CN114726703A - Power injection type multi-path self-adaptive digital predistortion algorithm and system - Google Patents
Power injection type multi-path self-adaptive digital predistortion algorithm and system Download PDFInfo
- Publication number
- CN114726703A CN114726703A CN202210166421.6A CN202210166421A CN114726703A CN 114726703 A CN114726703 A CN 114726703A CN 202210166421 A CN202210166421 A CN 202210166421A CN 114726703 A CN114726703 A CN 114726703A
- Authority
- CN
- China
- Prior art keywords
- predistortion
- submodule
- algorithm
- model
- power injection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L27/00—Modulated-carrier systems
- H04L27/32—Carrier systems characterised by combinations of two or more of the types covered by groups H04L27/02, H04L27/10, H04L27/18 or H04L27/26
- H04L27/34—Amplitude- and phase-modulated carrier systems, e.g. quadrature-amplitude modulated carrier systems
- H04L27/36—Modulator circuits; Transmitter circuits
- H04L27/366—Arrangements for compensating undesirable properties of the transmission path between the modulator and the demodulator
- H04L27/367—Arrangements for compensating undesirable properties of the transmission path between the modulator and the demodulator using predistortion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B7/00—Radio transmission systems, i.e. using radiation field
- H04B7/02—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
- H04B7/04—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas
- H04B7/0413—MIMO systems
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Amplifiers (AREA)
Abstract
The invention discloses a power injection type multi-path self-adaptive digital predistortion algorithm and a system, which adopt a direct learning framework to construct a module and have the characteristics of simple structure and convenient realization. The predistortion signal consists of three parts: an original baseband signal and two low power baseband signals generated by two predistortion sub-modules. The small-power signal is injected into the baseband input signal to compensate the nonlinearity, memory effect and crosstalk effect of the power amplifier in the MIMO system through the predistortion submodule I and the predistortion submodule II, and when the PAPR value of the baseband signal is large and the nonlinearity of the power amplifier is strong, a good linearization effect can still be obtained in each branch. The predistortion signal is amplified by a power amplifier, and the characteristics of the output signal are observed to find that out-of-band spectrum expansion and in-band distortion are well inhibited, so that the problems of nonlinearity, memory effect and crosstalk of the power amplifier in the MIMO system are solved, and the efficiency of the power amplifier in the MIMO system is improved.
Description
Technical Field
The invention relates to the technical field of wireless communication, in particular to a power injection type multi-channel adaptive Digital Predistortion (DPD) algorithm and a system.
Background
A typical multiple-input multiple-output-orthogonal frequency division multiplexing (MIMO-OFDM) transmitter architecture is proposed at 4G, and the problems of frequency selective fading and multipath interference can be effectively solved on the basis of remarkably improving the transmission rate and the spectrum efficiency of a system. The combination of the MIMO technology and the OFDM technology ensures that the communication system can achieve a high transmission rate and strong system reliability.
In 5G-NR, massive MIMO technology is very effective to improve the indexes of channel capacity, spectral efficiency, and data throughput, and 5G base stations need to comply with their respective transmission requirements or transmission requirements according to their types. The application indicators of the base station are: base station output power, transmission signal quality, signal bandwidth, adjacent channel leakage ratio, harmful radiation of working frequency band, transmitter stray radiation, transmitter intermodulation and the like; the influence of the performance of a Radio Frequency (RF) Power Amplifier (PA) in a base station on the indexes is particularly critical, and the inherent nonlinearity of the RF power amplifier, the memory effect when transmitting a broadband signal, and the crosstalk effect applied in a large-scale MIMO scenario deteriorate the performance of a communication system and affect the signal transmission quality.
Compared with a traditional single-input single-output (SISO) system, the predistorter at the transmitting end of the MIMO system is technically more challenging, as the transmitter carries out up-conversion on a plurality of signals simultaneously, crosstalk can be generated among a plurality of branches when a plurality of branches exist on the same chip, the effect of the crosstalk is more obvious when the size of the chip is smaller, the distortion of the output signals of the radio frequency power amplifier can be directly caused, the accuracy of the coefficient estimation stage of the predistorter is further influenced, and the performance of the predistortion technology is reduced. Therefore, crosstalk suppression is an important problem to be solved when applying predistortion technology to a transmitting end of a multi-antenna MIMO system.
