KR100324123B1 - Method and apparatus for amplifier linearization using a predistorter - Google Patents

Abstract

The present invention relates to a technique for linearizing the nonlinear characteristics of a power amplifier, and more particularly, to an apparatus and method for linearizing an output by pre-distorting a baseband input signal.

The present invention estimates a baseband equivalent representation using an adaptive algorithm for an arbitrary baseband input signal of an RF power amplifier and a frequency conversion of the baseband input signal and the baseband feedback signal demodulated by passing the power amplifier. Further, the present invention is characterized by calculating the back transfer characteristic of the power amplifier required for the predistorter from the baseband equivalent representation.

As a result, the predistortion linearization apparatus and method according to the present invention can estimate the change in the characteristics of the power amplifier without affecting the transmission operation of the power amplifier at all. In addition, since the predistortion scheme adapted to the change of the characteristics of the power amplifier in real time can be expected to improve the linearization characteristics compared to the prior art.

Description

METHOD AND APPARATUS FOR AMPLIFIER LINEARIZATION USING A PREDISTORTER

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for solving distortion caused by nonlinearity of RF power amplifiers. In particular, the present invention relates to predistortion of baseband input signals of power amplifiers. The present invention relates to a method and an apparatus for linearizing output.

As digital communication technology is applied to a mobile communication system, linearity of individual components constituting the communication system is emerging as an essential element for providing high quality communication service. In the case of RF power amplifiers, which are the main components of the communication system, the third or fifth order nonlinear characteristics of the RF power amplifier cause interference in adjacent communication channels, which is strictly regulated as a standard specification of the communication system.

There is a back-off method as a prior art for solving the nonlinear characteristic problem of the RF power amplifier, the technique is described in US Patent No. 4,637, 017. By the way, the back-off method according to the prior art can improve the linear characteristics of the power amplifier, but has a problem of low efficiency of the amplifier. Accordingly, a linearization technique using a baseband predistortion scheme has been proposed as an appropriate linearization technique for simultaneously satisfying the high efficiency and linearity of the RF amplifier.

FIG. 1A is a diagram illustrating a configuration of a linearization apparatus using a baseband predistortion method according to the prior art. Referring to FIG. 1A, the transfer characteristic of the predistorter 101 may be expressed as an inverse function of the transfer characteristic of the power amplifier 102, and an linearly amplified input signal may be obtained as an output of the power amplifier.

That is, according to the prior art pre-distorter linearization method, a linearly amplified power amplifier characteristic is obtained by applying the output of the pre-distorter 101 having the inverse function characteristic G ⁻¹ of the power amplifier transfer characteristic to the power amplifier 102. Can be obtained.

However, since the nonlinear characteristics of the power amplifier can be represented by the distortion of the magnitude and phase of the output signal according to the magnitude of the input signal, the reverse propagation characteristics required by the predistorter are as shown in FIG. 1b. ) Can be implemented by mapping the magnitude of the input signal.

Referring to FIG. 1B, the complex input signals I _i and Q _i 103 obtain a linearized output signal 105 by applying a pre-distorted signal 104 to the power amplifier using a look-up table.

The predistorter technique using this one-dimensional reference table is published by J. K Cavers on pages 374 to 382 of IEEE Transaction Vehecular Technology Proceedings Vol.30 No.4, published in 1990. Amplifier linearization using a digital predistorter with fast adaptation and low memory requirements.

Figure 1c is a view showing a pre-distortion linearization device according to the prior art. In the prior art linear distortion apparatus, since the distortion of the power amplifier occurs in proportion to the magnitude of the baseband complex signal, this value is stored in the memory in the form of a reference table according to the input size.

The polar coordinate converter 106 of FIG. 1C is a step for obtaining an input size and calculating the address of the lookup table according to this value. This process is performed in the lookup table address calculation indicated by reference numeral 107. Also, the lookup table address value As a result, the signal compensated through the rotation operation 110 or the like passes through the rectangular coordinate converter 111.

However, the predistortion linearizer according to the prior art shown in FIG. 1c applies an input signal corresponding to one element of the lookup table to an artificially known value in order to fill the amplitude lookup table and the phase lookup table. A method of updating the reference table by finding the point where the amplitude error 112 and the phase error 114 are minimized while changing the value by using an adaptive algorithm is employed.

This process is often defined as the training mode, which has the problem of stopping the signal that the power amplifier must transmit until the update of the lookup table is complete.

In addition, the above-described prior art linearization apparatus of the distortion method should correct the transfer characteristics of the predistorter with respect to the characteristic change of the power amplifier due to heat or aging.

