A method for improving linearity of a radio frequency power amplifier

Abstract

The application relates to a radio frequency power amplifier linearity improvement method, satisfies the requirement of a high-order modulation communication system on the linearity of a memory type power amplifier, effectively solves the problem of signal distortion caused by the nonlinearity of a radio frequency power amplifier, first, a modeling method is designed to effectively reflect the memory effect of the power amplifier and accurately describe the input-output relationship of the power amplifier, thereby providing support for subsequent predistortion design; then, a predistortion scheme is applied to adjust the input signal on the basis of the accurate model, and the predistortion scheme has the advantages of simple structure, fast calculation speed and relatively easy realization process; in addition, considering the aging and other problems in the actual use process of the device, a real-time updating measure for the predistortion adjustment scheme is designed, the adaptability of the whole system is increased, and the shortcoming that the memory type power amplifier model is only a mathematical model and cannot describe the changes in actual application is made up.

Description

Technical Field

[0001] This invention relates to a method for improving the linearity of a radio frequency power amplifier, belonging to the field of power amplifier technology. Background Technology

[0002] Radio frequency (RF) power amplifiers are a key component of a transmitting system, used in the front-end circuitry of the transmitter. They amplify low-power RF signals through a series of power amplification steps before feeding them to the transmitting antenna. When the input signal amplitude is large, the nonlinearity of the power amplifier can significantly impact the amplification effect. Furthermore, with the continuous improvement of modulation techniques, such as Quadrature Amplitude Modulation (QAM), the number of points in the constellation diagram is increasing in practical applications, placing increasingly higher demands on the linearity of RF components.

[0003] Linearization techniques for power amplifiers generally include predistortion, negative feedback, feedforward, and power back-off. Among these, negative feedback offers good stability but provides relatively poor linearity improvement; feedforward and power back-off methods have poor power-added efficiency (PAE). Predistortion, on the other hand, improves power amplifier linearity while exhibiting superior performance in PAE, bandwidth, stability, and other metrics.

[0004] Establishing a power amplifier model is a prerequisite for implementing predistortion technology. For different application scenarios, appropriate mathematical models for the power amplifier are required. The nonlinear models of RF power amplifiers can be divided into memoryless and memory-based models. Due to the unstable signal bandwidth and hardware temperature of current communication systems, the output signal of the power amplifier is usually related to the historical input components. Therefore, a memory-based model is used for digital modeling of the power amplifier.

[0005] Existing methods to improve the linearity of power amplifiers involve directly adding additional circuitry to the circuit structure. This leads to an increase in the overall area of ​​the integrated circuit, and problems such as temperature changes and device aging also alter the transistor characteristics of the dynamic feedback network itself, resulting in distortion of the feedback signal and a decrease in the ability to adjust linearity. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a method for improving the linearity of radio frequency power amplifiers. Based on a dynamic model, a new compensation control logic is introduced, and necessary factors are considered, which can effectively improve the input and output linearity of radio frequency power amplifiers.

[0007] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention designs a method for improving the linearity of an RF power amplifier, used to improve the linearity of a target memory-type power amplifier, comprising the following steps:

[0008] Step A. Construct the input-output dynamic model of the target memory power amplifier, and based on the input and output signals of the target memory power amplifier at each historical time point, and with the model accuracy as the objective, train the input-output dynamic model to obtain the actual input-output model of the target memory power amplifier, and then proceed to Step B;

[0009] Step B. Obtain the relationship curve between the input and output signals corresponding to the actual input and output model of the target memory power amplifier, and combine it with the preset ideal input and output signal relationship curve to obtain the input and output signal predistortion compensation curve. Further obtain the relationship between the actual input signal of the target memory power amplifier and the predistortion compensation input signal. Based on this relationship, construct a predistorter, and then proceed to step C.