Disclosure of Invention
The invention aims to provide a power injection type multi-path self-adaptive digital predistortion algorithm and a system.
In order to achieve the above purpose, the invention provides the following technical scheme:
a power injection type multi-path adaptive digital predistortion algorithm comprises the following steps:
s1, respectively constructing a digital PA behavior model under SISO conditions through baseband input and output signals of each branch;
s2, introducing crosstalk at the input end, and obtaining crosstalk output by using a memory polynomial model;
s3, constructing a digital PH behavior model under the condition of multiple input and multiple output by using the input and crosstalk output signals of the power amplifier;
s4, assembling a predistortion model for each branch, wherein the predistortion model comprises a predistortion submodule I and a predistortion submodule II, the submodule I is an inverse model of the power amplifier of the branch under the SISO condition, and the predistortion submodule II is an adaptive predistortion submodule based on a direct learning architecture;
and S5, adding a low-power injection signal output by the first predistortion submodule and a low-power injection signal output by the second predistortion submodule on the basis of baseband input of each branch, and taking the low-power injection signals as input of the branch power amplifier under the condition of parallel Hamming model representation.
Further, in step S1, the baseband input and output signals are all subjected to time domain alignment and normalization in advance.
Further, step S1 measures the model accuracy by normalizing the mean square error.
Further, step S3 measures the model accuracy by normalizing the mean square error.
Further, the construction process of the predistortion submodule one in the step S4 is as follows: after the highest order and the maximum memory depth of the inverse model are determined, the prediction output of the power amplifier is converted into a corresponding basis function matrix, a linear equation set is constructed, and a coefficient matrix of the inverse model is obtained by using an LS algorithm.
Further, the construction process of the predistortion submodule two in the step S4 is as follows: the method comprises the steps of firstly establishing an original low-power injection basis function set according to a PH model expression, carrying out orthogonalization treatment on the original low-power injection basis function set, obtaining the orthogonal basis function set after SVD decomposition, wherein under the condition that the original low-power injection basis function set is full-rank, the total number of orthogonal basis functions corresponding to each output signal is equal to the total number of basis functions corresponding to each output signal, and otherwise, the total number of the orthogonal basis functions corresponding to each output signal is smaller than the total number of basis functions corresponding to each output signal.
Further, in step S4, the predistortion submodule obtains the optimal coefficient matrix through the smart group optimization algorithm.
Further, in step S4, the predistortion submodule two times obtains the optimal coefficient matrix through the particle swarm optimization.
Further, the particle swarm algorithm comprises the following steps:
i. initializing the position of each particle;
calculating a fitness value for each particle;
updating the velocity vector and the position vector of each particle in the process of each iteration;
and iv, ending the algorithm when the iteration reaches the maximum number or the change condition of the iteration of a certain number is smaller than the set threshold value.
The invention also provides a power injection type multi-path self-adaptive digital predistortion system, wherein a predistortion module is assembled on each branch to realize the algorithm, the predistortion module comprises a first submodule and a second submodule, the first submodule is an inverse model of the power amplifier of the current branch, the second submodule is an optimal coefficient matrix obtained through a particle swarm optimization algorithm, a baseband input of the current branch outputs a predistortion signal through the predistortion module, and the predistortion signal compensates the nonlinearity and the memory effect of the power amplifier after passing through the power amplifier and is applied to the crosstalk effect generated by the MIMO scene.
Compared with the prior art, the invention has the beneficial effects that:
the power injection type multi-path self-adaptive digital predistortion algorithm and system provided by the invention adopt a direct learning architecture to construct the DPD module, and have the characteristics of simple structure and convenience in implementation. The predistortion signal consists of three parts: the original baseband signal and two low-power baseband signals generated by the two predistortion submodules are amplified by a power amplifier, and observation of the characteristics of output signals can find that out-band spectrum expansion and in-band distortion are well inhibited, so that the problems of nonlinearity, memory effect and crosstalk of the power amplifier in the MIMO system are solved, and the efficiency of the power amplifier in the MIMO system is improved.
