification method shown in FIG. 12 or the nonlinearity shown in FIG. Using the system inverse characteristic identification device, the inverse characteristic of the power amplifier can be identified to obtain the filter function F {·} and the lookup table function Q {·}.

  When the filter function F {·} and the look-up table function Q {·} are obtained, the power amplifier predistorter module is constructed and predistorted with respect to the power amplifier input signal x (n). The amplifier can obtain an excellent linear output. FIG. 15 shows a block diagram of a non-linear power amplifier capable of realizing predistortion according to the present embodiment.

  As shown in FIG. 15, the nonlinear power amplifying apparatus capable of realizing the predistortion performs predistortion on an input signal x (n) and outputs a predistortion added signal, and a predistorter module 400. A digital / analog converter 405 that converts the predistorted signal output from the tota module 400 into an analog signal (baseband signal), and a baseband signal output from the digital / analog converter 405 is upconverted to an RF signal. The up-converter 406 includes a power amplifier 407 that amplifies the RF signal output from the up-converter 406 and outputs a power-amplified output signal y (n). In this case, the signal y (n) output from the power amplifying device has an excellent linear characteristic from which the memoryless nonlinear characteristic and the memory effect of the power amplifier are removed.

  The predistorter module 400 shown in FIG. 15 takes an absolute value for the input signal x (n), performs quantization, and then converts it into a look-up table address. A one-dimensional lookup table 402 which is arranged according to the lookup table function Q {·} obtained by the method or the identification device and outputs a correction coefficient based on the lookup table address outputted from the address generator 401, original Multiplier 403 that multiplies input signal x (n) by the correction coefficient output from one-dimensional lookup table 402 to obtain an intermediate predistortion added signal, obtained using the identification method or identification device as described above. Filtering the intermediate predistorted signal output from the multiplier 403 and arranged according to the filter function F {·}. Achieving pre-distortion, comprising a filter 404 for outputting a predistorted additional signal to the digital-to-analog converter 405.

  Further, in order to construct a power amplifier having both a data collection and storage function and a predistortion function, a directional coupler 104, an attenuator 105 in the apparatus for collecting input / output data of the power amplifier shown in FIG. The down converter 106, the analog / digital converter 107, and the data collection / storage module 100 may be incorporated in the power amplification apparatus shown in FIG.

  According to the nonlinear power amplifying apparatus including the predistorter module arranged as described above, predistortion is realized in the baseband using digital technology, and the memoryless nonlinear characteristic and the memory effect of the power amplifier are compensated. In addition, the demand for the capacity of the storage device can be effectively reduced, and the demand for numerical calculation can be greatly reduced.

As described above, by using the power amplifier input signal x (n) and feedback output signal y _G (n) collected and stored in advance, the reverse characteristic of the power amplifier is identified offline, and the power amplifier is identified based on the identification result. The predistorter module is arranged.

  However, as mentioned above, the non-linear power amplifier is one of the non-linear systems, and its output characteristics change with time. Such changes are not apparent within a short time, but their characteristics always change over time, so the power amplifier's inverse characteristics are regularly identified so that its output always represents a good linear form. It is desirable.

  In order to avoid complicated operations associated with periodically identifying the reverse characteristics of the power amplifier and periodically updating the parameters of the predistorter module, this embodiment further includes an online configuration of the predistorter module. A power amplification method and apparatus for updating and upgrading parameters is provided. FIG. 16 shows a configuration diagram of a power amplifying apparatus that can update and upgrade the parameters of the predistorter module online according to the present embodiment.

  16 includes a predistorter module 400, a digital / analog converter 405, an up converter 406, a power amplifier 407, a directional coupler 104, an attenuator 105, a down converter 106, and an analog / digital converter 107. , Predistortion data updater 600, and delay unit 601. Each unit other than the predistortion data updater 600 and the delay unit 601 is denoted by the same reference numerals as those of the corresponding units in FIGS. 14 and 15, and the detailed description thereof will be omitted.

  In the power amplifying apparatus having a module for online updating the predistorter module shown in FIG. 16, first, the original digital input signal x (n) is input to the predistorter module 400 and the absolute value is obtained by the address generator 401. After being quantized, it is converted to a look-up table address. Based on this lookup table address, the correction coefficient is read from the one-dimensional lookup table 402 and the correction coefficient and the original digital input signal x (n) are multiplied by the multiplier 403. The result of the multiplication is further filtered by the filter 404, the predistortion process is completed, and a predistorted signal is generated. The predistorted signal is converted into a radio frequency analog signal by the digital / analog converter 405 and the up converter 406, transmitted to the power amplifier 407, power amplified, and output as a power amplified signal y (n). The

  The power amplified signal y (n) is fed back by the directional coupler 104, attenuated by the attenuator 105, and then the signal is downconverted to baseband by the downconverter 106, and then an analog to digital converter. The baseband signal is converted into a digital signal by 107 and input to the predistortion data updater 600.

  One branch of the original digital input signal x (n) is also delayed by the delay unit 601 and then input to the predistortion data updater 600. The predistortion data updater 600 corresponds to the identification module 300 shown in FIG. 11, and updates the filter function F {·} and the lookup table function Q {·} of the predistorter module 400 by the following algorithm.

ここで、P{・}は電力増幅器関数を表し、また、

であり、ルックアップテーブル関数Q{・}は、以下のように表示される。

フィルタ係数は、以下のように整理される。

Lは、メモリ深度を表す。
そして、ルックアップテーブル関数及びフィルタ係数は、次式により更新することができるようになる。

ここで、

であり、Uとμ_ｑの両方とも、収束ステップ長さである。 As can be seen from FIG. 16, the following error signal is obtained by comparing the original digital input signal x (n) and the fed back digital output signal.

Where P {·} represents the power amplifier function, and

And the lookup table function Q {•} is displayed as follows.

The filter coefficients are organized as follows.

L represents the memory depth.
Then, the look-up table function and the filter coefficient can be updated by the following equations.

here,

And both U and μ _q are convergence step lengths.

The predistortion data updater 600 uses the nonlinear inverse characteristic identification method as described above to obtain the inverse characteristic of the power amplifier using the latest acquired original input signal x (n) and feedback output signal y _G (n). The real-time lookup table function Q {•} and the filter function F {•} may be acquired by online identification.