To this end, the above-described prior art linearization apparatus of the distortion method is obtained by converging the values of the reference table using an adaptive algorithm so that the error of the input / output signal of the power amplifier is minimized. Since a specific training signal needs to be applied to the power amplifier until it converges, there is a problem in that the transmission interrupt of the power amplifier occurs.

Moreover, the prior-distortion linearization method according to the prior art has a possibility that the performance of the linearization device may suddenly degrade or become unstable when diverging any of the values in the reference table.

Accordingly, it is a first object of the present invention to provide a distortion linearization method and apparatus capable of maintaining a good degree of improvement with respect to a characteristic change of a power amplifier.

A second object of the present invention is to provide a predistortion linearization method and apparatus capable of maintaining a good degree of improvement by estimating a characteristic change of a power amplifier over time, irrespective of the output operation of the power amplifier, in addition to the first object. To provide.

A third object of the present invention is to provide a pre-distortion linearization method and apparatus capable of estimating a distortion component required for linearization by estimating a characteristic change of a power amplifier over time, in addition to the first object.

In addition to the first object, the fourth object of the present invention eliminates the amplitude error 112 and phase error 114 portions of the prior art, and inserts an equivalent model estimator into the power amplifier input / output stage for any input signal. To provide a method and apparatus for extracting the equivalent model of the power amplifier.

1a is a diagram showing the principle of a linearization device by a baseband predistortion method according to the prior art.

FIG. 1B is a diagram illustrating a trajectory on a complex plane for explaining a correlation between signals of stages in a linearization apparatus using a baseband predistortion method according to the related art.

Figure 1c is a diagram showing a linearization device by a baseband pre-distortion scheme according to the prior art.

2 is a diagram showing a configuration of a power amplifier predistortion linearization device according to the present invention. FIG. 3 is a diagram showing an embodiment of a digital predistorter configuration according to the present invention.

4 is a view showing the operation of the power amplifier pre-distortion linearization apparatus according to the present invention. FIGS. 5a and 5b are views showing the configuration of a digital predistorter using real-time equivalent model estimation according to the present invention. Fig. 6 is a diagram illustrating the measurement of the baseband complex transfer characteristics of a power amplifier using an IQ demodulator when the output of the power amplifier is varied from -3 dBm to 17 dBm according to the present invention. Fig. 7a shows a spectrum of an input signal used in an embodiment according to the present invention. Fig. 7b shows the spectrum of an input signal used in the embodiment according to the present invention. Nonlinear output spectrum of an embodiment according to FIG.

<Reference Description of Symbols for Main Parts> 10: Reference Table 30: Baseband Input Signal 40: Digital Signal 41: Power Amplifier Output 60: Digital Unit 61: DSP (Digital Signal Processor) 62: Signal Converter 70: Analog Apparatus 71, 72: IQ modulator 73: Power amplifier used for measurement 74: Directional coupler 75: Attenuator

103: Trajectory of the power amplifier input signal required for linear characteristics

104: Trajectory of the power amplifier input signal through the predistorter

105: Trajectory of the linearized power amplifier output signal

201: pre-distorter

202: power amplifier baseband equivalent representation estimator

203: right angle modulator

204: right angle demodulator

205: RF Nonlinear Power Amplifier

In order to achieve the above object, the present invention uses a baseband of a power amplifier using an adaptive algorithm to frequency-modulate any baseband input signal of the RF power amplifier and the baseband feedback signal demodulated by passing the power amplifier and demodulated through the power amplifier. A predistortion linearization technique is provided that estimates a band equivalent representation and calculates the reverse propagation characteristics of a power amplifier required for a predistorter. The complex transfer function of the power amplifier is estimated, and the look-up table (LUT) is updated using the equivalent model so that all items of the reference table can be modified at one time. In addition, since the present invention is not limited to the input signal, there is an advantage that the input of a specific signal is unnecessary and can be used in any communication system.

In addition, in order to compensate for the characteristic change of the power amplifier due to heat or aging, it is desirable to correct the transfer characteristic of the predistorter by re-estimating the baseband equivalent representation of the power amplifier periodically or according to a predetermined criterion.

Hereinafter, a linearization technique of an RF power amplifier according to the present invention and a preferred embodiment of the method will be described in detail with reference to FIGS. 2 to 7.