Drawings
In order to more clearly illustrate the embodiments of the present application or technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments described in the present invention, and other drawings can be obtained by those skilled in the art according to the drawings.
Fig. 1 is a schematic flow chart of a power injection type multi-path adaptive digital predistortion algorithm and system according to an embodiment of the present invention.
Fig. 2 is a modeling method of a SISO digital power amplifier model according to an embodiment of the present invention.
Fig. 3 is a power injection type multi-path adaptive digital predistortion model according to an embodiment of the present invention.
Fig. 4 is a schematic flow chart of a particle swarm algorithm adopted by the predistortion submodule two according to the embodiment of the present invention;
fig. 5 is a diagram of output spectrum of each branch according to an embodiment of the present invention. Fig. 5(a) is a graph of the output spectrum of branch 1, and fig. 5(b) is a graph of the output spectrum of branch 2.
Fig. 6 is a diagram of an output error vector magnitude table of each branch according to an embodiment of the present invention.
Detailed Description
The invention can improve the nonlinearity, memory effect and crosstalk effect of the power amplifier by constructing the power injection type digital predistortion module in the MIMO system so as to improve the efficiency of the transmitter, and can still obtain good effect when being applied to a broadband communication system.
The invention has excellent linearization effect when being applied to a 5G large-scale MIMO antenna array, a predistortion module is assembled on each branch, the first submodule is an inverse model of the current branch PA, the second submodule obtains an optimal coefficient matrix through a particle swarm optimization algorithm, the baseband input of the current branch outputs a predistortion signal through a DPD module, and the predistortion signal compensates the nonlinearity and memory effect of the PA after passing through the PA and is applied to a crosstalk effect generated by an MIMO scene.
The invention designs a power injection type multi-path self-adaptive digital predistortion algorithm, which comprises the following steps:
s1, respectively constructing a digital PA behavior model under SISO conditions through baseband input and output signals of each branch;
s2, introducing crosstalk at the input end, and obtaining crosstalk output by using a memory polynomial model;
s3, constructing a digital PH behavior model under the condition of multiple input and multiple output by using the input and crosstalk output signals of the power amplifier;
s4, assembling a predistortion model for each branch, wherein the predistortion model comprises a predistortion submodule I and a predistortion submodule II, the submodule I is an inverse model of the power amplifier of the branch under the SISO condition, and the predistortion submodule II is an adaptive predistortion submodule based on a direct learning architecture;
and S5, adding a low-power injection signal output by the first predistortion submodule and a low-power injection signal output by the second predistortion submodule on the basis of baseband input of each branch, and taking the low-power injection signals as input of the branch power amplifier under the condition of parallel Hamming model representation.
Taking 2 × 2MIMO antenna array as an example, the following steps are performed to construct a behavior model of the input/output relationship of each branch and design a corresponding predistortion model, and when branches increase, the corresponding behavior model and model of the predistortion device may be similarly constructed. The power injection type multi-path adaptive digital predistortion algorithm and system are explained by taking 2 × 2MIMO antenna array as an example, as shown in fig. 1, the specific steps are as follows:
step one, passing two paths of baseband signals x(1)、y(1)、x(2)、y(2)Respectively carrying out the digital baseband power amplifier line of the branch 1 and the branch 2 under the condition of Single Input Single Output (SISO)For model construction, it is worth noting that the two paths of baseband input and output signals are subjected to time domain alignment and normalization in advance, and are directly applied to modeling of a behavior model, x(1)=[x(1)(1),x(1)(2),L,x(1)(N)]T, y(1)=[y(1)(1),y(1)(2),L,y(1)(N)]T,x(2),y(2)Form and x(1),y(1)Similarly, N is the number of baseband signals]TRepresenting a matrix transposition.
Consider a Memory Polynomial (MP) model:
where K is the polynomial order, M is the memory depth, am,kFor the model coefficient, | | represents the operation of taking a modulus, the model coefficient is obtained by constructing a linear equation set by using a least square method (LS), and the construction of the linear equation set is shown in fig. 2.
