Compensation Method, Compensator and System for Nonlinear Distortion of Pulse Field Source

Abstract

The present invention belongs to the field of communication technology and specifically discloses a compensation method, compensator, and system for nonlinear distortion of a pulsed field source. The compensation method comprises: constructing a Volterra series model for simulating the nonlinear distortion of a digital receiver, and loading the signal to be compensated received by the digital receiver into the Volterra series model to obtain a distorted signal, wherein the distorted signal carries a nonlinear distortion amount, and constructing a compensation model for characterizing the nonlinear distortion amount; using the compensation model to eliminate the nonlinear distortion amount in the distorted signal to obtain a compensation output; based on the compensation output, using the least squares method to update the compensation kernel vector of the compensation model, so as to eliminate the nonlinear distortion amount in the distorted signal in real time using the compensation model after the compensation kernel vector is updated. The present invention solves the problem of nonlinear distortion of a pulsed field source and the difficulty of directly compensating for the nonlinearity of the receiver without adding an additional analog-to-digital converter (ADC).

Description

Technical Field

[0001] The present invention belongs to the field of communication technology, and in particular relates to a compensation method, a compensator and a system for nonlinear distortion of a pulse field source. Background Art

[0002] High-power pulsed field sources operate in the microwave frequency band to generate high-amplitude pulsed radiation fields. These typically utilize electronic vacuum devices such as klystrons and gyrotrons. However, these devices suffer from poor output power linearity, and the distortion of the output waveform varies. Improving the linearity of these pulsed field sources is a critical challenge that urgently needs to be addressed. Due to the physical structure and dispersion characteristics of electronic vacuum devices, improving output linearity from the pulse source itself is difficult. This is often achieved by improving the linearity of receiving devices such as broadband receivers and mixers. However, insufficient third-order output intercept points (IOPs) in RF front-end components such as amplifiers and mixers within broadband receivers can impact the linearity of the digital receiver front-end. Compared to transmitter PAs, high-speed ADCs in digital receivers are often considered ideal due to their reduced nonlinearity. However, in practice, the weak nonlinearity introduced by the receiver front-end ADC can still lead to oscillations, which are difficult to identify and calibrate. This can make it difficult to maintain a high spurious-free dynamic range (SFDR) and result in nonlinear distortion of the received signal. When nonlinear distortion occurs in the receiver, the spurious-free dynamic range of the system decreases, and the collected weak signal is easily overwhelmed by the nonlinear distortion component of the strong signal or other clutter.

[0003] For high-performance broadband digital receivers, weak nonlinearities in analog and hybrid analog-to-digital components can also reduce the potential for dynamic range improvement, impacting the ability to detect weak signals when strong and weak signals coexist. Therefore, the primary task in designing high-dynamic-range broadband digital receivers is to maximize the linearity of each component. Due to current process limitations, achieving ideal linearity simply through continuous optimization of analog circuits is difficult. Currently, utilizing digital compensation techniques to mitigate nonlinearities in the receiver front end is a more common approach. Currently, widely used linearization techniques include power backoff, feedback, feedforward, and predistortion. Predistortion is widely used to address nonlinear distortion at the transmitter end. If predistortion is applied to the receiver end, and the nonlinearities exhibited by the RF front end and analog-to-digital converter (ADC) are considered a nonlinear system similar to a PA, the compensation solution is to construct a compensation system similar to the inverse of digital predistortion (DPD) to ensure overall linearity and achieve receiver linearization.

[0004] However, the input signal to the nonlinear system is an analog signal. To obtain the digital input signal of the receiver, an additional analog-to-digital converter (ADC) is required to convert the signal. This structure has three disadvantages. First, the addition of a high-speed analog-to-digital converter (ADC) increases costs. Second, to directly convert the RF signal to baseband as a reference signal, the ADC must have an extremely high sampling rate and conversion accuracy. Third, the newly added analog-to-digital converter (ADC) still has nonlinear effects on the compensation effect. It can be seen that the existing technology cannot directly compensate for the nonlinearity of the receiver without adding an additional analog-to-digital converter (ADC). It is also difficult to improve the linearity of the pulsed field source radiation signal by improving the linearity of the receiving equipment without adding an additional analog-to-digital converter (ADC). Summary of the Invention

[0005] The purpose of the present invention is to provide a method, a compensator and a system for compensating for the nonlinear distortion of a pulsed field source, so as to solve the problem of the nonlinear distortion of the pulsed field source in the related art, that is, it is impossible to directly compensate for the nonlinearity of the receiver without adding an additional analog-to-digital converter ADC.

[0006] In order to achieve the above object, the first aspect of the present invention provides a method for compensating for nonlinear distortion of a pulsed field source, comprising:

[0007] A Volterra series model for simulating nonlinear distortion of a digital receiver is constructed using a Volterra series as a functional series, and a signal to be compensated received by the digital receiver is loaded into the Volterra series model to obtain a distorted signal, wherein the distorted signal carries a nonlinear distortion amount;

[0008] Constructing a compensation model for characterizing the nonlinear distortion amount according to the Volterra series model, the compensation model comprising a nonlinear memory matrix and a compensation kernel vector, the nonlinear memory matrix being composed of memory nonlinear column vectors of each order constructed by the Volterra series model, and the compensation kernel vector being composed of kernel coefficients of each order of the Volterra series model;

[0009] Eliminating the nonlinear distortion in the distorted signal by using the compensation model to obtain a compensated output;

[0010] Based on the compensation output, the compensation kernel vector of the compensation model is updated by using a least square method, so as to eliminate the nonlinear distortion in the distorted signal in real time by using the compensation model updated by the compensation kernel vector.

[0011] Furthermore, based on the compensation output, the step of updating the compensation kernel vector of the compensation model using the least squares method includes:

[0012] Extracting a nonlinear distortion component from the compensation output, the nonlinear distortion component being a value of the nonlinear distortion amount in the distorted signal obtained by loading the kth signal to be compensated into the Volterra series model, where k is a positive integer;

[0013] Determining the nonlinear signal power of the nonlinear distortion amount according to the extracted k nonlinear distortion components;

[0014] Based on the nonlinear signal power, the least squares method is used to iteratively operate the compensation kernel vector of the compensation model to obtain the compensation kernel vector after data update, and the updated compensation kernel vector is loaded into the compensation model to use the compensation model to eliminate the nonlinear distortion in the distorted signal in real time.