[0010] Step C. The predistorter compensates for the actual input signal of the target memory power amplifier based on the relationship between the actual input signal and the predistortion compensation input signal, forming a compensated input signal. This compensated input signal is then processed by the first filter modulation and input to the target memory power amplifier to obtain the output signal of the target memory power amplifier, thereby improving the linearity of the target memory power amplifier.

[0011] As a preferred technical solution of the present invention: while performing step C, the method further includes performing the following steps i to ii;

[0012] Step i. After sequentially performing coupling attenuation and second filtering demodulation processing on the output signal of the target memory power amplifier, the processed output signal is obtained, and then proceed to step ii;

[0013] Step ii. Based on the output signal obtained in step i, update the output signal predistortion compensation curve, and then update the relationship between the actual input signal and the predistortion compensation input signal of the target memory power amplifier.

[0014] As a preferred embodiment of the present invention: the first filtering modulation process in step C includes processing through a DAC, a filter, and a quadrature modulator in sequence; the second filtering demodulation process in step i includes processing through a quadrature demodulator, a filter, and an ADC in sequence, wherein the operation process of the quadrature modulator in the first filtering modulation process is the opposite of the operation process of the quadrature demodulator in the second filtering demodulation process.

[0015] As a preferred technical solution of the present invention: step A includes the following steps A1 to A4;

[0016] Step A1. Construct the input-output dynamic model of the target memory-type power amplifier as follows:

[0017]

[0018] Where Vd(t) and Vo(t) represent the input and output signals of the target memory power amplifier at historical time point t, respectively; k represents the order of the polynomial; q represents the maximum duration span from historical time point t in the historical time direction; and a ij Represent the coefficients in complex form, and then proceed to step A2;

[0019] Step A2. Based on the discrete values ​​of k and q corresponding to the upper and lower limits of each preset value range, obtain various combinations of k and q, and then proceed to step A3;

[0020] Step A3. For each combination of k and q, based on each historical time point in the q-time span of the combination from historical time point t to historical time direction, train the input-output dynamic model of the target memory power amplifier according to the input and output signals of the target memory power amplifier at each historical time point, obtain the comparison input-output model of the memory power amplifier under this combination, and obtain the corresponding model accuracy value; then proceed to step A4.

[0021] Step A4. Based on the input-output models to be compared and the model accuracy values ​​of the memory power amplifier under various combinations of k and q, obtain the input-output model to be compared corresponding to the maximum model accuracy value, and use it as the actual input-output model of the memory power amplifier.

[0022] As a preferred technical solution of the present invention: in step A3, the model accuracy is measured by the following normalized mean square error;

[0023]

[0024] In the formula, NMSE(dB) represents the model accuracy value, N represents the preset number of sampled data, and Vo(n) represents the actual output signal of the target memory power amplifier. This indicates the measured output of the target memory-type power amplifier.

[0025] The method for improving the linearity of an RF power amplifier described in this invention, compared with existing technologies, has the following technical advantages:

[0026] This invention presents a method for improving the linearity of radio frequency power amplifiers, meeting the linearity requirements of memory-type power amplifiers in high-order modulation communication systems. It effectively solves the problem of signal distortion caused by the nonlinearity of radio frequency power amplifiers. First, a modeling method is designed to effectively reflect the memory effect of the power amplifier and accurately describe the input-output relationship, providing support for subsequent pre-distortion design. Then, a pre-distortion scheme is applied to adjust the input signal based on the accurate model, offering advantages such as simple structure, fast calculation speed, and relatively easy implementation. Furthermore, considering issues such as aging during actual device use, a real-time update mechanism for the pre-distortion adjustment scheme is designed to increase the adaptability of the entire system, overcoming the limitation that the memory-type power amplifier model, being merely a mathematical model, cannot describe changes in actual applications. Attached Figure Description

[0027] Figure 1 This is a flowchart illustrating the method for improving the linearity of a radio frequency power amplifier according to the present invention.