2 is a block diagram showing a predistortion linearization device according to the present invention. Referring to FIG. 2, the baseband equivalent representation adaptive estimation apparatus 202 for receiving the baseband input signal V _d and the output feedback signal V ₀ of the power amplifier and estimating the baseband equivalent representation of the power amplifier is input according to the reference table. The predistorter 201 distorts the signal. Additionally, a quadrature modulator 203 and a quadrature demodulator 204 for frequency band conversion of the signal are provided, and a nonlinear power amplifier 205 to be linearized.

The predistortion linearizer according to the present invention inputs a signal obtained by frequency converting an arbitrary baseband input signal into an RF band to the power amplifier 205, and converts the output of the power amplifier into a baseband through a quadrature demodulator 204. Input to the baseband equivalent representation adaptive estimation apparatus 202. Subsequently, the predistorter 201 generates a pre-distortion signal capable of compensating for the current power amplifier distortion transfer characteristic by using a look-up table generated by the power amplifier baseband equivalent representation adaptive estimation apparatus 202 to generate a quadrature modulator ( 3 is a diagram showing an embodiment of a digital predistorter configuration according to the present invention. Article Referring to Figure 3, IQ form of ideal base band input signal V _i (30) is first converted to polar coordinate form. This is based on the assumption that the nonlinearity of the power amplifier is a function of only the magnitude of the input signal, and to refer to the reference table 10 using the magnitude of the input signal. As a preferred embodiment of the method of addressing a lookup table, the voltage of the input signal can be divided at equal intervals. [0046] In the lookup table (LUT) 10 according to the present invention, the AM-AM conversion and the AM of the power amplifier 205 are used. Information for compensating the -PM conversion is stored, and a baseband input signal V _d 40 is generated to be actually applied to the power amplifier 205 according to this reference table 10. As such, when the distorted V _d 40 is divided into magnitude and phase components, the distortion may be expressed as follows. here, Is the magnitude of the input signal corresponding to the i th LUT (10) item, Indicates the phase of the input signal at this time. Also, Is a value stored in the i th size LUT 10 to compensate for AM-AM conversion, Is the content of the i-th phase LUT 10 to compensate for AM-PM conversion. For reference, a capitalized signal means a complex envelope signal or value, and a lowercase letter indicates a real envelope signal or value. As such, a digital signal V _d (distorted according to the content of the LUT 10) is used. 40) is converted into an IQ form and then finally applied to the IQ modulator via a digital-to-analog converter. Therefore, the output V ₀ 41 of the final power amplifier can obtain the same signal as the input signal V _i 30. The non-linear characteristics of the power amplifier change with temperature or time of use, and to compensate for this, The output signal V ₀ of the amplifier should be measured from time to time and the contents of the LUT 10 should be updated if necessary using an adaptive algorithm. In general, the adaptive algorithm compares the output signal V ₀ with the input signal V _i and modifies the contents of the LUT 10 through several iterations to minimize the error signal V _e . When applying the least-mean-square (LMS) algorithm, which is generally used among many adaptation algorithms, to LUT update, the following equation can be obtained. here ego, to be. In addition, the subscript n + 1 in the equations (3) and (4) represents the number of iterations of the adaptive algorithm, Is a coefficient for adjusting the convergence speed of the adaptive algorithm, and can be determined in the range as shown in Equation 5 in consideration of stability. However, when the output V ₀ of the actual power amplifier is used, Equation 3 and Equation 4 must be repeated until V _e converges to the maximum allowable error range for each item of the LUT. Therefore, all items cannot be updated at the same time. Do. In addition, if the input signal of a certain size is to be maintained until the update of one item of the LUT is completed, it also affects the output operation of the power amplifier.

4 is a flowchart showing the operation procedure of the power amplifier baseband equivalent representation adaptive estimation apparatus. Referring to FIG. 4, the complex output V ₀ of the power amplifier with respect to the pulse shaping input signal V _i is sampled at time t 302 so that the adjacent channel power ratio (ACPR) of the output spectrum is adjacent. channel power ratio) is estimated (303).

A step 304 is then performed to compare the calculated adjacent channel power ACPR with a target threshold power P _th , and if the calculated adjacent channel power is greater than the threshold (ie, power amplifier's power amplifier). If the distortion nonlinearity is prominent, decide to update the lookup table. If the need for a lookup table update is determined (304), then the system delay time is estimated (305).

On the other hand, the method for estimating the system delay calculates an approximation of the correlation function of the input and output signals according to the system delay time d expressed in Equation 6, and estimates the time d indicated by the maximum value as the delay time of the entire system.

In Equation 6, t _d is an expected value of the actual delay time, and other variables may be determined by an experimental method.