[0015] Furthermore, based on the compensation output, the step of updating the compensation kernel vector of the compensation model using the least squares method further includes:

[0016] Determining a power spectral density of the distorted signal, and determining frequency band information of the nonlinear distortion amount according to the power spectral density;

[0017] constructing a multi-passband filter according to the frequency band information;

[0018] extracting the nonlinear distortion component from the compensation output using the multi-passband filter, wherein the nonlinear distortion component is a value of the nonlinear distortion amount in the distorted signal obtained by loading the kth signal to be compensated into the Volterra series model, where k is a positive integer;

[0019] Determining the nonlinear signal power of the nonlinear distortion amount according to the extracted k nonlinear distortion components;

[0020] Based on the nonlinear signal power, the least squares method is used to iteratively operate the compensation kernel vector of the compensation model to obtain the compensation kernel vector after data update, and the updated compensation kernel vector is loaded into the compensation model to use the compensation model to eliminate the nonlinear distortion in the distorted signal in real time.

[0021] Furthermore, the step of determining the power spectrum density of the distorted signal and determining the frequency band information of the nonlinear distortion amount according to the power spectrum density includes:

[0022] Performing a discrete Fourier transform on the distorted signal to obtain the power spectrum density, and drawing a power spectrum density graph based on the power spectrum density;

[0023] Based on a preset power spectrum threshold, a power spectrum density within a first Nyquist band in the power spectrum density graph and below the power spectrum threshold is determined as the power spectrum density of the nonlinear distortion amount.

[0024] Furthermore, the step of performing an iterative operation on the compensation kernel vector of the compensation model using a least squares method based on the nonlinear signal power includes:

[0025] Taking the operation goal of minimizing the nonlinear signal power as the cost function of the least squares method, and constructing an autocorrelation matrix according to the cost function;

[0026] Based on the inverse matrix of the autocorrelation matrix, the least square method is used to perform an iterative operation on the compensation kernel vector of the compensation model.

[0027] A second aspect of the present invention provides a compensator for nonlinear distortion of a digital receiver, comprising:

[0028] a nonlinear simulation module, configured to construct a Volterra series model for simulating nonlinear distortion of a digital receiver using a Volterra series as a functional series, and to load a signal to be compensated received by the digital receiver into the Volterra series model to obtain a distorted signal, wherein the distorted signal carries a nonlinear distortion amount;

[0029] a compensation model construction module, configured to construct a compensation model for characterizing the nonlinear distortion amount based on the Volterra series model, wherein the compensation model includes a nonlinear memory matrix and a compensation kernel vector, wherein the nonlinear memory matrix is composed of memory nonlinear column vectors of each order constructed by the Volterra series model, and the compensation kernel vector is composed of kernel coefficients of each order of the Volterra series model;

[0030] a compensation output acquisition module, configured to eliminate the nonlinear distortion in the distorted signal using the compensation model to obtain a compensation output;

[0031] A real-time elimination module is used to update the compensation kernel vector of the compensation model based on the compensation output using a least squares method, so as to eliminate the nonlinear distortion in the distorted signal in real time using the compensation model updated with the compensation kernel vector.

[0032] Furthermore, the real-time elimination module includes:

[0033] an extraction unit, configured to extract a nonlinear distortion component from the compensation output, wherein the nonlinear distortion component is a value of the nonlinear distortion amount in the distorted signal obtained by loading the kth signal to be compensated into the Volterra series model, where k is a positive integer;

[0034] a determining unit, configured to determine a nonlinear signal power of the nonlinear distortion amount according to the extracted k nonlinear distortion components;

[0035] An iterative operation unit is used to perform an iterative operation on the compensation kernel vector of the compensation model based on the nonlinear signal power using a least squares method to obtain the compensation kernel vector after data update, and load the updated compensation kernel vector into the compensation model to use the compensation model to eliminate the nonlinear distortion in the distorted signal in real time.

[0036] Furthermore, the real-time elimination module further includes:

[0037] a frequency band determining unit, configured to determine a power spectrum density of the distorted signal, and determine frequency band information of the nonlinear distortion amount according to the power spectrum density;

[0038] A filter construction unit, configured to construct a multi-passband filter according to the frequency band information;

[0039] an extraction unit, configured to extract the nonlinear distortion component from the compensation output using the multi-passband filter, wherein the nonlinear distortion component is a value of the nonlinear distortion amount in the distorted signal obtained by loading the kth signal to be compensated into the Volterra series model, where k is a positive integer;

[0040] a determining unit, configured to determine a nonlinear signal power of the nonlinear distortion amount according to the extracted k nonlinear distortion components;

[0041] An iterative operation unit is used to perform an iterative operation on the compensation kernel vector of the compensation model based on the nonlinear signal power using a least squares method to obtain the compensation kernel vector after data update, and load the updated compensation kernel vector into the compensation model to use the compensation model to eliminate the nonlinear distortion in the distorted signal in real time.

[0042] A third aspect of the present invention provides a signal compensation system, comprising:

[0043] A digital receiver, comprising a compensator for nonlinear distortion of the digital receiver;

[0044] A pulsed field source, wherein the pulsed field source is used to generate a pulsed radiation signal;

[0045] The digital receiver receives the pulse radiation signal and loads the pulse radiation signal as a signal to be compensated into the compensator for nonlinear distortion of the digital receiver.

[0046] Furthermore, the digital receiver further includes:

[0047] RF front end;

[0048] An analog-to-digital converter, wherein the input end of the analog-to-digital converter is electrically connected to the output end of the radio frequency front end, and the output end of the analog-to-digital converter is electrically connected to the compensator for nonlinear distortion of a digital receiver.