[0028] Figure 2 This is a schematic diagram of the three curves involved in step B of the design and implementation of this invention;

[0029] Figure 3 This is a schematic diagram illustrating the simultaneous execution of step C and steps i to ii in the design of this invention;

[0030] Figure 4 This is a schematic diagram illustrating the principle of simultaneous signal processing in step C and steps i to ii in the design of this invention. Detailed Implementation

[0031] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0032] As the bandwidth of input signals increases, the memory effect of power amplifiers becomes more pronounced, making the modeling of memory-type power amplifiers increasingly important. The memory effect of a power amplifier refers to the fact that the operating state of the device is not only related to the current input signal but also influenced by past envelope levels. Ignoring the memory effect during power amplifier modeling can lead to inaccurate model creation, impacting the reliability of subsequent digital predistortion operations.

[0033] Commonly used memory power amplifier models include: Volterra model, Hammerstein model, Wiener model, Wiener-Hammerstein model, memory polynomial model, and neural network model, etc. Among these, the memory polynomial model is a simplified version of the Volterra model, which can simulate the AM-AM and AM-PM characteristics of multiple power amplifiers. This invention designs a method for improving the linearity of RF power amplifiers, specifically for improving the linearity of a target memory power amplifier. In practical applications, such as... Figure 1 As shown, the specific steps are as follows.

[0034] Step A. Construct the input-output dynamic model of the target memory power amplifier, and based on the input and output signals of the target memory power amplifier at each historical time point, and with the model accuracy as the objective, train the input-output dynamic model to obtain the actual input-output model of the target memory power amplifier, and then proceed to Step B.

[0035] In practical applications, step A above is specifically executed as steps A1 to A4.

[0036] Step A1. Construct the input-output dynamic model of the target memory-type power amplifier as follows:

[0037]

[0038] Where vd(t) and Vo(t) represent the input and output signals of the target memory power amplifier at historical time point t, respectively; k represents the order of the polynomial, reflecting the nonlinearity of the memory polynomial; q represents the maximum duration span from historical time point t to the historical time direction, reflecting the memory effect of the model; a ij Represent the coefficients in complex form, and then proceed to step A2.

[0039] The input-output dynamic model of the target memory-type power amplifier includes higher-order terms and delay terms of the independent variables, which can provide a good description of the memory-type power amplifier. The memory depth q can be understood as: how much of the previous output signal has an impact on the current signal that the model can express. The higher the q, the more accurately it describes the memory effect of the power amplifier, but an excessively high q value will lead to model complexity and affect system operation. The polynomial order k represents the accuracy of the model description in the current time period. The higher the k, the more accurate the model, but like the q value, the higher the k, the more it affects the performance of the system.

[0040] The input-output dynamic model of the target memory power amplifier can be written in matrix form:

[0041] Y = XA

[0042] in

[0043]

[0044]

[0045] Y = [Vo(t)Vo(t-1)...Vo(tq)] T

[0046] From equations (1) and (2), we can obtain:

[0047] A = X -1 Y (3)

[0048] It can be seen that the polynomial coefficient matrix can be obtained from the input and output signal matrices.

[0049] Step A2. Based on the discrete values ​​of k and q corresponding to the upper and lower limits of each preset value range, obtain various combinations of k and q, and then proceed to step A3.

[0050] Step A3. For each combination of k and q, based on each historical time point in the q-time span of the combination from historical time point t to historical time direction, train the input-output dynamic model of the target memory power amplifier according to the input and output signals of the target memory power amplifier at each historical time point, obtain the comparison input-output model of the memory power amplifier under this combination, and obtain the corresponding model accuracy value; then proceed to step A4.

[0051] The model accuracy is measured by the following normalized mean square error;

[0052]

[0053] In the formula, NMSE(dB) represents the model accuracy value, N represents the preset number of sampled data, and Vo(n) represents the actual output signal of the target memory power amplifier. This indicates the measured output of the target memory-type power amplifier.

[0054] Step A4. Based on the input-output models to be compared and the model accuracy values ​​of the memory power amplifier under various combinations of k and q, obtain the input-output model to be compared corresponding to the maximum model accuracy value, and use it as the actual input-output model of the memory power amplifier.