  Acquiring new look-up table functions and filter functions allows the corresponding parameters in the predistorter module 400 to be modified. Here, the detailed description of the flow of the processing method executed by the predistorter module 400 and the predistortion data updater 600 shown in FIG. 16 is omitted.

  According to the identification apparatus and identification method disclosed in the present embodiment, it is possible to simultaneously reduce the requirements for the storage device and the numerical calculation. Further, online updating of the parameters of the predistorter module can be realized, and as a result, the linear characteristics of the power amplifier can be secured in real time.

  As described above, taking the nonlinear power amplifier as an example, detailed description has been given of how to perform inverse characteristic identification of a time-varying nonlinear system offline and online and to perform predistortion on the nonlinear system based on the identification result.

  As described above, the present embodiment describes the nonlinear characteristics of a power amplifier in series with a finite impulse response (FIR) filter and a single one-dimensional lookup table, so that many of the memory polynomial models used. Avoid accurately describing the nonlinear characteristics of the power amplifier in higher order terms. This is because the use of higher-order terms increases the condition number of the corresponding matrix, causing difficulties in the accurate acquisition and predistortion of memory polynomial parameters. In addition, the power amplifier predistorter according to the present embodiment and the power amplifier using the power amplifier reverse characteristic identification device and the predistorter employ a one-dimensional look-up table, which requires a memory capacity requirement. Reduce significantly.

  Further, the above series of processing and apparatus may be realized by software and / or firmware. When realized by software and / or firmware, a program constituting the software is installed from a storage medium or a network into a computer having a dedicated hardware configuration, for example, a general-purpose personal computer 700 shown in FIG. When this program is installed, various functions can be executed.

  In FIG. 17, a central processing unit (CPU) 701 executes various processes by a program stored in a read-only memory (ROM) 702 or a program loaded from a storage unit 708 to a random access memory (RAM) 703. . The RAM 703 stores data necessary for executing various processes of the CPU 701 as necessary.

  The CPU 701, ROM 702, and RAM 703 are connected to each other via a bus 704. An input / output interface 705 is also connected to the bus 704.

  The input / output interface 705 includes an input unit 706 including a keyboard and a mouse, an output unit 707 including a display and a speaker such as a cathode ray tube (CRT) and a liquid crystal display (LCD), and a hard disk. A storage unit 708 including a communication unit 709 including a network interface card including, for example, a LAN card and a modem is connected. The communication unit 709 executes communication processing via a network such as the Internet.

  A driver 710 is also connected to the input / output interface 705 as necessary. For example, a removable medium 711 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is attached to the driver 710 as necessary, and a computer program read from the medium 711 is stored in the storage unit 708 as necessary. To be installed.

  When the above-described series of processing is realized by software, a program constituting the software is installed from a network such as the Internet or a storage medium such as a removable medium 711, for example.

  Such a storage medium is not limited to the removable medium 711 shown in FIG. 17 in which the program is stored and which provides the program to the user by delivering it away from the device. Examples of the removable medium 711 include a magnetic disk (including a floppy disk (registered trademark)), an optical disk (including a compact disk read-only memory (CD-ROM) and a digital versatile disk (DVD)), and a magneto-optical disk ( Including a mini disk (MD) (registered trademark), and a semiconductor memory. Alternatively, the storage medium may be a ROM 702, a hard disk included in the storage unit 708, or the like. A program is stored in the hard disk and delivered to a user together with a device including these programs.

  In addition, the steps for executing the series of processes described above may be executed in time order according to the order of description, but may not necessarily be executed in time order. Certain steps may be performed in parallel or separately from each other.

に従って、オリジナル入力信号x(n)に対する逆フィルタリングを行って逆フィルタ信号x_F(n)を生成する逆フィルタステップと、
設定された条件を満たすまで、以上の各ステップを繰り返すことを含み、
ルックアップテーブル関数Q{・}とフィルタ関数F{・}とは、非線形システムの逆特性を表す、ことを特徴とする、非線形システム逆特性同定方法。
（付記２）
第一回目にルックアップテーブル関数Q{・}を生成する際に、フィードバック出力信号y_G(n)とともに、オリジナル入力信号x(n)を逆フィルタ信号x_F(n)として、ルックアップテーブル関数Q{・}を算出する、ことを特徴とする、付記１に記載の非線形システム逆特性同定方法。
（付記３）
前記設定された条件は、次式

〔ここで、式における

は、次式

によりフィルタ関数F{・}と中間予歪付加信号z(n)とを使用して生成された、オリジナル入力信号x(n)の推定値

である〕
に示す正規化平均二乗誤差NMSEが収束すること、又は繰り返し数が規定の閾値より大きいこと、を確定することである、ことを特徴とする、付記２に記載の非線形システム逆特性同定方法。
（付記４）
次式

〔ここで、[|・|]は、絶対値取りと量子化とを表す〕
によりフィードバック出力信号y_G(n)をルックアップテーブルにおける一つのアドレスｋに量子化する絶対値取りと量子化ステップをさらに含む、ことを特徴とする、付記３に記載の非線形システム逆特性同定方法。
（付記５）
ルックアップテーブル関数Q{・}生成ステップは、
式

に従って、各サンプリングポイント毎の利得を算出すること、
同じアドレスを有する

を、次式

に従って１グループに分類することで、ルックアップテーブルの１アドレスｋに対応するようにすること、及び
各グループ毎の利得を取得して、ルックアップテーブル関数Q{・}を生成すること、
を含む、ことを特徴とする、付記４に記載の非線形システム逆特性同定方法。
（付記６）
各グループ毎の利得を取得するステップは、
次式

により各グループ毎の

に対して、それぞれ、その絶対値と位相の平均を取って、ルックアップテーブル関数Q{・}

〔ここで、

であり、

は、ハダマード（Hadamard）積を表し、上付き文字Tは、マトリックスの転置を表す〕
を取得する、ことを特徴とする、付記５に記載の非線形システム逆特性同定方法。
（付記７）
中間予歪付加信号z(n)生成ステップにおいて、次式

によりフィードバック出力信号y_G(n)とルックアップテーブル関数Q{・}とを使用して中間予歪付加信号z(n)を生成する、ことを特徴とする、付記６に記載の非線形システム逆特性同定方法。
（付記８）
フィルタ関数F{・}作成ステップにおいて、
次式

