JP2010011348A - Distortion compensation circuit and distortion compensation method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a distortion compensation circuit and distortion compensation method for preventing excessive compensation. <P>SOLUTION: A distortion compensation circuit 1 comprises a model estimation unit 2 for estimating a model that represents an output signal of an amplifier 5 in the form of a polynomial having n-th order power series of an input signal X, and a distortion compensation unit 4 that performs distortion compensation for canceling distortion in input/output characteristics of the amplifier 5 by adding an inverse model, with respect to the model, to the input signal. In the distortion compensation circuit 1, the model estimation unit 2 increases/decreases a value of (n) on the basis of whether or not regularity is maintained that a coefficient of each order in the polynomial becomes smaller as a degree becomes larger. In particular, if compensation may become excessive, a maximum degree can be decreased, thereby preventing excessive compensation. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a distortion compensation circuit having a function of compensating for nonlinear characteristics of a high-power amplifier used in, for example, a radio transceiver, and a distortion compensation method thereof.

  In general, a high power amplifier (HPA) has low linearity of input / output characteristics because priority is given to efficiency characteristics. Therefore, when power is amplified using such an amplifier, a desired output may not be obtained due to distortion of input / output characteristics. Therefore, as a distortion compensation method for compensating such distortion, a reverse distortion characteristic opposite to the distortion characteristic of the amplifier is generated by digital signal processing for the input signal of the amplifier and added to the input of the amplifier. A technique for obtaining a desired amplifier output by performing DPD (Digital Pre-Distortion) processing has been proposed (for example, see Non-Patent Document 1).

  In addition, it has been proposed to perform distortion compensation represented by a polynomial in order to perform highly accurate DPD processing on a high-power amplifier that amplifies a broadband signal (see, for example, Patent Document 1).

Thesis by Lei Ding, "Digital predistortion of power amplifiers for wireless application", Georgia institute of Technology, March 2004 JP 2007-282066 A

  However, the reverse distortion characteristics may not be estimated correctly, which results in promoting the nonlinear characteristics of the amplifier. FIG. 5 is a graph showing two examples of input / output characteristics of the amplifier, in which the horizontal axis indicates the input signal amplitude and the vertical axis indicates the output signal amplitude. The characteristic shown in (a) has a complicated shape and a large nonlinearity, while the characteristic shown in (b) has a simple shape and a small nonlinearity. The dotted line shows ideal linear characteristics. With respect to the characteristics of (a) and (b), when a model is estimated with a power series polynomial of a predetermined order and the inverse model is obtained, the result is as shown by broken lines in FIGS. 6 (a) and 6 (b). The solid line shows the same input / output characteristics as those shown in FIGS.

  In the case of FIG. 6A, an inverse model suitable for compensating the distortion of the input / output characteristics is given, and as a result, the input / output characteristics with high linearity from which the distortion is removed as shown in FIG. Is obtained. However, in the case of FIG. 6B, the inverse model becomes inappropriate and excessive compensation is performed. As a result, as shown in (d), distortion appears on the opposite side of the dotted line. This is considered to be due to overcompensation of distortion of a simple shape by a high-order polynomial. A focus on such overcompensation and solutions have not yet been proposed.

  In view of such a conventional problem, an object of the present invention is to provide a distortion compensation circuit and a distortion compensation method that prevent overcompensation.

  The present invention provides a model estimation unit that estimates a model in which an output signal of an amplifier is expressed in the form of a polynomial of an n-th power series of the input signal, and an inverse model for the model is added to the input signal to add the input of the amplifier. A distortion compensation circuit that performs distortion compensation that cancels distortion of output characteristics, wherein the model estimation unit maintains a regularity that each order coefficient in the polynomial is smaller as the degree is larger. It has a function to increase or decrease the value of n based on whether or not it is present.

  In the distortion compensation circuit configured as described above, n that is the maximum degree of the polynomial is not set to a fixed value, and the suitability of the model is determined by paying attention to the regularity of the coefficient, and is increased or decreased so as to become an appropriate model. Can do. In particular, the maximum order can be lowered in the case of overcompensation.

In the distortion compensation circuit, when a coefficient contrary to regularity appears, the model estimation unit may reduce the order by 1 and estimate the model.
In this case, the model can be estimated with the order immediately before overcompensation.

Further, in the distortion compensation circuit, each order coefficient in order from 1 maintains the regularity, and the order when the error between the model and the actual output signal is within a predetermined value is calculated as a model estimation unit. Can be the value of n.
In this case, an accurate model can be estimated within a range in which overcompensation can be eliminated by maintaining regularity.

  On the other hand, the present invention estimates a model representing the output signal of the amplifier in the form of a polynomial of the nth power series of the input signal, and adds an inverse model to the input signal to the input / output characteristics of the amplifier. A distortion compensation method for performing distortion compensation that cancels out the distortion of the equation, wherein the value of n is increased or decreased based on whether or not each order coefficient in the polynomial maintains the regularity that the degree is smaller as the degree is larger. In this order, each order coefficient maintains the regularity, and the order when the error between the model and the actual output signal is within a predetermined value is the value of n.