By x(1),y(1)Obtain the power amplification coefficient of branch 1Then, by utilizing a memory polynomial model, the predicted output y of the power amplifier can be obtained(1) FitThe accuracy of the model is measured by Normalized Mean Square Error (NMSE):
where d (N) represents the ideal output signal, s (N) represents the actual measured output signal, and N represents the number of sample points. After the model precision meets the requirement, similar operation obtains the power amplification coefficient of the branch 2And predicted output y of power amplifier(2) Fit。
Step (ii) ofAnd secondly, introducing-20 dB crosstalk by using the memory polynomial models of the branch circuit 1 and the branch circuit 2 obtained in the step one, wherein the baseband input signals through the power amplifier are not only related to the baseband signals passing through the branch circuit, but also influenced by the baseband input signals of other branch circuits. For branch 1, the baseband input signal at this timePower amplifier output y generated by memory polynomial model under condition of existing crosstalk(1) Fit_crosstalkSimilar operation can obtain the power amplifier output y under the condition that the branch 2 has crosstalk(2) Fit_crosstalk。
Step three, utilizing the baseband signal x(1)、x(2)、y(1) Fit_crosstalk、y(2) Fit_crosstalkRespectively constructing a behavior model of the digital baseband power amplifier under the condition of multi-input and multi-output of the branch 1 and the branch 2, and considering a Parallel Hamming (PH) model:
wherein:
i is the i-th branch (1, 2, …, S), x(1)(n-m)、…、x(S)(n-m), baseband signal x for each branch(1)(n)、…、x(S)A delay term of order m of (n).
PH model corresponding to 2 x 2MIMO system, of the form:
i is the i (1, 2) th branch, K is the order, M is the memory depth, b(i) m,k2,k1The model coefficients are obtained by constructing a linear equation set and using a least square method. By x(1)、x(2)、 y(i) Fit_crosstalkObtaining power amplification coefficient of branch iThen, the predictive output y of the power amplifier can be obtained by utilizing the parallel Hamming model(i) crosstalkModel accuracy is measured by Normalized Mean Square Error (NMSE). And executing the step four after the model precision reaches the specified requirement.
And step four, designing a digital predistortion model, wherein the predistortion model comprises two sub-modules, and the design of the first sub-module is firstly carried out as shown in figure 3. In the invention, each branch of the MIMO system is provided with a predistortion device, in the branch 1, a submodule 1 is an inverse model of the branch power amplifier under the condition of SISO, a memory polynomial model is also considered, and after the highest order and the maximum memory depth of the inverse model are determined, in order to obtain a coefficient matrix of the inverse modelWill y(1) FitConversion into corresponding basis function matrix Y(1) FitBuilding a system of linear equationsAnd obtaining a coefficient matrix of the inverse model by using an LS algorithm. It can be seen that the design of the predistortion sub-module 1 in the present invention is similar to the design of the inverse model for digital predistortion of a single PA. However, the input signal of submodule 1 is not already a baseband signal, but x(1)-y(1) crosstalkBecause our power amplifier model is through normalizing the input and output signals x(i)、y(i)The module value of the power amplifier complex gain is approximate to 1, and the inverse model complex gain module value is ideally the reciprocal of the power amplifier complex gain module value and is also a constant close to 1, so that the original baseband signals are normalized in the invention, but represent nonlinearity, memory effect and crosstalkThe components of the effect remain in the data, which is also an important issue that we need to deal with for digital pre-distortion. From the above analysis, we know that x(1)-y(1) crosstalkAs input signal for the predistortion submodule, the modulus of which is relative to the base band signal x(1)Is very small, so the output z of the submodule(1) inject1Also a low power signal, forms part of the predistortion signal. The predistortion submodule of branch 2 is designed in a similar way to branch 1.