[0049] Compared with the existing technology, the technical solution provided by this application has at least the following technical effects:

[0050] The present invention does not require an additional analog-to-digital converter (ADC) to collect original input data, and directly compensates the signal received by the digital receiver. The original input signal can still largely eliminate the nonlinear distortion of the digital receiver without the need to obtain the original input signal. By identifying the frequency distribution of the nonlinear component, a filter is constructed to extract the nonlinear amount, and a cost function of the signal energy of the nonlinear distortion amount in the compensated signal is constructed to identify and update the compensation model parameters. This method does not require an additional ADC to obtain the actual input signal, and can better improve the spurious free dynamic range (SFDR) performance of the signal compensation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are intended to explain the present invention and do not constitute an undue limitation of the present invention. In the accompanying drawings:

[0052] Figure 1 A schematic flow chart of a method for compensating for nonlinear distortion of a pulsed field source provided by one embodiment of the present invention;

[0053] Figure 2 for Figure 1 This is the specific execution process intention of step S14;

[0054] Figure 3 A schematic structural diagram of a compensator for nonlinear distortion of a pulsed field source provided by one embodiment of the present invention;

[0055] Figure 4 A schematic structural diagram of a signal compensation system provided by an embodiment of the present invention;

[0056] Figure 5 A diagram of a nonlinear post-compensation structure established in one embodiment of the present invention;

[0057] Figure 6Schematic diagram of the post-compensation simulation experiment steps of a digital receiver according to an embodiment of the present invention;

[0058] Figure 7 A power spectrum diagram of a simulated dual-tone signal with nonlinear distortion in one embodiment of the present invention;

[0059] Figure 8 A flow chart of updating adaptive parameters of a compensation model provided by an embodiment of the present invention;

[0060] Figure 9 for Figure 6 Power spectrum of the output signal after simulation compensation.

[0061] Description of reference numerals:

[0062] 10. Digital receiver; 11. Compensator; 111. Nonlinear simulation module; 112. Compensation model construction module; 113. Compensation output acquisition module; 114. Real-time elimination module; 141. Frequency band determination unit; 142. Filter construction unit; 143. Extraction unit; 144. Determination unit; 145. Iterative operation unit; 12. Analog-to-digital converter; 13. RF front end; 20. Pulse field source. DETAILED DESCRIPTION

[0063] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The advantages and features of the present invention will become more apparent from the following description and claims. It should be noted that the drawings are all in a very simplified form and are not to exact scale. They are only used for the purpose of conveniently and clearly illustrating the embodiments of the present invention.

[0064] It should be noted that, in order to clearly illustrate the contents of the present invention, the present invention specifically provides multiple embodiments to further illustrate different implementations of the present invention. These multiple embodiments are provided in an enumerated manner rather than an exhaustive manner. Furthermore, for the sake of brevity, the contents mentioned in the previous embodiments are often omitted in the subsequent embodiments. Therefore, for the contents not mentioned in the subsequent embodiments, reference may be made to the previous embodiments accordingly.

[0065] The first embodiment of the present invention provides a method for compensating nonlinear distortion of a pulsed field source, such as Figure 1 As shown, the compensation method includes the following steps:

[0066] Step S11: using the Volterra series as a functional series to construct a Volterra series model for simulating the nonlinear distortion of the digital receiver 10, and loading the signal to be compensated received by the digital receiver 10 into the Volterra series model to obtain a distorted signal, wherein the distorted signal carries a nonlinear distortion amount.

[0067] Step S12: Based on the Volterra series model, a compensation model for characterizing the amount of nonlinear distortion is constructed. The compensation model includes a nonlinear memory matrix and a compensation kernel vector. The nonlinear memory matrix is composed of memory nonlinear column vectors of each order constructed by the Volterra series model. The compensation kernel vector is composed of kernel coefficients of each order of the Volterra series model.

[0068] Step S13: using the compensation model to eliminate the nonlinear distortion in the distorted signal to obtain a compensated output.

[0069] Step S14: Based on the compensation output, the compensation kernel vector of the compensation model is updated using the least square method, so as to eliminate the nonlinear distortion in the distorted signal in real time using the compensation model updated with the compensation kernel vector.

[0070] Specifically, since the memory effect of the broadband digital receiver 10 is particularly obvious, the Volterra series, as a functional series, can theoretically simulate all systems that can be described by continuous functionals. It is a commonly used memory nonlinear model with universal applicability. Therefore, the embodiment of the present invention uses a discrete Volterra series model to simulate the nonlinear distortion of the digital receiver 10. The expression of the Volterra series model is:

[0071]

[0072] Where x(k) is the kth discrete input signal (i.e., the signal to be compensated), k is a positive integer, y(k) is the corresponding output after x(k) enters the Volterra series model, d is the nonlinear order of the input, and D is the maximum value of d. d is the maximum memory depth of the d-th order Volterra kernel. h(r1, r2, ..., r d ) is the kernel coefficient corresponding to the d-th order Volterra series. The total number of kernel coefficients of the Volterra series model is:

[0073]

[0074] In practice, as the memory depth increases, the signal amplitude attenuates, and the impact of the attenuated high-order nonlinear distortion on system linearity decreases sharply. For a receiver, low-order nonlinear distortion has the greatest impact on the system's spurious-free dynamic range. Therefore, a Volterra series model with low-order short-memory effects can be used to simulate receiver nonlinear distortion. For example, a Volterra series model with a maximum memory depth of 2 for the second and third order terms can be used.

[0075] The compensation model in step S12 should simulate the nonlinear distortion of the digital receiver 10 as closely as possible. Therefore, the compensation model uses a structure similar to the Volterra series model in step S11 to characterize the nonlinear memory portion of the distorted signal. The expression of the compensation model B(k) is:

[0076] B(k)=A T (k)×w.

[0077] Where T is the matrix transpose operator, and the matrix A representing the nonlinear memory part is defined as T (k) (i.e., nonlinear memory matrix) is a G*N matrix, where the expression of A(k) is:

[0078] A(k)=[v(k-G+1)v(k-G+2)...v(k-1)v(k)].

[0079] Where v(k) is a memory nonlinear column vector of each order constructed by y(k) that does not contain linear terms, and w is the compensation kernel vector to be identified.

[0080] v(k)=[y 2 (k)y(k)y(k-1)…y 2 (kN d +1)y 3 (k)y 2 (k)y(k-1)…y D (kN d +1)] T .