  According to the above distortion compensation method, it is possible to determine whether or not a model is appropriate by paying attention to the regularity of a coefficient without increasing n, which is the maximum degree of a polynomial, so as to become an appropriate model. it can. In particular, the maximum order can be lowered in the case of overcompensation. In addition, an accurate model can be estimated within a range in which overcompensation can be eliminated by maintaining regularity.

  According to the distortion compensation circuit of the present invention, it is possible to reduce the maximum order in the case of overcompensation, thereby preventing overcompensation. By not increasing the order more than necessary in this way, relatively low-order operations increase, which contributes to high-speed distortion compensation.

  FIG. 1 is a block diagram showing a main part of a distortion compensation circuit 1 according to an embodiment of the present invention. The distortion compensation circuit 1 includes a model estimation unit 2 that estimates a model in which an output signal Y of a high output amplifier (HPA, hereinafter simply referred to as an amplifier) 5 is expressed in the form of a polynomial of an nth power series of the input signal X; A coefficient table 3 that temporarily stores the coefficients of the model, and a distortion compensation unit that performs distortion compensation that cancels the distortion of the input / output characteristics of the amplifier 5 by adding an inverse model (inverse function of the model) to the input signal X. 4 is provided. The model estimation unit 2, the coefficient table 3, and the distortion compensation unit 4 are configured by, for example, a DSP (Digital Signal Processor). Note that both the input signal X and the output signal Y are input to the model estimation unit 2.

The nonlinear characteristic of the amplifier 5 is expressed by a power series polynomial of the input signal X. That is, when the output signal is Y, it can be expressed as follows. Note that a _i is a coefficient of each order.

Here, it has been experimentally known that the value (absolute value) of the coefficient a _i of X becomes smaller as the order increases. That is, │a ₁ │> │a ₂ │>. . . There is a regularity of | │a _n-1 │> │a _n │.
FIG. 2A is a graph showing a harmonic spectrum of an amplifier having strong nonlinearity. The harmonic spectrum is closely related to the nonlinear characteristics of the amplifier, and an amplifier having a large harmonic spectrum intensity has a strong nonlinearity. In this case, the polynomial of the model is required up to higher order (for example, n = 10).

On the other hand, FIG. 2B is a graph showing a harmonic spectrum of an amplifier having weak nonlinearity. In the case of such a weak non-linearity, for example, the spectral intensity of the fourth order or higher is small, and a very high order component is unnecessary. If you try to force a coefficient at an order higher than necessary, the coefficient will diverge. That is, a _i whose order is larger than the coefficient under the first order appears, and the regularity cannot be maintained. If distortion compensation is performed using an inverse model for such a model, overcompensation occurs. That is, when the coefficient diverges as described above, it is an overcompensation state.

Therefore, the model estimation unit 2 is provided with a function of increasing / decreasing the value of n based on whether or not the regularity of each degree in the polynomial is smaller as the degree is larger.
FIG. 3 is a flowchart showing processing in the model estimation unit 2. After the start of the process, the model estimation unit 2 estimates the model by setting the initial value of n to 1 (step S1) (step S2). Thereafter, the model estimation unit 2 determines whether or not there is excessive compensation (step S3). Here, since the coefficient is only a ₁ , an event that a ₂ larger than a ₁ exists cannot occur, and the determination result in step S3 is “No”.

  Next, the model estimation unit 2 determines whether or not the error between the output represented by the model and the actual output is within a predetermined value (slight difference), that is, whether the model is appropriate (step S4). ). Since the model has not yet reached an appropriate level at this point, the determination result is “No”. Accordingly, the model estimation unit 2 adds 1 to n (step S6) and performs model estimation again (step S2). Such processing is repeated, and when the error is initially within a predetermined value (Yes in step S4), the order is once determined, and model estimation (step S2) at that order is repeatedly performed.

On the other hand, when the error is overcompensated before it falls within the predetermined value, or after the order is determined once as described above, the characteristic of the amplifier 5 is changed due to a temperature change or the like. In other words, when a coefficient whose order is larger than the coefficient of the first order appears, the model estimation unit 2 performs a process of subtracting n from 1 (step S5). For example, when the fifth-order model estimation results in that the fifth-order coefficient a ₅ is equal to or greater than the fourth-order coefficient a ₄ , the process returns to the fourth-order model estimation. Therefore, when overcompensation occurs, the order is reduced by 1, and as a result, model estimation with an order greater than necessary is not performed.

  In addition, after the order is once determined as described above, the model becomes inappropriate due to a change in the characteristics of the amplifier 5 due to a temperature change or the like, and when the error is not within a predetermined value, the model estimation unit 2 , N is incremented by 1 (step S6). Thereby, an appropriate model is reconstructed.