〔ここで、

であり、Nは、収集されたデータの長さであり、Lは、フィルタの階数である〕
により、中間予歪付加信号z(n)とオリジナル入力信号x(n)とを使用して、フィルタ係数を作成し、
これによって、フィルタ関数F{・}は、マトリックスとベクトルの形式で次式

のように表示される、ことを特徴とする、付記７に記載の非線形システム逆特性同定方法。
（付記９）
設定された条件を満たす場合、ルックアップテーブル関数Q{・}とフィルタ関数F{・}とを非線形システムの逆特性として出力する、ことを特徴とする、付記８に記載の非線形システム逆特性同定方法。
（付記１０）
フィードバック出力信号y_G(n)は、次式

により生成され、Gは、非線形システムの所望利得であり、y(n)は、非線形システムの出力である、ことを特徴とする、付記１〜９の何れか一項に記載の非線形システム逆特性同定方法。
（付記１１）
非線形システムに特定のオリジナル入力信号x(n)を入力し、相応の出力信号y(n)を取得するデータ収集ステップをさらに含む、ことを特徴とする、付記１０に記載の非線形システム逆特性同定方法。
（付記１２）
非線形システムのオリジナル入力信号x(n)、および相応のフィードバック出力信号y_G(n)により、非線形システムの逆特性を同定する非線形システム逆特性同定装置であって、
非線形システムの逆特性を同定して、非線形システムの逆特性を表すルックアップテーブル関数Q{・}とフィルタ関数F{・}とを生成するパラメータ算出器と、
パラメータ算出器で生成されたルックアップテーブル関数Q{・}とフィルタ関数F{・}とが設定条件を満たすか否かを判定し、ルックアップテーブル関数Q{・}とフィルタ関数F{・}とを繰り返し算出するようにパラメータ算出器を制御する収束判断器とを含み、
前記パラメータ算出器は、
逆フィルタ信号x_F(n)とフィードバック出力信号y_G(n)とによりルックアップテーブル関数Q{・}を算出するルックアップテーブル生成器と、
フィードバック出力信号y_G(n)とルックアップテーブル生成器で算出されたルックアップテーブル関数Q{・}とにより中間予歪付加信号z(n)を生成する中間プリディストータと、
中間プリディストータで生成された中間予歪付加信号z(n)とオリジナル入力信号x(n)とによりフィルタ関数F{・}を作成するフィルタパラメータ算出器と、
フィルタパラメータ算出器で作成されたフィルタ関数F{・}を使用し、式

に従って、オリジナル入力信号x(n)に対する逆フィルタリングを行って逆フィルタ信号x_F(n)を生成する逆フィルタとを含む、ことを特徴とする、非線形システム逆特性同定装置。
（付記１３）
さらにセレクタを含み、このセレクタは、第一回目にルックアップテーブル関数Q{・}を生成する際に、オリジナル入力信号x(n)を逆フィルタ信号x_F(n)として選択し、ルックアップテーブル生成器に入力して、ルックアップテーブル生成器にオリジナル入力信号x(n)とフィードバック出力信号y_G(n)とによりルックアップテーブル関数Q{・}を算出させるようにし、その後にルックアップテーブル関数Q{・}を生成する際に、逆フィルタで生成された逆フィルタ信号x_F(n)を選択し、ルックアップテーブル生成器に入力して、ルックアップテーブル生成器に逆フィルタ信号x_F(n)とフィードバック出力信号y_G(n)とに基づいてルックアップテーブル関数Q{・}を算出させるのに用いられる、ことを特徴とする、付記１２に記載の非線形システム逆特性同定装置。
（付記１４）
式

により、フィルタパラメータ算出器で作成されたフィルタ関数F{・}と中間予歪付加信号z(n)を使用してオリジナル入力信号x(n)の推定値

を生成するフィルタと、
次式

に従って正規化平均二乗誤差NMSEを算出する正規化平均二乗誤差NMSE算出器とをさらに含み、ここで、収束判断器は、正規化平均二乗誤差NMSEが収束するか否か又は繰り返し数が規定の閾値より大きいか否かによって、設定された条件を満たすか否かを判定する、ことを特徴とする、付記１３に記載の非線形システム逆特性同定装置。
（付記１５）
次式

〔ここで、[|・|]は、絶対値取りと量子化とを表す〕
により、フィードバック出力信号y_G(n)をルックアップテーブル生成器で生成されたルックアップテーブルにおける一つのアドレスｋに量子化するアドレス発生器をさらに含む、ことを特徴とする、付記１４に記載の非線形システム逆特性同定装置。
（付記１６）
そのルックアップテーブル生成器は、
式

に従って各サンプリングポイント毎の利得を算出し、
同じアドレスを有する

を、次式

により１グループに分類することで、ルックアップテーブルの１アドレスｋに対応するようにし、各グループ毎の利得を取得して、ルックアップテーブル関数Q{・}を生成するように構成される、ことを特徴とする、付記１５に記載の非線形システム逆特性同定装置。
（付記１７）
そのルックアップテーブル生成器は、
次式

により、各グループ毎の

に対して、それぞれ、その絶対値と位相の平均を取って、ルックアップテーブル関数Q{・}

〔ここで、

は、Hadamard積を表し、上付き文字Tは、マトリックスの転置を表す〕
を取得するように構成される、ことを特徴とする、付記１６に記載の非線形システム逆特性同定装置。
（付記１８）
中間プリディストータは、次式

により、フィードバック出力信号y_G(n)とルックアップテーブル生成器で算出されたルックアップテーブル関数Q{・}とを使用して、中間予歪付加信号z(n)を生成する、ことを特徴とする、付記１７に記載の非線形システム逆特性同定装置。
（付記１９）
フィルタパラメータ算出器は、次式

〔ここで、

Nは、収集されたデータの長さであり、Lは、フィルタの階数である〕
により、中間プリディストータで生成された中間予歪付加信号z(n)とオリジナル入力信号x(n)とを使用して、フィルタ係数を作成し、これによって、フィルタ関数F{・}は、マトリックスとベクトルの形式で次式

のように表示される、ことを特徴とする、付記１８に記載の非線形システム逆特性同定装置。
（付記２０）
設定された条件を満たす場合、ルックアップテーブル関数Q{・}とフィルタ関数F{・}とを非線形システムの逆特性として出力する、ことを特徴とする、付記１９に記載の非線形システム逆特性同定装置。
（付記２１）
そのフィードバック出力信号y_G(n)は、次式