[0055] Step B. Figure 2As shown, the relationship curve between the input and output signals corresponding to the actual input and output model of the target memory power amplifier is obtained. Combined with the preset ideal input and output signal relationship curve, the input and output signal predistortion compensation curve is obtained. Furthermore, the relationship between the actual input signal and the predistortion compensation input signal corresponding to the target memory power amplifier is obtained. Based on this relationship, a predistorter is constructed, and then the process proceeds to step C.

[0056] Step C. Figure 3 and Figure 4 As shown, the predistorter compensates for the actual input signal of the target memory power amplifier based on the relationship between the actual input signal and the predistortion compensation input signal, forming a compensated input signal. This compensated input signal is then processed by the first filter modulation and input to the target memory power amplifier to obtain the output signal of the target memory power amplifier, thereby improving the linearity of the target memory power amplifier.

[0057] In practical applications, while performing step C, such as Figure 3 and Figure 4 As shown, it also includes performing steps i to ii as follows.

[0058] Step i. After sequentially performing coupling attenuation and second filtering demodulation processing on the output signal of the target memory power amplifier, the processed output signal is obtained, and then proceed to step ii.

[0059] Step ii. Based on the output signal obtained in step i, update the output signal predistortion compensation curve, and then update the relationship between the actual input signal and the predistortion compensation input signal of the target memory power amplifier.

[0060] In practical applications, the first filtering and modulation process described above includes processing through a DAC, a filter, and a quadrature modulator in sequence; the second filtering and demodulation process in step i includes processing through a quadrature demodulator, a filter, and an ADC in sequence. The quadrature modulator is used to modulate the received baseband signal onto a high-frequency signal, that is, to place the baseband signal onto the carrier (local oscillator). The quadrature demodulator is used to demodulate the received high-frequency signal onto the baseband signal, that is, to extract the baseband signal from the carrier (local oscillator). In other words, the operation process of the quadrature modulator in the first filtering and modulation process is the opposite of the operation process of the quadrature demodulator in the second filtering and demodulation process.

[0061] like Figure 4 As shown, in specific execution, the amplitude ρ is obtained by performing an r / p transformation (converting rectangular coordinates to polar coordinates) on the actual input signal U(t). t and phase Based on the predistortion compensation curve of the target memory power amplifier, the ρ of the U(t) signal... tMultiplying by the predistortion factor β, the U(t) signal After adjusting the angle σ, it is transformed into θ. t After undergoing p / r transformation (converting polar coordinates to rectangular coordinates), the input X(t) to the upconversion channel is obtained. Then, the output V is obtained through the upconversion channel (DAC, filter, quadrature modulator). d (t) represents the RF input signal of the target memory power amplifier, and then the power amplifier outputs the amplified V. o (t) improves the linearity of the power amplifier. However, since the nonlinear characteristics of the target memory power amplifier are affected by factors such as temperature, humidity, and device aging, a feedback channel is needed to increase the system's adaptability. For example... Figure 4 As shown, the feedback path designed in this invention acquires the real-time output of the power amplifier, and obtains the feedback signal Y(t) through coupling attenuation and a down-conversion channel (quadrature demodulator, filter, ADC). Y(t) is then transformed by r / p to obtain the amplitude Rn and phase ψ. t By comparing the results with the expected results, the predistortion factor β and the adjustment angle σ are updated in real time, which means that the feedback signal is adaptively processed and the relationship between the actual input signal and the predistortion compensation input signal of the target memory power amplifier is updated, so that the predistortion system has higher application capability and realizes the linearization of the power amplifier.