Step five, designing a predistortion submodule II, and in the step three, representing x under the condition of-20 dB crosstalk by a PH model(1)、x(2)、y(1) crosstalk、y(2) crosstalkIn branch 1, x(1)、x(2)、y(1) crosstalkOn the basis of the PH model expression, the idea of designing a predistortion submodule II is as follows: using the PH model, the input to Branch 1 is at x(1)On the basis of the low-power injection signal z output by the predistortion submodule 1(1) inject1At this time, the output signal corresponding to the model generates a certain predistortion effect, which indicates that the baseband signal is injected into the low-power signal generated by the predistortion submodule 1 and then passes through the power amplifier, thereby compensating the nonlinearity and memory effect of the power amplifier and the crosstalk effect applied in the MIMO scene to a certain extent. We now need to use the low power injection signal z generated by the predistortion sub-module two(1) inject2To construct a complete digital predistortion device, baseband signal x(1)After passing through the predistortion device, the outputWhen the branch 1 power amplifier is used as input under the PH model representation condition, the in-band distortion and out-of-band distortion of the power amplifier output signal are obviously improved.
The specific implementation steps for constructing the predistortion submodule II are as follows:
firstly, establishing an original low-power injection basis function set U according to a PH model expression(1)0 (N-M)*KN is the total number of baseband data, M is the memory depth, and K is the number of basis functions in the PH model expression. In order to improve the convergence performance of a subsequent algorithm and improve the robustness of the algorithm, an original low-power injection basis function set is subjected to orthogonalization processing, and an orthogonal basis function set U is obtained after SVD decomposition(1)orth (N-M)*K*K is the number of the orthogonal basis functions corresponding to each output signal, and a basis function set U is injected at the original low power(1)0 (N-M)*KIn the case of full rank, K ═ K; otherwise K<K。
Output of submodule twoα(1)Considering the coefficient matrix from the perspective of adaptive filtering, we construct an adaptive predistortion submodule two, and consider the coefficient matrix as a point in a high-dimensional space, and in order to find the optimal coefficient, we first need to let α be the most significant(1)Close to alpha(1) optAn iterative LMS algorithm may be used to accomplish:
wherein i is the iteration number, mu is the iteration step length of the LMS algorithm, e(1)*(i) In order to consider the PH model and under the participation of a low-power injection signal of the predistortion submodule I, the difference value of an output signal and a corresponding baseband input signal is conjugated. Alpha is alpha(1)Initially as a 0 vector, i.e. alpha(1)(0)=[0,0,...,0]TOnly tens of iterations are required.
Step six, making alpha(1)(i +1) is close to α(1) optIn predistortion submodule 2, alpha is obtained(1)And (i +1) finishing all functions of the second submodule by considering an intelligent group optimization algorithm. The present invention uses Particle Swarm Optimization (PSO) to achieve this goal.
i initialize the position of each particle:wherein p is(0)(n) represents the position coordinates of the nth particle at iteration 0; r ═ σ | | | | α(1)(i +1) |, σ is a constant less than 1, | | | | | | represents taking the norm of L2; and () represents generating an and α(1)A random vector of the same size that is,representing the Hadamard product.
ii calculating the fitness value of each particle:
wherein the pH is(1)Representing the parallel Hamming model for the output of branch 1, the form of which has been given in step three, f(0)(n, x) represents the fitness value of the nth particle at the 0 th iteration (initialization), and the position coordinate corresponding to the lowest fitness value of each particle in the ith iteration process is recorded and recorded as pbest(i)(ii) a And the position coordinates corresponding to the particle with the lowest fitness value in the ith iteration of all the particles are recorded as gbest(i). Then:
iii updating the velocity vector and the position vector of each particle in the process of each iteration:
p(i+1)(n)=p(i)(n)+v(i+1)(n)
where N is 1, 2, …, N being the total number of particles, i being the number of iterations, c1、c2Are learning factors and are typically set to 2.
iv maximum number of iterations or gbest(i)Is changed less than a certain number of timesThe algorithm ends at the set threshold, at which time gbest(i)The corresponding position vector can be regarded as alpha(1) opt。
The design process of the predistortion submodule two of the branch 2 is similar to that of the branch 1. Thus, the design of the predistortion device is completed.