[0081] w=[h(0,0)h(0,1)…h(N2-1,N2-1)h(0,0,0)h(0,0,1)…h(N D -1,…,N D -1)] T , where h is the kernel coefficient of each order obtained by expanding the Volterra series model expression.

[0082] In step S13, the compensation model is used to eliminate the nonlinear distortion in the distorted signal, and the obtained compensation output is: s(k) = y(k) - B(k). To facilitate subsequent calculations, A(k) and y(k) are stored as memory parts of the same length for subsequent calculations. Therefore, the expression of y(k) can be expressed as:

[0083] y(k)=[y(k-G+1)y(k-G+2)…y(k-1)y(k)] T , since v(k) in A(k) is a memory nonlinear column vector constructed by y(k), y(k) in the y(k) expression is the quantity processed by x(k) through memory nonlinearity.

[0084] like Figure 2 As shown, in step S14, based on the compensation output, the step of using the least squares method to update the compensation kernel vector of the compensation model includes:

[0085] Step S141: determining the power spectrum density of the distorted signal, and determining frequency band information of the nonlinear distortion amount according to the power spectrum density.

[0086] Step S142: constructing a multi-passband filter according to the frequency band information.

[0087] Step S143: extracting a nonlinear distortion component from the compensation output using a multi-passband filter. The nonlinear distortion component is the value of the nonlinear distortion amount in the distorted signal obtained by loading the kth signal to be compensated into the Volterra series model.

[0088] Step S144: determining the nonlinear signal power of the nonlinear distortion amount according to the extracted k nonlinear distortion components.

[0089] Step S145: Based on the nonlinear signal power, the least squares method is used to iteratively calculate the compensation kernel vector of the compensation model to obtain the compensation kernel vector after data update, and the updated compensation kernel vector is loaded into the compensation model to use the compensation model to eliminate the nonlinear distortion in the distorted signal in real time.

[0090] In step S141, the steps of determining the power spectrum density of the distorted signal and determining the frequency band information of the nonlinear distortion amount based on the power spectrum density include: performing a discrete Fourier transform on the distorted signal to obtain the power spectrum density, and drawing a power spectrum density graph based on the power spectrum density, and based on a pre-set power spectrum threshold, determining the power spectrum density within the first Nyquist band of the power spectrum density graph that is below the power spectrum threshold as the power spectrum density of the nonlinear distortion amount. Specifically, for the entire receiving frequency band of the broadband digital receiver 10, the received strong signals are usually distributed relatively sparsely, and the nonlinear distortion components of the digital receiver 10 system can be considered to be mainly caused by the strong signals. The N-point discrete Fourier transform of the distorted signal is obtained to obtain the power spectrum density of the distorted signal. The power spectrum density expression is:

[0091]

[0092] Then draw the power spectrum density diagram of the distorted signal and set the appropriate power spectrum threshold P hold , then it can be considered that the power spectrum density in the first Nyquist band in the power spectrum density diagram exceeds the power spectrum threshold P hold The signal is the strong signal to be received. hold The weak signal below can be considered as the amount of nonlinear distortion, from which the approximate distribution frequency band of the nonlinear distortion can be obtained.

[0093] To suppress or even eliminate the nonlinear distortion of the compensated digital receiver 10, in step S145, the step of iteratively calculating the compensation kernel vector of the compensation model using the least squares method based on the nonlinear signal power includes: using the operational objective of minimizing the nonlinear signal power as the cost function of the least squares method, constructing an autocorrelation matrix based on the cost function, and iteratively calculating the compensation kernel vector of the compensation model using the least squares method based on the inverse matrix of the autocorrelation matrix. Due to the limitations of matrix inversion in some hardware implementations, in the embodiment of the present invention, the inverse matrix of the autocorrelation matrix can be replaced by the cross-correlation vector of the cost function during the specific calculation. After obtaining the updated compensation kernel vector, the updated compensation kernel vector is loaded into the compensation model to eliminate the nonlinear distortion.

[0094] The compensation method provided by the embodiment of the present invention does not require an additional analog-to-digital converter (ADC) to collect original input data, and directly compensates the signal received by the digital receiver 10. It is not necessary to obtain the original input signal and can still largely eliminate the nonlinear distortion of the digital receiver 10. By identifying the frequency distribution of the nonlinear component, constructing a filter to extract the nonlinear amount, constructing a cost function of the signal energy of the nonlinear distortion amount in the compensated signal, and using this to identify and update the compensation model parameters, this method does not require an additional ADC to obtain the actual input signal, and can better improve the system's spurious-free dynamic range (SFDR) performance.

[0095] The second embodiment of the present invention provides a compensator for nonlinear distortion of a pulsed field source, such as Figure 3 As shown, the system includes a nonlinear simulation module 111, a compensation model construction module 112, a compensation output acquisition module 113, and a real-time elimination module 114. The nonlinear simulation module 111 is configured to use the Volterra series, a functional series, to construct a Volterra series model for simulating the nonlinear distortion of the digital receiver 10. The system then loads the signal to be compensated received by the digital receiver 10 into the Volterra series model to obtain a distorted signal, wherein the distorted signal carries a nonlinear distortion amount. The compensation model construction module 112 is configured to construct a compensation model for characterizing the nonlinear distortion amount based on the Volterra series model. The compensation model includes a nonlinear memory matrix and a compensation kernel vector. The nonlinear memory matrix is composed of memory nonlinear column vectors of various orders constructed by the Volterra series model, and the compensation kernel vector is composed of kernel coefficients of various orders of the Volterra series model. The compensation output acquisition module 113 is configured to use the compensation model to eliminate the nonlinear distortion amount in the distorted signal to obtain a compensated output. The real-time elimination module 114 is configured to update the compensation kernel vector of the compensation model using the least square method based on the compensation output, so as to eliminate the nonlinear distortion in the distorted signal in real time using the compensation model updated with the compensation kernel vector.