〔Gは、非線形システムの所望利得であり、y(n)は、非線形システムの出力である〕
により生成される、ことを特徴とする、付記１２〜２０のいずれか一項に記載の非線形システム逆特性同定装置。
（付記２２）
非線形システムに特定の入力信号x(n)を入力し、相応の出力信号y(n)を取得するデータ収集記憶装置をさらに含む、ことを特徴とする、付記２１に記載の非線形システム逆特性同定装置。
（付記２３）
電力増幅器のプリディストータであって、
オリジナル入力信号に対して、絶対値を取り、量子化を行ってから、ルックアップテーブルアドレスに変換するアドレス発生器と、
ルックアップテーブル関数Q{・}に従って配置され、アドレス発生器から出力されたルックアップテーブルアドレスにより、補正係数を出力する一次元ルックアップテーブルと、
オリジナル入力信号を一次元ルックアップテーブルから出力された補正係数に乗算して、中間予歪付加信号を取得する乗算器と、
フィルタ関数F{・}に従って配置され、乗算器から出力された中間予歪付加信号に対するフィルタリングを行って、予歪付加信号を生成するフィルタとを含み、
ルックアップテーブル関数Q{・}とフィルタ関数F{・}とは、付記１〜９のいずれか一項に記載の非線形システム逆特性同定方法により生成され、又は、付記１２〜２０のいずれか一項に記載の非線形システム逆特性同定装置により生成される、ことを特徴とする、電力増幅器のプリディストータ。
（付記２４）
電力増幅器のオリジナル入力信号とフィードバック出力信号とを収集するデータ収集記憶装置をさらに含み、
当該データ収集メモリ装置は、
オリジナル入力信号をアナログ信号に変換するデジタル・アナログ変換器と、
デジタル・アナログ変換器から出力されるアナログ信号をRF信号にアップコンバーティングするアップコンバータと、
アップコンバータから出力されるRF信号を電力増幅する電力増幅器と、
電力増幅器の出力を二つ分岐の信号に分岐して、その一つの信号を出力信号として出力し、他の一つの信号を減衰器にフィードバックする方向性結合器と、
方向性結合器からフィードバックされる信号を受信し、当該フィードバック信号を減衰する減衰器と、
減衰器から出力された減衰信号をベースバンド信号にダウンコンバーティングするダウンコンバータと、
ダウンコンバータから出力されたベースバンド信号をデジタル信号に変換して、フィードバック出力信号とするアナログ・デジタル変換器と、
オリジナル入力信号とフィードバック出力信号を収集して記憶するデータ収集記憶モジュールとを含む、ことを特徴とする、付記２３に記載の電力増幅器のプリディストータ。
（付記２５）
電力増幅装置であって、
オリジナル入力信号に対してプリディストーションを行って、予歪付加信号を出力するプリディストータモジュールと、
プリディストータモジュールから出力された予歪付加信号をアナログ信号に変換するデジタル・アナログ変換器と、
デジタル・アナログ変換器から出力されたアナログ信号をRF信号にアップコンバーティングするアップコンバータと、
アップコンバータから出力されたRF信号を電力増幅して、電力増幅された信号を出力する電力増幅器とを含み、
前記プリディストータモジュールの構成は、付記２３に記載の電力増幅器のプリディストータの構成と同じである、ことを特徴とする、電力増幅装置。
（付記２６）
電力増幅器の出力を二つの信号に分岐して、その一つの信号を出力信号として出力し、他の一つの信号を減衰器にフィードバックする方向性結合器と、
方向性結合器からフィードバックされた信号を受信し、当該フィードバック信号を減衰する減衰器と、
減衰器から出力された減衰信号をベースバンド信号にダウンコンバーティングするダウンコンバータと、
ダウンコンバータから出力されたベースバンド信号をデジタル信号に変換して、フィードバック出力信号とするアナログ・デジタル変換器と、
オリジナル入力信号とフィードバック出力信号を収集して記憶するデータ収集記憶モジュールとを含む、ことを特徴とする、付記２５に記載の電力増幅装置。
（付記２７）
電力増幅装置であって、
付記２３に記載の電力増幅器のプリディストータと同じ構成を有し、オリジナル入力信号に対してプリディストーションを行って予歪付加入力信号を出力するのに用いられるプリディストータモジュールと、
プリディストータモジュールから出力される予歪付加信号をアナログ信号に変換するデジタル・アナログ変換器と、
デジタル・アナログ変換器から出力されるアナログ信号をRF信号にアップコンバーティングするアップコンバータと、
アップコンバータから出力されたRF信号を電力増幅して、電力増幅された信号を出力する電力増幅器と、
電力増幅器の出力を二つの信号に分岐して、その一つの信号を出力信号として出力し、他の一つの信号を減衰器にフィードバックする方向性結合器と、
方向性結合器からフィードバックされた信号を受信し、当該フィードバック信号を減衰する減衰器と、
減衰器から出力された減衰信号をベースバンド信号にダウンコンバーティングするダウンコンバータと、
ダウンコンバータから出力されたベースバンド信号をデジタル信号に変換して、フィードバック出力信号とするアナログ・デジタル変換器と、
オリジナル入力信号を遅延して、遅延入力信号を生成する遅延器と、
アナログ・デジタル変換器から出力されたフィードバック出力信号と遅延器から出力された遅延入力信号とを受信し、受信されたフィードバック出力信号と遅延入力信号とにより、ルックアップテーブル関数とフィルタ係数とをオンラインで更新するプリディストーションデータ更新器とを含む、ことを特徴とする、電力増幅装置。
（付記２８）
プリディストーションデータ更新器は、付記１〜９のいずれか一項に記載の非線形システム逆特性同定方法により、プリディストータモジュールのルックアップテーブル関数とフィルタ関数とをオンラインで更新するように構成される、ことを特徴とする、付記２７に記載の電力増幅装置。
（付記２９）
プリディストーションデータ更新器は、下記のような方式により、プリディストータモジュールのルックアップテーブル関数とフィルタ関数とをオンラインで更新するように構成され、すなわち、
オリジナル入力信号x(n)とフィードバック出力信号とを比較して、次式