In Equation 6, D is greater than t _d , and T is greater than or equal to N + D. In the equivalent model estimation step, the baseband equivalent representation of the power amplifier is assumed to be a complex polynomial of the order (2M + 1), as shown in Equation 7, and the coefficients of (M + 1) complex polynomials are represented by the recursive least squares (RLS). Estimate using adaptive algorithm.

Here, there is a relationship where V _d = I _d + jQ _d , V ₀ = I ₀ + jQ ₀ , and C _k = a _k + jb _k .

An iteration of the RLS algorithm using Equation 7 is as follows.

here,

ego,

d is the delay time of the feedback loop estimated in the previous step. According to Equation 8, the adaptive estimating apparatus proposed in the present invention uses only the real data of the feedback signal as an input, so that the feedback path can be reduced to a single channel in two channels of the real part (I) and the imaginary part (Q). .

Finally, the baseband equivalent representation of the estimated power amplifier is used to update the lookup table required for the predistorter in the digital domain. The method may use the LMS (Least Mean Squares) algorithm so that the error of input and output is minimized with respect to the magnitude and phase as in the digital predistorter using a general one-dimensional reference table. It includes both AM-AM conversion characteristics and AM-PM conversion characteristics. By knowing the complex coefficient of the transfer function in the above-described manner, it is possible to estimate the AM-AM conversion characteristics and the AM-PM conversion characteristics of the power amplifier. In addition, using the complex coefficient vector θ and the input signal, the output signal V ₀ of the power amplifier can be known, and based on this, the LUT of the digital predistorter can be updated. Referring to FIG. Sample T consecutively at the real part of the output signal and use it to analyze the spectrum of the output of the power amplifier. The output continues to be monitored unless the ACPR (adjacent channel power ratio) in the output spectrum exceeds the internally set threshold P _th in the adaptive predistorter. The reference value P _th used here should be set more marginally than required by the power amplifier specification, taking into account the time it takes to update the LUT. If this threshold is exceeded, the characteristics of the power amplifier are regarded as changed and the new virtual model is estimated using the estimation method of the present invention, and the equations 3 and 4 are performed by performing the equations (3) and (4). Finish by creating a LUT and replacing it with the LUT used for the actual predistortion. By repeating this process continuously, it is possible to compensate for the change over time of the power amplifier and always satisfy the set ACPR specification. FIGS. 5A and 5B are digital using real-time equivalent model estimation according to the present invention. It is a figure which shows the structure of a predistorter. Referring to FIG. 5A, the digital predistorter according to the present invention is composed of a digital unit 60 and an analog unit 70. The digital unit 60 is used to implement a real-time complex transfer function estimator and a digital predistorter. It is composed of a digital signal processor (DSP) 61, a signal converter 62, and the like. As a preferred embodiment of the DSP 61 according to the present invention, a pair of TMS320C44 for real-time operation of 20 mips processing speed can be used, and each pair of DSP 61 has a three-channel digital-to-analog converter and an analog-to-digital converter, respectively. The analog unit 70 is composed of IQ modulators 71 and 72 for frequency conversion and a directional coupler 74 for feeding back an output, and an attenuator for adjusting the size of the feedback signal. 75), the IQ modulators 71 and 72 can use a product that can directly modulate and demodulate into the 900 MHz band in order to minimize the effects of nonlinearity. The power amplifier 73 used for the measurement is the operation of the linearizer. In order to verify the performance, a relatively small output power class 2 amplifier with a 1 dB compression point of 17 dBm in the cellular band can be used. The amplifier stage used consists of three stages, but the effects of the amplifiers at the front end except the amplifier at the last stage operate at a sufficiently lower point than the 1 dB compression point so that their effects can be neglected. Can be divided into 100 equal intervals, and the input signal required for estimating the equivalent model of the power amplifier is a square-root RRC (square-root) symbol with a random bandwidth of 416.6 KHz in consideration of the DSP processing power and the bandwidth of the system. raised cosine), which can be pulse shaped and stored in the DSP storage area for use. At this time, the roll-off coefficient of the RRC filter may be determined to be 0.2. FIG. 6A shows the baseband of the power amplifier using an IQ demodulator when the output of the power amplifier is changed from -3 dBm to 17 dBm according to the present invention. The complex transmission characteristic is measured and shown in the IQ coordinate system. FIG. 6b is a diagram illustrating the transfer function of the power amplifier estimated according to the present invention on the IQ plane by normalizing the output function based on the maximum value of the output. Comparing Figures 6a and 6b shows that the estimated model is consistent with the measurements. Thus, the frequency output when the LUT is generated using the estimated model and the digital predistorter is operated is shown in FIG. 7. FIG. 7a is a spectrum of an input signal used, and FIG. 7b is a diagram showing a nonlinear output spectrum. . 7c is also an output spectrum of the power amplifier linearized using the predistorter according to the present invention. At this time, the average output of the power amplifier is 9.6 dBm, which is about 7.4 dB back off at the output 1 dB compression point. Comparing FIG. 7B and FIG. 7C, it can be seen that the ACPR is improved by about 8.4 dB at the same output.