[0096] The real-time elimination module 114 primarily includes an extraction unit 143, a determination unit 144, and an iterative operation unit 145. The extraction unit 143 is configured to extract a nonlinear distortion component from the compensation output. The nonlinear distortion component is the value of the nonlinear distortion in the distorted signal obtained by loading the kth signal to be compensated into the Volterra series model. The determination unit 144 is configured to determine the nonlinear signal power of the nonlinear distortion based on the k extracted nonlinear distortion components. The iterative operation unit 145 is configured to iteratively operate on the compensation kernel vector of the compensation model using the least squares method based on the nonlinear signal power to obtain an updated compensation kernel vector. The updated compensation kernel vector is then loaded into the compensation model to eliminate the nonlinear distortion in the distorted signal in real time using the compensation model.

[0097] In a specific application example of an embodiment of the present invention, the real-time elimination module 114 further includes a frequency band determination unit 141 and a frequency band determination unit 142. The frequency band determination unit 141 is configured to determine the power spectral density of the distorted signal and determine the frequency band information of the nonlinear distortion amount based on the power spectral density. The filter construction unit 142 is configured to construct a multi-passband filter based on the frequency band information. The extraction unit 143 is configured to extract the nonlinear distortion component from the compensation output using the multi-passband filter constructed by the filter construction unit 142. The nonlinear distortion component is the value of the nonlinear distortion amount in the distorted signal obtained by loading the kth signal to be compensated into the Volterra series model. The determination unit 144 is configured to determine the nonlinear signal power of the nonlinear distortion amount based on the extracted k nonlinear distortion components. The iterative operation unit 145 is configured to iteratively operate the compensation kernel vector of the compensation model using the least squares method based on the nonlinear signal power to obtain the updated compensation kernel vector, and load the updated compensation kernel vector into the compensation model to eliminate the nonlinear distortion amount in the distorted signal in real time using the compensation model.

[0098] The third embodiment of the present invention provides a signal compensation system, such as Figure 4 As shown, the signal compensation system includes a pulsed field source 20 and a digital receiver 10. The pulsed field source 20 is used to generate a pulsed radiation signal. The digital receiver 10 includes a compensator 11 for nonlinear distortion of the digital receiver 10. The digital receiver 10 receives the pulsed radiation signal and loads the pulsed radiation signal as the signal to be compensated into the compensator 11 for nonlinear distortion of the digital receiver 10. This improves the linearity of the signal radiated by the pulsed field source 20 by improving the linearity of the digital receiver 10 without adding an additional analog-to-digital converter (ADC).

[0099] The digital receiver 10 also includes a radio frequency front end 13 and an analog-to-digital converter 12. The input of the analog-to-digital converter 12 is electrically connected to the output of the radio frequency front end 13, and the output of the analog-to-digital converter 12 is electrically connected to a compensator 11 for nonlinear distortion of the digital receiver 10. The radio frequency front end 13 receives the pulsed radiation signal generated by the pulsed field source 20 and sends it to the analog-to-digital converter 12. The analog-to-digital converter 12 converts the pulsed radiation signal into a digital signal and sends it as a signal to be compensated to the compensator 11 for compensation processing, thereby eliminating the nonlinear distortion in the pulsed radiation signal and improving the spurious free dynamic range (SFDR) performance of the signal compensation system.

[0100] The fourth embodiment of the present invention combines the above three embodiments and the appended Figure 1 To the attached Figure 9 In order to solve the nonlinear distortion problem existing in the broadband digital receiver 10, a post-compensation model construction method based on the least squares method is proposed, and an application embodiment of the invention is provided in which the nonlinear distortion component of the signal is eliminated by using the compensation model to restore the received signal.

[0101] The post-compensation model construction method based on the least squares method in the embodiment of the present invention has the following specific steps:

[0102] Step 1: Select a suitable nonlinear model to simulate the nonlinear distortion of the RF front end 13 and the analog-to-digital converter 12 (hereinafter referred to as ADC in the embodiment of the present invention) of the digital receiver 10, load the original signal, and obtain the distorted signal, such as Figure 5 shown.

[0103] Since most transceivers have a certain memory effect, the memory effect is particularly obvious for the wideband digital receiver 10. In this embodiment of the present invention, a discrete Volterra series model is selected to simulate the nonlinear distortion of the RF front end 13 and the analog-to-digital converter 12 of the digital receiver 10. The Volterra series model expression is:

[0104]

[0105] Where x(k) is the kth discrete input signal (i.e., the kth discrete input signal to be compensated), y(k) is the corresponding output of x(k) entering the Volterra series model, d is the nonlinear order of the input, D is the maximum value of d, and N is the maximum value of d. d is the maximum memory depth of the d-th order Volterra kernel, h(r1, r2, ..., r d ) is the kernel coefficient corresponding to the d-th order Volterra. The total number of kernel coefficients of the model is:

[0106]

[0107] In practice, as the memory depth increases, the signal amplitude will attenuate, and the impact of the attenuated high-order nonlinear distortion on system linearity will be sharply reduced. For digital receiver 10, low-order nonlinear distortion has the greatest impact on the system's spurious-free dynamic range. Therefore, a Volterra series model with low-order short-memory effects can be used to simulate system distortion. For example, a Volterra series model with a maximum memory depth of 2 for the second and third order terms can be used to simulate the nonlinear distortion of digital receiver 10.

[0108] Step 2: Identify strong signals and distorted signals. For the entire receiving frequency band of the broadband digital receiver 10, the received strong signals are usually sparsely distributed, and the nonlinear distortion components of the receiver system can be considered to be mainly caused by strong signals. An N-point discrete Fourier transform is performed on the distorted signal to obtain the power spectral density of the distorted signal. The power spectral density expression is:

[0109]

[0110] Draw the power spectrum density diagram of the distorted signal and set the appropriate power spectrum threshold P hold , then it can be considered that the power spectrum density in the first Nyquist band in the power spectrum density graph exceeds the threshold P hold The signal is the strong signal to be received. hold The weak signal below can be considered as nonlinear distortion, from which the approximate distribution frequency band of nonlinear distortion can be obtained. Based on the frequency information (or frequency band information) of nonlinear distortion, a multi-band filter is constructed to obtain the filter coefficients. The FIR tap coefficient vector of the filter coefficients is defined as g = [g0g1…g G-1 ] T , the matrix dimension is G*l, and T is the transpose operator.