As a result of the above processing, the model is estimated so that the higher order is required, and the higher order is not required. The estimated model coefficients are temporarily stored in the coefficient table 3. The distortion compensator 4 obtains an inverse model while referring to the coefficient table 3, and adds this to the input signal to perform distortion compensation.
In addition to the above, the model estimation unit 2 may obtain the inverse model and store the coefficients in the coefficient table 3, and the distortion compensation unit 4 may perform distortion compensation while referring to the coefficient table 3. Good.

  4A and 4B are graphs showing two examples of input / output characteristics (solid line) and inverse model (broken line) of the amplifier, where the horizontal axis represents the input signal amplitude and the vertical axis represents the output signal amplitude. , Respectively. The characteristic shown in (a) has a complicated shape and large nonlinearity. The order of the inverse model reaches the 10th order, and an appropriate inverse model is given. On the other hand, the characteristic shown in (b) has a simple shape and small nonlinearity, and the order of the inverse model is limited to the third order, and an appropriate model is given. The dotted line shows ideal linear characteristics. As a result of such compensation, in both cases (a) and (b), as shown in (c) and (d), input / output characteristics from which distortion has been removed are obtained.

  In the case of (b), since the order is low, the calculation of model estimation is simplified, and the calculation speed is increased accordingly. That is, if the order is not increased more than necessary, relatively low-order operations increase, which contributes to high-speed distortion compensation.

  As described above, in the distortion compensation circuit 1 according to the present embodiment or the distortion compensation method executed by the circuit 1, the maximum degree n of the polynomial is not set to a fixed value, and attention is paid to the regularity of the coefficient. Appropriateness can be determined and increased or decreased to become an appropriate model. In particular, since the maximum order can be lowered in the case of overcompensation, the distortion compensation circuit 1 or the distortion compensation method that prevents overcompensation can be provided.

In addition, when a coefficient contrary to regularity appears, the model estimation unit 2 can estimate the model with the order immediately before overcompensation by reducing the order by 1 and estimating the model.
Further, each order coefficient maintains regularity in order from 1, and when the error between the model and the actual output signal is within a predetermined value, the model estimation unit 2 determines that the value of n By doing so, an accurate model can be estimated within a range in which overcompensation can be eliminated by maintaining regularity.

  This logic is similarly applied to distortion compensation for amplifiers that require memory effect compensation. In other words, even in an amplifier whose past state affects the current nonlinearity, the coefficient of each order of the power series of the inverse model characteristic at any past time that affects the current nonlinearity is If the value of the coefficient does not decrease as it decreases, overcompensation can be prevented by reducing the order.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined not by the above-mentioned meaning but by the scope of claims for patent, and is intended to include all modifications within the scope and meaning equivalent to the scope of claims for patent.

It is a block diagram which shows the principal part of the distortion compensation circuit which concerns on one Embodiment of this invention. (A) is a graph which shows the harmonic spectrum of the amplifier which has strong nonlinearity, (b) is a graph which shows the harmonic spectrum of the amplifier which has weak nonlinearity. It is a flowchart which shows the process in a model estimation part. It is a graph which shows two examples of the input-output characteristic (solid line) and inverse model (dashed line) of an amplifier. It is a graph which shows two examples of the input-output characteristic of an amplifier. It is a graph (conventional example) which shows the characteristic after an inverse model and distortion compensation about two examples of the input-output characteristic of an amplifier.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Distortion compensation circuit 2 Model estimation part 3 Coefficient table 4 Distortion compensation part 5 Amplifier

Claims (4)

A model estimator for estimating a model representing the output signal of the amplifier in the form of a polynomial of the nth power series of the input signal;
A distortion compensation circuit that performs distortion compensation to cancel distortion of input and output characteristics of the amplifier by adding an inverse model to the input signal to the input signal,
The model estimation unit has a function of increasing / decreasing the value of n based on whether or not each order coefficient in the polynomial maintains the regularity that the degree is smaller as the degree is larger. circuit.   The distortion compensation circuit according to claim 1, wherein, when a coefficient contrary to the regularity appears, the model estimation unit estimates the model by reducing the order by one.   The model estimation unit calculates the order when the coefficient of each order in order from 1 maintains the regularity and the error between the model and the actual output signal is within a predetermined value. The distortion compensation circuit according to claim 1 or 2, wherein the value is a value. Distortion compensation that estimates the model of the output signal of the amplifier in the form of a polynomial of the nth power series of the input signal, and adds the inverse model to the model to the input signal to cancel the distortion of the input / output characteristics of the amplifier A distortion compensation method for performing
The value of n is increased or decreased based on whether or not each degree of coefficient in the polynomial maintains the regularity that the larger the degree is, the smaller the degree is.
Each order coefficient in order from 1 maintains the regularity, and the order when the error between the model and the actual output signal is within a predetermined value is the value of n. Distortion compensation method.

Images