The branch 1 of the invention uses an OFDM signal with the bandwidth of 100MHz and the PAPR of 8.2dB as a test signal, the branch 2 uses an OFDM signal with the bandwidth of 100MHz and the PAPR of 9.0dB as a test signal to carry out experiments, each branch is added with a predistortion model after introducing-20 dB crosstalk, the out-band distortion and the in-band distortion of the output signal of the power amplifier are analyzed, and the effect of applying the power injection type multi-path self-adaptive digital predistortion algorithm to a 2 x 2MIMO system is evaluated.
As shown in fig. 5, the out-of-band distortion change of the output signal of the power amplifier before and after the predistortion module is added is shown. By the application of the power injection type digital predistortion algorithm, as shown in fig. 5(a), the output signal ACPR of branch 1 is improved from-31.2378 dBc/-30.4613dBc to-47.3696 dBc/-46.5634dBc, an improvement of about 16 dB. As shown in FIG. 5(b), the output signal ACPR of Branch 2 is improved from-31.5914 dBc/31.1235 dBc to-46.0921 dBc/46.3370 dBc, which is about 15 dB.
Fig. 6 shows the in-band distortion change of the output signal of the power amplifier before and after the predistortion module is added. The output signal EVM of branch 1 is improved from 11.23% to 0.92%. The output signal EVM of branch 2 is improved from 13.37% to 1.07%.
The invention has the advantages that:
1. the small-power signal is injected into the baseband input signal to compensate the nonlinearity, memory effect and crosstalk effect of the power amplifier of the MIMO system through the predistortion submodule I and the predistortion submodule II, and when the PAPR value of the baseband signal is large and the nonlinearity of the power amplifier is strong, a good linearization effect can still be obtained in each branch.
2. The coefficient of the first predistortion submodule can be obtained by off-line calculation, the total operand of DPD can be reduced to a certain extent, and the submodules can be shared by the branches similar to the power amplifier behavior model, so that the structure of the actual circuit is further simplified.
The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: it is to be understood that modifications may be made to the technical solutions described in the foregoing embodiments, or equivalents may be substituted for some of the technical features thereof, but such modifications or substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A power injection type multi-path self-adaptive digital predistortion algorithm is characterized by comprising the following steps:
s1, respectively constructing a digital PA behavior model under SISO conditions through baseband input and output signals of each branch;
s2, introducing crosstalk at the input end, and obtaining crosstalk output by using a memory polynomial model;
s3, constructing a digital PH behavior model under the condition of multiple input and multiple output by using the input and crosstalk output signals of the power amplifier;
s4, assembling a predistortion model for each branch, wherein the predistortion model comprises a predistortion submodule I and a predistortion submodule II, the submodule I is an inverse model of the power amplifier of the branch under the SISO condition, and the predistortion submodule II is an adaptive predistortion submodule based on a direct learning architecture;
and S5, adding a low-power injection signal output by the first predistortion submodule and a low-power injection signal output by the second predistortion submodule on the basis of baseband input of each branch, and taking the low-power injection signals as input of the branch power amplifier under the condition of parallel Hamming model representation.
2. The power injection type multi-path adaptive digital predistortion algorithm of claim 1, wherein in step S1, each path of baseband input and output signals is subjected to time domain alignment and normalization processing in advance.
3. The power injection type multi-path adaptive digital predistortion algorithm of claim 1, wherein the step S1 measures the model accuracy by normalizing the mean square error.
4. The power injection type multi-path adaptive digital predistortion algorithm of claim 1, wherein the step S3 measures the model accuracy by normalizing the mean square error.
5. The power injection type multi-path adaptive digital predistortion algorithm of claim 1, wherein the construction process of the predistortion submodule one in the step S4 is: after the highest order and the maximum memory depth of the inverse model are determined, the prediction output of the power amplifier is converted into a corresponding basis function matrix, a linear equation set is constructed, and a coefficient matrix of the inverse model is obtained by using an LS algorithm.
6. The power injection type multi-path adaptive digital predistortion algorithm of claim 1, wherein the construction process of the predistortion submodule two in the step S4 is as follows: the method comprises the steps of firstly establishing an original low-power injection basis function set according to a PH model expression, carrying out orthogonalization treatment on the original low-power injection basis function set, obtaining the orthogonal basis function set after SVD decomposition, wherein under the condition that the original low-power injection basis function set is full-rank, the total number of orthogonal basis functions corresponding to each output signal is equal to the total number of basis functions corresponding to each output signal, and otherwise, the total number of the orthogonal basis functions corresponding to each output signal is smaller than the total number of basis functions corresponding to each output signal.