[0111] Step 3: The compensation model of nonlinear distortion should simulate the nonlinear distortion of the system as much as possible. Therefore, the compensation model adopts a structure similar to the Volterra series model in step 1 to characterize the nonlinear memory part. The expression of the compensation model B(k) is

[0112] B(k)=A T (k)×w

[0113] Define the matrix A that represents the nonlinear memory part of the compensation system T (k), is the G*N matrix

[0114] A(k)=[v(k-G+1)v(k-G+2)…v(k-1)v(k)]

[0115] Definition: Where v(k) is a memory nonlinear column vector of each order constructed by y(k) that does not contain linear terms, and w is the compensation kernel vector of the compensation model to be identified.

[0116] v(k)=[y 2 (k)y(k)y(k-1)…y 2 (kN d +1)y 3 (k)y 2 (k)y(k-1)…y D (kN d +1)] T .

[0117] w=[h(0,0)h(0,1)…h(N2-1,N2-1)h(0,0,0)h(0,0,1)…h(N D -1,…,N D -1)] T .

[0118] The compensation output after compensation is: s(k) = y(k) - B(k), where:

[0119] y(k)=[y(k-G+1)y(k-G+2)…y(k-1)y(k)] T

[0120] The purpose of the compensation algorithm provided by the embodiment of the present invention is to suppress or even eliminate the nonlinear distortion of the system after compensation. To this end, the embodiment of the present invention extracts the kth nonlinear component value from the compensation output s(k):

[0121] s f (k, w) = g T ×s(k)=g r ×[y(k)-B(k)].

[0122] In the subsequent step 4, when calculating the kernel vector of the compensation model, the goal should be to minimize the energy value of its nonlinear component.

[0123] Step 4: Use the least squares method to iteratively calculate the kernel vector of the compensation model:

[0124] The extracted nonlinear signal power is expressed by the sum of the squares of the K discrete data (nonlinear component values). The nonlinear signal power (or signal energy) can be expressed as:

[0125]

[0126] in represents the short-term energy of the compensation output of the digital receiver 10 at a unit time point at the kth discrete input signal. Taking the minimum value of P(w) as the cost function of the least squares method, the iterative formula can be obtained:

[0127] w(i)=w(i-1)-{Q T [w(i-1)]*Q[w(i-1)]}-1*Q T [w(i-1)]*s f [w(i-1)];

[0128] Where Q(w) is the K*N dimensional first-order derivative matrix. The iterative formula can be further expressed as:

[0129] in, Represents the accumulation of each element in a nonlinear vector.

[0130] Since the least squares RLS algorithm converges quickly, it can improve the parameter identification capability under the premise of real-time changes in the input signal. The cost function can be written as:

[0131]

[0132] Where τ is the forgetting factor, 0<τ<1, and τ is related to the rate of change of the system's nonlinear distortion.

[0133] The autocorrelation matrix R(i) and cross-correlation vector M(i) are obtained as follows:

[0134] R(i)=τR(i-1)+V(k)gg T V T (k)

[0135] M(i)=τM(i-1)+g T y(k)V(k)g

[0136] The update formula of the compensation kernel vector is deduced as follows:

[0137] w(i)=w(i-1)-R -1 (i){[V(k)gg T V T (k)]w(i-1)-V(k)gg T y(k))}

[0138] Since matrix inversion has limitations in some hardware implementations, it can also be assumed that R -1 (i)=M(i) instead of calculation.

[0139] Finally, the kernel coefficients of the updated compensation kernel vector are loaded into the compensation model to complete the nonlinear distortion cancellation. The specific elimination principle is as follows: Figure 5 shown.

[0140] Therefore, the advantages of the embodiments of the present invention are: it can identify the frequency distribution of nonlinear distortion, construct a filter to extract the nonlinear distortion component, construct a nonlinear energy cost function in the compensation signal and use this to identify the compensation model parameters. This method does not require adding an additional ADC to obtain the actual input signal, and can still eliminate a large amount of nonlinear distortion in the system, and can better improve the system's spurious free dynamic range (SFDR) performance.

[0141] The present invention addresses the problem that nonlinear distortion in a broadband digital receiver 10 affects the system's spurious-free dynamic range (SFDR), and proposes a post-compensation model construction method based on the least squares method. This method does not require adding an additional ADC to acquire a reference signal, can weaken the nonlinear distortion component of the signal, reduce the normalized mean square error (NMSE) of the signal, and improve the system's spurious-free dynamic reception range.

[0142] like Figure 6 As shown, for the convenience of explanation, the embodiment of the present invention uses a dual-tone signal to perform a simulation experiment to verify the reliability of the above-mentioned post-compensation model in eliminating nonlinear distortion. The specific steps are as follows:

[0143] Step 1: Select a suitable nonlinear model to simulate the nonlinear distortion of the RF front end 13 and the ADC, load the test signal, and obtain the distorted signal. Generate input data x(k):

[0144]

[0145] Where fs is the sampling frequency, set to 100 MHz. f1 and f2 are the dual-tone signal frequencies, set to 6.3 MHz and 10.3 MHz respectively. Taking the maximum memory depth of the second and third order terms of the Volterra series as 2, the constructed nonlinear series model is:

[0146] y(k)=Y(k)*W

[0147] Y(k)=[x(k),x 2 (k), x(k)x(k-1), x 2 (k-1), x 3 (k), x 2 (k)x(k-1), x(k)x 2 (k-1), x 3 (k-1)], W=10 -2 ×[100, 0.028, 1.232, 0.667, 0.923, 1.188, 1.383, 1.107].

[0148] W is the kernel vector of the nonlinear series model. Substituting x(k) into the model, we can obtain the distorted output signal y(k) of the receiver.

[0149] Step 2: Draw a power density graph, set a threshold, and identify strong signals and distorted signals. For the entire receiving frequency band of a broadband receiver, the received strong signals are usually sparsely distributed, and the nonlinear distortion components of the receiver system can be considered to be mainly caused by strong signals. Calculate the N-point discrete Fourier transform of the signal, calculate the power spectrum density of the signal, and draw a power spectrum density graph as shown in the figure. Figure 7 shown.