[0062] The above-described method for improving the linearity of RF power amplifiers meets the linearity requirements of memory-type power amplifiers in high-order modulation communication systems, effectively solving the signal distortion problem caused by the nonlinearity of RF power amplifiers. Firstly, a modeling method is designed to effectively reflect the memory effect of the power amplifier and accurately describe the input-output relationship, providing support for subsequent pre-distortion design. Then, the pre-distortion scheme is applied to adjust the input signal based on the accurate model, offering advantages such as simple structure, fast calculation speed, and relatively easy implementation. Furthermore, considering issues such as aging during actual device use, a real-time update mechanism for the pre-distortion adjustment scheme is designed to increase the adaptability of the entire system, overcoming the limitation that the memory-type power amplifier model, being merely a mathematical model, cannot describe changes in actual applications.

[0063] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A method for improving the linearity of a radio frequency power amplifier, used to improve the linearity of a target memory-type power amplifier, characterized in that, Includes the following steps: Step A. Construct the input-output dynamic model of the target memory power amplifier, and based on the input and output signals of the target memory power amplifier at each historical time point, and with the model accuracy as the objective, train the input-output dynamic model to obtain the actual input-output model of the target memory power amplifier, and then proceed to Step B; Step A includes steps A1 to A4 as follows; Step A1. Construct the input-output dynamic model of the target memory-type power amplifier as follows: ； in, , These represent the historical time points corresponding to the target memory-type power amplifier. Input signals and output signals Denotes the order of a polynomial. Indicates from a historical point in time The maximum duration span in the historical timeline. Represent the coefficients in complex form, and then proceed to step A2; Step A2. Based on , For each discrete value within the preset upper and lower limits, obtain and Various combinations, then proceed to step A3; Step A3. Specifically for and Various combinations, based on historical time points In the combination For each historical time point in the time span, based on the input and output signals of the target memory power amplifier corresponding to each historical time point, the input-output dynamic model of the target memory power amplifier is trained to obtain the input-output model of the memory power amplifier under this combination, and at the same time obtain the corresponding model accuracy value; then proceed to step A4; Step A4. Based on and Under various combinations, the input-output models to be compared with the memory power amplifier and the model accuracy values ​​are obtained. The input-output model to be compared with the maximum model accuracy value is obtained as the actual input-output model of the memory power amplifier. Step B. Obtain the relationship curve between the input and output signals corresponding to the actual input and output model of the target memory power amplifier, and combine it with the preset ideal input and output signal relationship curve to obtain the input and output signal predistortion compensation curve. Further obtain the relationship between the actual input signal of the target memory power amplifier and the predistortion compensation input signal. Based on this relationship, construct a predistorter, and then proceed to step C. Step C. The predistorter compensates for the actual input signal of the target memory power amplifier based on the relationship between the actual input signal and the predistortion compensation input signal, forming a compensated input signal. This compensated input signal is then processed by the first filter modulation and input to the target memory power amplifier to obtain the output signal of the target memory power amplifier, thereby improving the linearity of the target memory power amplifier. While performing step C, the process also includes performing steps i to i. ; Step i. After sequentially performing coupling attenuation and second filtering demodulation processing on the output signal of the target memory power amplifier, the processed output signal is obtained, and then proceeds to step i. ; step Based on the output signal obtained in step i, update the output signal predistortion compensation curve, and then update the relationship between the actual input signal and the predistortion compensation input signal of the target memory power amplifier.

2. The method for improving the linearity of an RF power amplifier according to claim 1, characterized in that: The first filtering and modulation process in step C includes processing through a DAC, a filter, and a quadrature modulator in sequence; the second filtering and demodulation process in step i includes processing through a quadrature demodulator, a filter, and an ADC in sequence, wherein the operation process of the quadrature modulator in the first filtering and modulation process is the opposite of the operation process of the quadrature demodulator in the second filtering and demodulation process.

3. The method for improving the linearity of an RF power amplifier according to claim 1, characterized in that: In step A3, the model accuracy is measured by the following normalized mean square error; ； In the formula, This represents the model accuracy value. Indicates the preset number of sampled data. This represents the actual output signal of the target memory-type power amplifier. This indicates the measured output of the target memory-type power amplifier. .

Images