7. The power injection type multi-path adaptive digital predistortion algorithm of claim 1, wherein the predistortion submodule in step S4 obtains the optimal coefficient matrix through an intelligent group optimization algorithm.
8. The power injection type multi-path adaptive digital predistortion algorithm of claim 7, wherein the predistortion submodule in step S4 obtains the optimal coefficient matrix through the particle swarm optimization.
9. The power injection type multi-path adaptive digital predistortion algorithm of claim 8, characterized in that the particle swarm algorithm comprises the steps of:
i. initializing the position of each particle;
calculating a fitness value for each particle;
updating the velocity vector and the position vector of each particle in the process of each iteration;
and iv, ending the algorithm when the iteration reaches the maximum number or the change condition of the iteration of a certain number is smaller than the set threshold value.
10. A power injection type multi-path self-adaptive digital predistortion system is characterized in that a predistortion module is assembled on each branch to realize the algorithm of any one of claims 1 to 9, the predistortion module comprises a first submodule and a second submodule, the first submodule is an inverse model of a current branch power amplifier, the second submodule obtains an optimal coefficient matrix through a particle swarm optimization algorithm, a baseband input of the current branch outputs a predistortion signal through the predistortion module, and the predistortion signal compensates nonlinearity and memory effect of the power amplifier after passing through the power amplifier and is applied to a crosstalk effect generated by an MIMO scene.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210166421.6A CN114726703B (en) | 2022-02-23 | 2022-02-23 | Power injection type multipath self-adaptive digital predistortion algorithm and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210166421.6A CN114726703B (en) | 2022-02-23 | 2022-02-23 | Power injection type multipath self-adaptive digital predistortion algorithm and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114726703A true CN114726703A (en) | 2022-07-08 |
CN114726703B CN114726703B (en) | 2023-08-15 |
Family
ID=82236459
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210166421.6A Active CN114726703B (en) | 2022-02-23 | 2022-02-23 | Power injection type multipath self-adaptive digital predistortion algorithm and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114726703B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117729078A (en) * | 2024-02-07 | 2024-03-19 | 厦门大学 | Dual crosstalk cancellation digital predistortion system for MIMO transmitter |
CN117729078B (en) * | 2024-02-07 | 2024-06-04 | 厦门大学 | Dual crosstalk cancellation digital predistortion system for MIMO transmitter |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1746720A1 (en) * | 2005-07-21 | 2007-01-24 | Alcatel | Adaptive Digital Pre-Distortion System |
US20150043678A1 (en) * | 2013-05-09 | 2015-02-12 | King Fahd University Of Petroleum And Minerals | Scalable digital predistortion system |
CN105471783A (en) * | 2015-06-28 | 2016-04-06 | 知鑫知识产权服务(上海)有限公司 | MIMO system transmitting end digital pre-distortion optimization method based on input sequence |
WO2018185532A1 (en) * | 2017-04-07 | 2018-10-11 | Telefonaktiebolaget Lm Ericsson (Publ) | Low complexity mimo digital pre-distortion |
WO2020187384A1 (en) * | 2019-03-15 | 2020-09-24 | Telefonaktiebolaget Lm Ericsson (Publ) | Per-branch, combined, and grouped combined mimo dpd |
CN112803972A (en) * | 2021-02-01 | 2021-05-14 | 西安电子科技大学 | Digital predistortion method and system suitable for MIMO transmitter full loop distortion compensation |
CN113037226A (en) * | 2021-03-01 | 2021-06-25 | 北京邮电大学 | Digital predistortion design method and device based on adaptive step length clipping method |
-
2022
- 2022-02-23 CN CN202210166421.6A patent/CN114726703B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1746720A1 (en) * | 2005-07-21 | 2007-01-24 | Alcatel | Adaptive Digital Pre-Distortion System |