〔ここで、F{・}、Q{・}、P{・}は、それぞれフィルタ関数、ルックアップテーブル関数、電力増幅器関数を表し、

であり、ルックアップテーブル関数Q{・}は、次式

であり、
フィルタ係数は、次式

であり、
Lは、フィルタの記憶長さを表し、[|・|]は、絶対値取りと量子化とを表す〕
に従って誤差信号を算出し、次式

〔ここで、

であり、Uとμ_ｑの両方とも、収束までのステップ数である〕
により、ルックアップテーブル関数とフィルタ係数とをオンラインで更新する、ことを特徴とする、付記２７に記載の電力増幅装置。
（付記３０）
電力増幅装置のプリディストータモジュールのパラメータをオンラインで更新する方法であって、
電力増幅装置のオリジナル入力信号x(n)とフィードバック出力信号とを比較して、次式

〔ここで、F{・}、Q{・}、P{・}は、それぞれフィルタのフィルタ関数、ルックアップテーブルのルックアップテーブル関数、電力増幅器の増幅関数を表し、ただし

であり、ルックアップテーブル関数Q{・}は、次式

であり、
フィルタ係数は、次式

であり、
Lは、フィルタの記憶長さを表し、[|・|]は、絶対値取りと量子化とを表す〕
に従って誤差信号を算出し、次式

〔ここで、

であり、Uとμ_ｑの両方とも、収束までのステップ数である〕
により、ルックアップテーブル関数とフィルタ係数とをオンラインで更新することを含む、ことを特徴とする、電力増幅装置のプリディストータモジュールのパラメータをオンラインで更新する方法。 Although this embodiment and its advantages have been described in detail, it should be understood that various changes, replacements and conversions can be made without departing from the spirit and scope of the matters described in the claims. It is. And “having”, “including” or any other variation is meant to cover non-exclusive “including”. Thus, a process, method, article, or device that includes a series of elements includes not only those elements, but also other elements not explicitly described, or such processes, methods, articles, or devices Includes unique elements. In the absence of more restrictions, an element that is limited by the phrase “including one” is that there are other similar elements in the process, method, article or device that include the element. Are not excluded.
Regarding the above embodiment, the following additional notes are disclosed.
(Appendix 1)
A nonlinear system inverse characteristic identification method for identifying an inverse characteristic of a nonlinear system from an original input signal x (n) of the nonlinear system and a corresponding feedback output signal y _G (n),
A lookup table function Q {•} generation step for calculating a lookup table function Q {•} from the inverse filter signal x _F (n) and the feedback output signal y _G (n);
An intermediate predistortion added signal z (n) generating step for generating an intermediate predistortion added signal z (n) by the feedback output signal y _G (n) and the lookup table function Q {·};
A filter function F {•} creating step for creating a filter function F {•} from the intermediate predistortion added signal z (n) and the original input signal x (n);
Use the filter function F {

An inverse filter step of performing inverse filtering on the original input signal x (n) to generate an inverse filtered signal x _F (n),
Including repeating each of the above steps until the set conditions are met,
The nonlinear system inverse characteristic identification method, wherein the lookup table function Q {·} and the filter function F {·} represent inverse characteristics of the nonlinear system.
(Appendix 2)
When generating the look-up table function Q {·} First time, along with the feedback output signal y _G (n), the original input signal x (n) as an inverse filter signal x _F (n), the look-up table function The nonlinear system inverse characteristic identification method according to attachment 1, wherein Q {·} is calculated.
(Appendix 3)
The set condition is expressed by the following formula:

[Where

Is

Is an estimate of the original input signal x (n) generated using the filter function F {•} and the intermediate predistorted signal z (n)

Is)
The non-linear system inverse characteristic identification method according to appendix 2, characterized in that the normalized mean square error NMSE shown in (2) converges or that the number of repetitions is larger than a prescribed threshold value.
(Appendix 4)
Next formula

[Where [| ・ |] represents absolute value acquisition and quantization]
The non-linear system inverse characteristic identification method according to appendix 3, further comprising: taking an absolute value and quantizing the feedback output signal y _G (n) into one address k in the lookup table according to .
(Appendix 5)
The lookup table function Q {·} generation step is
formula

To calculate the gain for each sampling point according to
Have the same address

The following formula

Categorizing into one group according to the above, so as to correspond to one address k of the lookup table, and obtaining a gain for each group to generate a lookup table function Q {·},
The non-linear system inverse characteristic identification method according to appendix 4, characterized by comprising:
(Appendix 6)
The step of obtaining the gain for each group is as follows:
Next formula

For each group

For each, take the average of its absolute value and phase, and look up the table function Q {•}

〔here,

And

Represents the Hadamard product, and the superscript T represents the transpose of the matrix)
The non-linear system inverse characteristic identification method according to appendix 5, characterized in that:
(Appendix 7)
In the intermediate predistortion added signal z (n) generation step,

The non-linear system inverse according to appendix 6, wherein the intermediate predistortion added signal z (n) is generated by using the feedback output signal y _G (n) and the look-up table function Q {·} by Characteristic identification method.
(Appendix 8)
In the filter function F {·} creation step,
Next formula

〔here,

Where N is the length of the collected data and L is the filter rank)
To create a filter coefficient using the intermediate predistortion added signal z (n) and the original input signal x (n),
This allows the filter function F {·} to be expressed in the form of matrix and vector:

The non-linear system inverse characteristic identification method according to appendix 7, wherein the nonlinear system inverse characteristic identification method is displayed.
(Appendix 9)
The nonlinear system inverse characteristic identification according to appendix 8, characterized in that when the set condition is satisfied, the lookup table function Q {·} and the filter function F {·} are output as inverse characteristics of the nonlinear system. Method.
(Appendix 10)
The feedback output signal y _G (n) is