[0150] from Figure 7 As can be seen from the figure, the signal distortion produces multiple distortion components with high power values. The system's maximum spurious-free power spectrum range (SFDR) is only 31dB.

[0151] Step 3: The frequency distribution of the distorted signal can be roughly estimated from the power spectrum. Based on the frequency information of the nonlinear distortion, a multi-band filter is constructed and the filter coefficients are derived. The FIR tap coefficient vector is defined as g = [g0g1…g G-1 ] T , the dimension is G*1. Here the coefficient length of the filter is set to 150.

[0152] Step 4: Use the least squares method to iteratively calculate the kernel vector of the compensation model. The entire adaptive parameter update process is as follows: Figure 8 shown.

[0153] Construct the autocorrelation matrix R(i): R(i)=τR(i-1)+V(k)gg T V T (k).

[0154] The least squares RLS algorithm is used to update the compensation model parameters, and the number of training times is set to 200.

[0155] w(i)=w(i-1)-R -1 (i){[V(k)gg T V T (k)]w(i-1)-V(k)gg T y(k)}}

[0156] Finally, the calculated coefficients are updated to the compensation model B(k): B(k) = A T (k)×w, where:

[0157] A(k)=[v(k-G+1)v(k-G+2)...v(k-1)v(k)].

[0158] v(k)=[y 2 (k)y(k)y(k-1)…y 2 (kN d +1)y 3 (k)y 2 (k)y(k-1)…y D (kN d +1)] T .

[0159] The spectrum of the compensated signal is as follows: Figure 9 shown.

[0160] from Figure 9 It can be seen that the nonlinear distortion has been effectively suppressed. After compensation using this method, the system's SFDR value has increased from the original 32dB to 53dB, a 22dB improvement. In addition, the normalized mean square error (NMSE) is used to measure the linearization effect of this solution. Its mathematical expression is:

[0161]

[0162] Among them, y r (i) is the actual input data, which is used as the standard reference signal, y s (i) is the compensated signal output after compensation by this scheme, and K is the number of output data involved in the calculation. NMSE compares the deviation between the compensated input signal and the actual signal, so the smaller the value, the better. Through calculation, the NMSE of the system before compensation is negative 20dB, and the NMSE of the system after compensation is significantly reduced to negative 40dB. This shows that the quality of signal reception has been greatly improved after the implementation of this invention, and the nonlinear distortion of the pulse field source output signal has been improved.

[0163] For ease of description, spatially relative terms such as "above", "above", "on the upper surface of", "above", etc. may be used herein to describe the spatial positional relationship of a device or feature to other devices or features as shown in the figures. It should be understood that spatially relative terms are intended to include different orientations of the device in use or operation in addition to the orientation described in the figures. For example, if the device in the drawings is inverted, the device described as "above other devices or structures" or "above other devices or structures" will be positioned as "below other devices or structures" or "below other devices or structures". Thus, the exemplary term "above" can include both "above" and "below". The device can also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatially relative descriptions used here are interpreted accordingly.

[0164] In addition, it should be noted that the use of terms such as "first" and "second" to limit components is only for the convenience of distinguishing the corresponding components. Unless otherwise stated, the above terms have no special meaning and therefore cannot be understood as limiting the scope of protection of this application.

[0165] The above description is merely a preferred embodiment of the present application and is not intended to limit the present application. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A method for compensating nonlinear distortion of a pulsed field source, characterized in that: include: A Volterra series model for simulating nonlinear distortion of a digital receiver (10) is constructed using a Volterra series as a functional series, and a signal to be compensated received by the digital receiver (10) is loaded into the Volterra series model to obtain a distorted signal, wherein the distorted signal carries a nonlinear distortion amount; Constructing a compensation model for characterizing the nonlinear distortion amount according to the Volterra series model, the compensation model comprising a nonlinear memory matrix and a compensation kernel vector, the nonlinear memory matrix being composed of memory nonlinear column vectors of each order constructed by the Volterra series model, and the compensation kernel vector being composed of kernel coefficients of each order of the Volterra series model; Eliminating the nonlinear distortion in the distorted signal by using the compensation model to obtain a compensated output; Based on the compensation output, the compensation kernel vector of the compensation model is updated using a least squares method, so as to eliminate the nonlinear distortion in the distorted signal in real time using the compensation model updated with the compensation kernel vector; Based on the compensation output, the step of updating the compensation kernel vector of the compensation model using the least squares method further includes: Determining a power spectral density of the distorted signal, and determining frequency band information of the nonlinear distortion amount according to the power spectral density; constructing a multi-passband filter according to the frequency band information; extracting the nonlinear distortion component from the compensation output using the multi-passband filter, wherein the nonlinear distortion component is a value of the nonlinear distortion amount in the distorted signal obtained by loading the kth signal to be compensated into the Volterra series model, where k is a positive integer; Determining the nonlinear signal power of the nonlinear distortion amount according to the extracted k nonlinear distortion components; Based on the nonlinear signal power, using a least squares method to iteratively calculate the compensation kernel vector of the compensation model to obtain the compensation kernel vector after data update, and loading the updated compensation kernel vector into the compensation model to eliminate the nonlinear distortion in the distorted signal in real time using the compensation model; The step of performing an iterative operation on the compensation kernel vector of the compensation model using a least squares method based on the nonlinear signal power includes: The nonlinear signal power is expressed as: in The short-time energy of the compensation output of the digital receiver (10) at a unit time point where the k-th discrete input signal is located is represented, and the minimum value of P(w) is used as the cost function of the least square method. The cost function is: Where τ is the forgetting factor, 0<τ<1, and τ is related to the rate of change of the system's nonlinear distortion. The cross-correlation vector M(i) is obtained as: M(i)=τM(i-1)+g T y(k)V(k)g; Then the update formula of the compensation kernel vector is: w(i)=w(i-1)-M(i){[V(k)gg T V T (k)]w(i-1)-V(k)gg T y(k)}}。 2. The method for compensating for nonlinear distortion of a pulsed field source according to claim 1, wherein: Based on the compensation output, the step of updating the compensation kernel vector of the compensation model using the least squares method includes: Extracting a nonlinear distortion component from the compensation output, the nonlinear distortion component being a value of the nonlinear distortion amount in the distorted signal obtained by loading the kth signal to be compensated into the Volterra series model, where k is a positive integer; Determining the nonlinear signal power of the nonlinear distortion amount according to the extracted k nonlinear distortion components; Based on the nonlinear signal power, the least squares method is used to iteratively operate the compensation kernel vector of the compensation model to obtain the compensation kernel vector after data update, and the updated compensation kernel vector is loaded into the compensation model to use the compensation model to eliminate the nonlinear distortion in the distorted signal in real time.