The foregoing has outlined rather broadly the features and technical advantages of the present invention to better understand the claims of the invention which will be described later. Additional features and advantages that make up the claims of the present invention will be described below. It should be appreciated by those skilled in the art that the conception and specific embodiments of the disclosed subject matter can be used immediately as a basis for designing or modifying other structures for carrying out similar purposes to the present invention.

In addition, the inventive concepts and embodiments disclosed herein may be used by those skilled in the art as a basis for modifying or designing other structures for carrying out the same purposes of the present invention. In addition, such modifications or altered equivalent structures by those skilled in the art may be variously changed, substituted, and changed without departing from the spirit or scope of the invention described in the claims.

As described above, the predistortion linearizer proposed in the present invention can estimate the nonlinear characteristics of the power amplifier using an arbitrary input signal using a real-time baseband equivalent representation estimator of the RF power amplifier.

Therefore, since the input of a specific training signal is unnecessary as in the conventional predistortion linearizer, it is possible to estimate the characteristic change of the power amplifier over time without affecting the output operation of the power amplifier. In addition, since the back transfer characteristics of the power amplifier required for the predistorter can be accurately calculated using the equivalent representation of the power amplifier, the performance of the improved linearizer can be expected.

In addition, the equivalent representation estimating apparatus of the power amplifier of the present invention can choose between the real part and the imaginary part of the feedback complex signal as input data, thereby simplifying the feedback circuit from two channels to a single channel.

Claims (9)

An apparatus for linearizing a power amplifier having nonlinear characteristics, Adaptive estimation means for receiving a baseband input signal and an output feedback signal of the power amplifier and estimating a baseband equivalent representation of the power amplifier; Pre-distortion means for distorting the input signal according to a look-up table updated using the baseband equivalent representation generated by the adaptive estimation means Pre-distortion linearization device comprising a. The predistortion linearization device of claim 1, wherein the predistortion linearization device further comprises a quadrature modulator and a quadrature demodulator for frequency band conversion of a signal. A method for linearizing a power amplifier having nonlinear characteristics, (a) sampling a predetermined number of outputs of the power amplifier for the pulsed input signal; (b) estimating the adjacent channel power ratio of the output spectrum of the power amplifier; (c) comparing the magnitude of the adjacent channel power ratio with a predetermined threshold; (d) calculating a delay time of the entire system; (e) estimating a baseband equivalent model of the power amplifier; (f) generating a look-up table from the baseband equivalent model; (g) generating an input distortion signal using the reference table Pre-distortion linearization method comprising a. The method of claim 3, wherein comparing the magnitude of the adjacent channel power ratio with a predetermined threshold value comprises: (d), (e), when the adjacent channel power ratio is larger than a predetermined threshold value as a result of the magnitude comparison. If steps (f) and (g) are performed and the adjacent channel power ratio is smaller than a predetermined threshold, steps (d), (e), (f) and (g) are omitted and step (a) is performed. A predistortion linearization method characterized in that it performs. 4. The method of claim 3, wherein calculating a delay time of the entire system comprises calculating a correlation function between an input signal and a feedback signal. 4. The method of claim 3, wherein estimating the baseband model of the power amplifier sets the baseband equivalent representation of the power amplifier to a complex polynomial of (2M + 1) differences and coefficients of (M + 1) complex polynomials. Pre-distortion linearization method characterized in that for calculating the RLS (Recursive Least Square) method. 4. The method of claim 3, wherein estimating the baseband model of the power amplifier Frequency shifting any baseband signal to the RF band; Frequency shifting the output signal of the power amplifier to baseband Pre-distortion linearization method comprising a. 8. The method of claim 7, wherein the frequency shifting of the output signal of the power amplifier to the baseband uses one of a real part and an imaginary part as an output of the power amplifier. 4. The method of claim 3, wherein estimating the baseband equivalent model of the power amplifier repeatedly estimates the power amplifier based on a change in transmission characteristics of the power amplifier.