10. The nonlinear system inverse characteristic according to any one of appendices 1 to 9, wherein G is a desired gain of the nonlinear system, and y (n) is an output of the nonlinear system. Identification method.
(Appendix 11)
The nonlinear system inverse characteristic identification according to appendix 10, further comprising a data collection step of inputting a specific original input signal x (n) to the nonlinear system and obtaining a corresponding output signal y (n). Method.
(Appendix 12)
A nonlinear system inverse characteristic identification device that identifies the inverse characteristic of a nonlinear system from the original input signal x (n) of the nonlinear system and the corresponding feedback output signal y _G (n),
A parameter calculator that identifies the inverse characteristics of the nonlinear system and generates a look-up table function Q {•} and a filter function F {•} representing the inverse characteristics of the nonlinear system;
It is determined whether the lookup table function Q {•} and the filter function F {•} generated by the parameter calculator satisfy the setting condition, and the lookup table function Q {•} and the filter function F {•} And a convergence determiner that controls the parameter calculator to repeatedly calculate
The parameter calculator is
A look-up table generator for calculating a look-up table function Q {•} from the inverse filter signal x _F (n) and the feedback output signal y _G (n);
An intermediate predistorter that generates an intermediate predistortion added signal z (n) from the feedback output signal y _G (n) and the lookup table function Q {·} calculated by the lookup table generator;
A filter parameter calculator that creates a filter function F {•} from the intermediate predistortion added signal z (n) generated by the intermediate predistorter and the original input signal x (n);
Using the filter function F {•} created by the filter parameter calculator,

And an inverse filter that performs inverse filtering on the original input signal x (n) to generate an inverse filter signal x _F (n).
(Appendix 13)
The selector further includes a selector that selects the original input signal x (n) as the inverse filter signal x _F (n) when generating the lookup table function Q {·} for the first time, Input to the generator and let the lookup table generator calculate the lookup table function Q {•} from the original input signal x (n) and the feedback output signal y _G (n), and then the lookup table When generating the function Q {·}, the inverse filter signal x _F (n) generated by the inverse filter is selected and input to the lookup table generator, and the inverse filter signal x _F is input to the lookup table generator. The nonlinear system inverse characteristic identification device according to appendix 12, characterized in that the lookup table function Q {·} is calculated based on (n) and the feedback output signal y _G (n). .
(Appendix 14)
formula

The estimated value of the original input signal x (n) using the filter function F {•} created by the filter parameter calculator and the intermediate predistorted signal z (n)

A filter that generates
Next formula

And a normalized mean square error NMSE calculator for calculating a normalized mean square error NMSE according to the threshold value, wherein the convergence judging unit determines whether the normalized mean square error NMSE converges or the number of repetitions is a prescribed threshold value. 14. The nonlinear system inverse characteristic identification device according to appendix 13, wherein it is determined whether or not a set condition is satisfied depending on whether or not it is larger.
(Appendix 15)
Next formula

[Where [| ・ |] represents absolute value acquisition and quantization]
15. The supplementary note 14, further comprising: an address generator that quantizes the feedback output signal y _G (n) into one address k in the lookup table generated by the lookup table generator. Nonlinear system reverse characteristic identification device.
(Appendix 16)
The lookup table generator is
formula

To calculate the gain for each sampling point according to
Have the same address

The following formula

By classifying into one group by the above, it is configured to correspond to one address k of the lookup table, obtain the gain for each group, and generate the lookup table function Q {·}. The non-linear system inverse characteristic identification device according to appendix 15, characterized by:
(Appendix 17)
The lookup table generator is
Next formula

By each group

For each, take the average of its absolute value and phase, and look up the table function Q {•}

〔here,

Represents the Hadamard product, and the superscript T represents the transpose of the matrix)
The non-linear system inverse characteristic identification device according to appendix 16, wherein the non-linear system inverse characteristic identification device is configured to acquire the following.
(Appendix 18)
The intermediate predistorter is

To generate an intermediate predistorted signal z (n) using the feedback output signal y _G (n) and the lookup table function Q {·} calculated by the lookup table generator. The non-linear system inverse characteristic identification device according to appendix 17.
(Appendix 19)
The filter parameter calculator is

〔here,

(N is the length of the collected data, L is the filter rank)
By using the intermediate predistortion added signal z (n) generated by the intermediate predistorter and the original input signal x (n), a filter coefficient is created, whereby the filter function F {·} is In matrix and vector form

The non-linear system inverse characteristic identification device according to appendix 18, characterized by being displayed as follows:
(Appendix 20)
The nonlinear system inverse characteristic identification according to appendix 19, wherein when the set condition is satisfied, the lookup table function Q {·} and the filter function F {·} are output as inverse characteristics of the nonlinear system. apparatus.
(Appendix 21)
The feedback output signal y _G (n) is given by