3. The method for compensating for nonlinear distortion of a pulsed field source according to claim 1, wherein: The steps of determining the power spectrum density of the distorted signal and determining the frequency band information of the nonlinear distortion amount according to the power spectrum density include: Performing a discrete Fourier transform on the distorted signal to obtain the power spectrum density, and drawing a power spectrum density graph based on the power spectrum density; Based on a preset power spectrum threshold, a power spectrum density within a first Nyquist band in the power spectrum density graph and below the power spectrum threshold is determined as the power spectrum density of the nonlinear distortion amount.

4. The method for compensating for nonlinear distortion of a pulsed field source according to claim 2, wherein: The step of performing an iterative operation on the compensation kernel vector of the compensation model using a least squares method based on the nonlinear signal power includes: Taking the operation goal of minimizing the nonlinear signal power as the cost function of the least squares method, and constructing an autocorrelation matrix according to the cost function; Based on the inverse matrix of the autocorrelation matrix, the least square method is used to perform an iterative operation on the compensation kernel vector of the compensation model.

5. A compensator for nonlinear distortion of a pulsed field source, characterized in that: include: A nonlinear simulation module (111) is configured to construct a Volterra series model for simulating nonlinear distortion of a digital receiver (10) using a Volterra series as a functional series, and to load a signal to be compensated received by the digital receiver (10) into the Volterra series model to obtain a distorted signal, wherein the distorted signal carries a nonlinear distortion amount; A compensation model construction module (112) is used to construct a compensation model for characterizing the nonlinear distortion amount based on the Volterra series model, wherein the compensation model includes a nonlinear memory matrix and a compensation kernel vector, wherein the nonlinear memory matrix is composed of memory nonlinear column vectors of various orders constructed by the Volterra series model, and the compensation kernel vector is composed of kernel coefficients of various orders of the Volterra series model; A compensation output acquisition module (113) is used to eliminate the nonlinear distortion amount in the distorted signal using the compensation model to obtain a compensation output; A real-time elimination module (114) is configured to update the compensation kernel vector of the compensation model using a least square method based on the compensation output, so as to eliminate the nonlinear distortion amount in the distorted signal in real time using the compensation model updated with the compensation kernel vector; The real-time elimination module (114) further includes: A frequency band determination unit (141) is used to determine the power spectrum density of the distorted signal and determine the frequency band information of the nonlinear distortion amount according to the power spectrum density; A filter construction unit (142), configured to construct a multi-passband filter according to the frequency band information; an extraction unit (143) configured to extract the nonlinear distortion component from the compensation output using the multi-passband filter, wherein the nonlinear distortion component is a value of the nonlinear distortion amount in the distorted signal obtained by loading the kth signal to be compensated into the Volterra series model, where k is a positive integer; A determination unit (144) is used to determine the nonlinear signal power of the nonlinear distortion amount based on the extracted k nonlinear distortion components; An iterative operation unit (145) is used to perform an iterative operation on the compensation kernel vector of the compensation model using a least square method based on the nonlinear signal power to obtain the compensation kernel vector after data update, and load the updated compensation kernel vector into the compensation model to eliminate the nonlinear distortion amount in the distorted signal in real time using the compensation model; The iterative operation unit (145) performs iterative operation on the compensation kernel vector of the compensation model using the least square method based on the nonlinear signal power, including the following steps: The nonlinear signal power is expressed as: in The short-time energy of the compensation output of the digital receiver (10) at a unit time point where the k-th discrete input signal is located is represented, and the minimum value of P(w) is used as the cost function of the least square method. The cost function is: Where τ is the forgetting factor, 0<τ<1, and τ is related to the rate of change of the system's nonlinear distortion. The cross-correlation vector M(i) is obtained as: M(i)=τM(i-1)+g T y(k)V(k)g; Then the update formula of the compensation kernel vector is: w(i)=w(i-1)-M(i){[V(k)gg T V T (k)]w(i-1)-V(k)gg T y(k)}}。 6. The compensator for nonlinear distortion of a pulsed field source according to claim 5, characterized in that: The real-time elimination module (114) includes: An extraction unit (143) is used to extract a nonlinear distortion component from the compensation output, wherein the nonlinear distortion component is a value of the nonlinear distortion amount in the distorted signal obtained by loading the kth signal to be compensated into the Volterra series model, where k is a positive integer; A determination unit (144) is used to determine the nonlinear signal power of the nonlinear distortion amount based on the extracted k nonlinear distortion components; An iterative operation unit (145) is used to perform an iterative operation on the compensation kernel vector of the compensation model using a least square method based on the nonlinear signal power to obtain the compensation kernel vector after data update, and load the updated compensation kernel vector into the compensation model to eliminate the nonlinear distortion amount in the distorted signal in real time using the compensation model.

7. A signal compensation system, characterized in that: include: A digital receiver (10), comprising the compensator (11) for nonlinear distortion of a pulsed field source according to any one of claims 5 to 6; A pulsed field source (20), the pulsed field source (20) being used to generate a pulsed radiation signal; The digital receiver (10) receives the pulse radiation signal and loads the pulse radiation signal as a signal to be compensated into a compensator (11) for nonlinear distortion of a digital receiver.

8. The signal compensation system according to claim 7, wherein: The digital receiver further comprises: RF front end (13); An analog-to-digital converter (12), wherein an input end of the analog-to-digital converter (12) is electrically connected to an output end of the radio frequency front end (13), and an output end of the analog-to-digital converter (12) is electrically connected to the compensator (11) for nonlinear distortion of a digital receiver.

Images