[G is the desired gain of the nonlinear system and y (n) is the output of the nonlinear system]
The nonlinear system inverse characteristic identification device according to any one of appendices 12 to 20, wherein the nonlinear system inverse property identification device is generated by:
(Appendix 22)
The nonlinear system inverse characteristic identification according to appendix 21, further comprising a data acquisition and storage device for inputting a specific input signal x (n) to the nonlinear system and obtaining a corresponding output signal y (n). apparatus.
(Appendix 23)
A power amplifier predistorter,
An address generator that takes an absolute value for the original input signal, performs quantization, and then converts it to a lookup table address;
A one-dimensional look-up table that is arranged according to the look-up table function Q {·} and outputs a correction coefficient according to the look-up table address output from the address generator;
A multiplier for multiplying the original input signal by the correction coefficient output from the one-dimensional lookup table to obtain an intermediate predistorted signal;
A filter that is arranged according to a filter function F {•} and performs filtering on the intermediate predistorted signal output from the multiplier to generate a predistorted signal,
The look-up table function Q {·} and the filter function F {·} are generated by the nonlinear system inverse characteristic identification method according to any one of appendices 1 to 9, or any one of appendices 12 to 20. A predistorter for a power amplifier, characterized in that the predistorter is generated by the non-linear system inverse characteristic identification device according to the item.
(Appendix 24)
A data acquisition and storage device for collecting the original input signal and the feedback output signal of the power amplifier;
The data collection memory device is
A digital-to-analog converter that converts the original input signal into an analog signal;
An up-converter that up-converts the analog signal output from the digital-to-analog converter into an RF signal;
A power amplifier that amplifies the RF signal output from the upconverter; and
A directional coupler that branches the output of the power amplifier into two branched signals, outputs the one signal as an output signal, and feeds back the other one signal to the attenuator;
An attenuator that receives a signal fed back from the directional coupler and attenuates the feedback signal;
A down converter that down-converts the attenuated signal output from the attenuator into a baseband signal;
An analog-to-digital converter that converts the baseband signal output from the downconverter into a digital signal and uses it as a feedback output signal;
24. The predistorter for a power amplifier according to appendix 23, comprising: a data acquisition and storage module for collecting and storing an original input signal and a feedback output signal.
(Appendix 25)
A power amplifying device,
A predistorter module that predistorts the original input signal and outputs a predistorted signal;
A digital-to-analog converter that converts the predistorted signal output from the predistorter module into an analog signal;
An up-converter that up-converts the analog signal output from the digital-to-analog converter into an RF signal;
A power amplifier that amplifies the RF signal output from the up-converter and outputs the power-amplified signal;
The configuration of the predistorter module is the same as the configuration of the predistorter of the power amplifier described in appendix 23.
(Appendix 26)
A directional coupler that branches the output of the power amplifier into two signals, outputs the one signal as an output signal, and feeds back the other signal to the attenuator;
An attenuator that receives the signal fed back from the directional coupler and attenuates the feedback signal;
A down converter that down-converts the attenuated signal output from the attenuator into a baseband signal;
An analog-to-digital converter that converts the baseband signal output from the downconverter into a digital signal and uses it as a feedback output signal;
26. The power amplifying apparatus according to appendix 25, further comprising: a data collection storage module that collects and stores an original input signal and a feedback output signal.
(Appendix 27)
A power amplifying device,
A predistorter module having the same configuration as that of the power amplifier predistorter according to appendix 23, used for predistorting the original input signal and outputting a predistorted input signal;
A digital-to-analog converter that converts the predistorted signal output from the predistorter module into an analog signal;
An up-converter that up-converts the analog signal output from the digital-to-analog converter into an RF signal;
A power amplifier that amplifies the RF signal output from the upconverter and outputs a power amplified signal;
A directional coupler that branches the output of the power amplifier into two signals, outputs the one signal as an output signal, and feeds back the other signal to the attenuator;
An attenuator that receives the signal fed back from the directional coupler and attenuates the feedback signal;
A down converter that down-converts the attenuated signal output from the attenuator into a baseband signal;
An analog-to-digital converter that converts the baseband signal output from the downconverter into a digital signal and uses it as a feedback output signal;
A delay unit that delays the original input signal to generate a delayed input signal;
Receives the feedback output signal output from the analog-to-digital converter and the delayed input signal output from the delay circuit, and uses the received feedback output signal and delayed input signal to make the lookup table function and filter coefficient online And a predistortion data updater for updating at a power amplifier.
(Appendix 28)
The predistortion data updater is configured to update the look-up table function and the filter function of the predistorter module online by the nonlinear system inverse characteristic identification method according to any one of appendices 1 to 9. 28. The power amplifying device according to appendix 27, wherein
(Appendix 29)
The predistortion data updater is configured to update the look-up table function and filter function of the predistorter module online in the following manner:
Compare the original input signal x (n) with the feedback output signal and

[Where F {•}, Q {•}, and P {•} represent a filter function, a lookup table function, and a power amplifier function, respectively.

And the lookup table function Q {•} is given by

And
The filter coefficient is

And
L represents the storage length of the filter, and [| · |] represents absolute value acquisition and quantization)
The error signal is calculated according to

〔here,

And both U and μ _q are steps to convergence]
The power amplifying apparatus according to appendix 27, wherein the lookup table function and the filter coefficient are updated online by
(Appendix 30)
A method of updating parameters of a predistorter module of a power amplifying device online,
Compare the original input signal x (n) of the power amplifying device with the feedback output signal.

[Where F {•}, Q {•}, and P {•} represent the filter function of the filter, the lookup table function of the lookup table, and the amplification function of the power amplifier, respectively.

And the lookup table function Q {•} is given by

And
The filter coefficient is

And
L represents the storage length of the filter, and [| · |] represents absolute value acquisition and quantization)
The error signal is calculated according to

〔here,

And both U and μ _q are steps to convergence]
A method for online updating parameters of a predistorter module of a power amplifying device, comprising: online updating a look-up table function and filter coefficients.

It is a schematic diagram which shows the nonlinear characteristic of the input-output signal of a power amplifier. It is a schematic diagram which shows the spectrum of the signal amplified by the nonlinear power amplifier. It is a schematic diagram which shows the characteristic of the input-output signal of the predistorter of the power amplifier which cancels the nonlinear characteristic shown in FIG. It is a schematic diagram which shows the characteristic of the input-output signal of a power amplifier provided with a predistorter. It is a schematic diagram which shows the characteristic of the input-output signal of the nonlinear power amplifier which has a memory effect. It is a schematic diagram which shows the spectrum of the signal amplified by the nonlinear power amplifier which has a memory effect. It is a schematic diagram which shows performing a predistortion process with respect to the input signal of the nonlinear power amplifier which has a memory effect using a memory polynomial model by a prior art. It is a schematic diagram which shows performing a predistortion process with respect to the input signal of the nonlinear power amplifier which has a memory effect using a two-dimensional lookup table by a prior art. It is a schematic diagram of the structure which produces the reverse characteristic of a nonlinear system using a Hammerstein model. It is a schematic diagram which shows an example of the indirect learning method for acquiring the reverse characteristic of the nonlinear system which a present Example employ | adopts. It is a schematic diagram which shows an example of the identification apparatus which acquires the reverse characteristic of a nonlinear system offline by the indirect learning method of a present Example. It is a figure which shows the flow of the optimization process process of the identification method which acquires the reverse characteristic of a nonlinear system offline by the indirect learning method of a present Example. It is a block diagram which shows the nonlinear system reverse characteristic identification apparatus which concerns on a present Example. It is a block diagram which shows the structure of the apparatus which collects the input / output data of a power amplifier. It is a block diagram which shows the structure of the nonlinear power amplifier which can implement | achieve the predistortion by a present Example. It is a block diagram which shows the structure of the power amplifier which can update and upgrade the parameter of the predistorter module by a present Example online. It is a block diagram which shows the example structure of a personal computer.

Claims (10)

A nonlinear system inverse characteristic identification method for identifying an inverse characteristic of a nonlinear system from an original input signal x (n) of the nonlinear system and a corresponding feedback output signal y _G (n),
A lookup table function Q {•} generating step for calculating a lookup table function Q {•} from the inverse filter signal x _F (n) and the feedback output signal y _G (n);
An intermediate predistortion added signal z (n) generating step for generating an intermediate predistortion added signal z (n) by the feedback output signal y _G (n) and the lookup table function Q {·};
A filter function F {•} creating step for creating a filter function F {•} from the intermediate predistortion added signal z (n) and the original input signal x (n);
Using the filter function F {

And performing an inverse filtering on the original input signal x (n) to generate the inverse filtered signal x _F (n), and
Including repeating each of the above steps until the set conditions are met,
The nonlinear system inverse characteristic identification method, wherein the lookup table function Q {·} and the filter function F {·} represent inverse characteristics of the nonlinear system. When generating the lookup table function Q {·} for the first time, together with the feedback output signal y _G (n), the original input signal x (n) as the inverse filter signal x _F (n), The nonlinear system inverse characteristic identification method according to claim 1, wherein the lookup table function Q {·} is calculated. The set condition is expressed by the following formula:

[Where

Is

The estimated value of the original input signal x (n) generated using the filter function F {·} and the intermediate predistorted signal z (n) by

Is)
The non-linear system inverse characteristic identification method according to claim 2, characterized in that the normalized mean square error NMSE shown in (2) converges or that the number of repetitions is larger than a prescribed threshold value. Next formula

[Where [| ・ |] represents absolute value acquisition and quantization]
The non-linear system inverse according to claim 3, further comprising: taking an absolute value and quantizing the feedback output signal y _G (n) into one address k in a lookup table according to Characteristic identification method. A nonlinear system inverse characteristic identification device that identifies the inverse characteristic of a nonlinear system from the original input signal x (n) of the nonlinear system and the corresponding feedback output signal y _G (n),
A parameter calculator that identifies inverse characteristics of the nonlinear system and generates a look-up table function Q {·} and a filter function F {·} representing the inverse characteristics of the nonlinear system;
It is determined whether the lookup table function Q {·} and the filter function F {·} generated by the parameter calculator satisfy a set condition, and the lookup table function Q {·} A convergence determination unit that controls the parameter calculator to repeatedly calculate the filter function F {·},
The parameter calculator is
A look-up table generator for calculating the look-up table function Q {•} from an inverse filter signal x _F (n) and the feedback output signal y _G (n);
An intermediate predistorter for generating an intermediate predistortion added signal z (n) from the feedback output signal y _G (n) and the lookup table function Q {·} calculated by the lookup table generator;
A filter parameter calculator that creates the filter function F {·} from the intermediate predistortion added signal z (n) and the original input signal x (n) generated by the intermediate predistorter;
Using the filter function F {·} created by the filter parameter calculator,

And a reverse filter that performs reverse filtering on the original input signal x (n) to generate the reverse filter signal x _F (n). A power amplifier predistorter,
An address generator that takes an absolute value for the original input signal, performs quantization, and then converts it to a lookup table address;
A one-dimensional lookup table arranged according to a lookup table function Q {·} and outputting a correction coefficient according to the lookup table address outputted from the address generator;
A multiplier for multiplying the original input signal by the correction coefficient output from the one-dimensional lookup table to obtain an intermediate predistorted signal;
A filter that is arranged according to a filter function F {·} and performs filtering on the intermediate predistorted signal output from the multiplier to generate a predistorted signal,
The lookup table function Q {·} and the filter function F {·} are generated by the nonlinear system inverse characteristic identification method according to any one of claims 1 to 4, or according to claim 5. A predistorter for a power amplifier, characterized in that the predistorter is generated by a non-linear system inverse characteristic identification device. Further comprising a data collection and storage device for collecting the original input signal and feedback output signal of the power amplifier;
The data collection storage device
A digital-to-analog converter for converting the original input signal into an analog signal;
An up-converter that up-converts the analog signal output from the digital-analog converter into an RF signal;
The power amplifier for power amplifying the RF signal output from the up-converter;
A directional coupler that branches the output of the power amplifier into two branched signals, outputs one of the signals as an output signal, and feeds back the other signal to the attenuator;
Receiving the signal fed back from the directional coupler and attenuating the feedback signal;
A downconverter that downconverts the attenuated signal output from the attenuator into a baseband signal;
An analog-to-digital converter that converts the baseband signal output from the down-converter into a digital signal and sets the feedback output signal;
The power amplifier predistorter according to claim 6, further comprising a data acquisition and storage module that collects and stores the original input signal and the feedback output signal. A power amplifying device,
A predistorter module that predistorts the original input signal and outputs a predistorted signal;
A digital-to-analog converter that converts the predistorted signal output from the predistorter module into an analog signal;
An up-converter that up-converts the analog signal output from the digital-analog converter into an RF signal;
A power amplifier that amplifies the RF signal output from the upconverter and outputs a power amplified signal;
The configuration of the predistorter module is the same as the configuration of the predistorter of the power amplifier according to claim 6. A power amplifying device,
A predistorter module having the same configuration as the predistorter of the power amplifier according to claim 6 and used for predistorting an original input signal and outputting a predistorted signal;
A digital-to-analog converter that converts the predistorted signal output from the predistorter module into an analog signal;
An up-converter that up-converts the analog signal output from the digital-analog converter into an RF signal;
A power amplifier that amplifies the RF signal output from the up-converter and outputs a power-amplified signal;
A directional coupler for branching the output of the power amplifier into two signals, outputting the one signal as an output signal, and feeding back the other signal to the attenuator;
Receiving the signal fed back from the directional coupler and attenuating the feedback signal;
A downconverter that downconverts the attenuated signal output from the attenuator into a baseband signal;
An analog-to-digital converter that converts the baseband signal output from the down-converter into a digital signal and sets the feedback output signal;
A delay unit that delays the original input signal to generate a delayed input signal;
The feedback output signal output from the analog-to-digital converter and the delayed input signal output from the delay unit are received, and a lookup table function is obtained according to the received feedback output signal and the delayed input signal. And a predistortion data updater that updates the filter coefficient online. A method of updating parameters of a predistorter module of a power amplifying device online,
Comparing the original input signal x (n) of the power amplifier and the feedback output signal,

[Where F {•}, Q {•}, and P {•} represent the filter function of the filter, the lookup table function of the lookup table, and the amplification function of the power amplifier, respectively.

And the lookup table function Q {•} is given by

And
The filter coefficient is

And
L represents the storage length of the filter, and [| · |] represents absolute value acquisition and quantization)
The error signal is calculated according to

〔here,

And both U and μ _q are steps to convergence]
The method of updating the parameters of the predistorter module of the power amplifying device online, comprising: updating the lookup table function and the filter coefficient online.

Images