US20150043678A1 (en) * | 2013-05-09 | 2015-02-12 | King Fahd University Of Petroleum And Minerals | Scalable digital predistortion system |
CN105471783A (en) * | 2015-06-28 | 2016-04-06 | 知鑫知识产权服务(上海)有限公司 | MIMO system transmitting end digital pre-distortion optimization method based on input sequence |
WO2018185532A1 (en) * | 2017-04-07 | 2018-10-11 | Telefonaktiebolaget Lm Ericsson (Publ) | Low complexity mimo digital pre-distortion |
WO2020187384A1 (en) * | 2019-03-15 | 2020-09-24 | Telefonaktiebolaget Lm Ericsson (Publ) | Per-branch, combined, and grouped combined mimo dpd |
CN112803972A (en) * | 2021-02-01 | 2021-05-14 | 西安电子科技大学 | Digital predistortion method and system suitable for MIMO transmitter full loop distortion compensation |
CN113037226A (en) * | 2021-03-01 | 2021-06-25 | 北京邮电大学 | Digital predistortion design method and device based on adaptive step length clipping method |
Non-Patent Citations (1)
Title |
---|
程鑫;康凯;赵铖;: "MIMO发射端的非线性串扰校正", 电子设计工程, no. 11 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117729078A (en) * | 2024-02-07 | 2024-03-19 | 厦门大学 | Dual crosstalk cancellation digital predistortion system for MIMO transmitter |
CN117729078B (en) * | 2024-02-07 | 2024-06-04 | 厦门大学 | Dual crosstalk cancellation digital predistortion system for MIMO transmitter |
Also Published As
Publication number | Publication date |
---|---|
CN114726703B (en) | 2023-08-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Fager et al. | Linearity and efficiency in 5G transmitters: New techniques for analyzing efficiency, linearity, and linearization in a 5G active antenna transmitter context | |
US11057081B2 (en) | Multi-cell processing architectures for modeling and impairment compensation in multi-input multi-output systems | |
US8610500B2 (en) | Orthogonal basis function set for digital predistorter | |
US11936343B2 (en) | Per-branch, combined, and grouped combined MIMO DPD | |
CN115278746B (en) | Self-adaptive fast error correction digital predistortion method for 5G broadband power amplifier | |
CN108833318B (en) | Pre-distortion calibration method based on spatial coupling in large-scale MIMO communication system | |
Suryasarman et al. | Digital pre-distortion for multiple antenna transmitters | |
Qian et al. | Calibration of nonlinear crosstalk in MIMO transmitter | |
Wang et al. | Digital predistortion of 5G multiuser MIMO transmitters using low-dimensional feature-based model generation | |
US10498295B2 (en) | Feedback in MISO systems | |
Li et al. | Forward modeling assisted digital predistortion method for hybrid beamforming transmitters with a single PA feedback | |
CN114726703B (en) | Power injection type multipath self-adaptive digital predistortion algorithm and system | |
US20230336215A1 (en) | Multi-cell processing architectures for modeling and impairment compensation in multi-input multi-output systems | |
CN110086438B (en) | Digital predistortion system and method for passive multi-beam transmitter | |
Zhao et al. | A general digital predistortion linearization scheme for hybrid beamforming system | |
Park et al. | A new digital predistortion technique for analog beamforming systems | |
Yu et al. | Dynamic Activation Digital Predistortion of RF Power Amplifiers for 6G Dynamic Spectrum Aggregation Applications | |
Almoneer et al. | Massive MIMO precoding methods that minimize the variation in average power and active impedance with channel conditions | |
Hesami et al. | Digital predistorter in crosstalk compensation of MIMO transmitters | |
Khan et al. | Statistical linearization of phased arrays using power adaptive power amplifier model | |
López-Bueno et al. | Training data selection and dimensionality reduction for polynomial and artificial neural network MIMO adaptive digital predistortion | |
Gulo et al. | Implementation of Tiny Machine Learning (TinyML) as Pre-distorter for High Power Amplifier (HPA) Linearization of SDR-based MIMO-OFDM | |
Fernandez et al. | Linearization of Active Array Transmitters Under Crosstalk via Over-the-Air Observations | |
Campo et al. | Closed-loop sign algorithms for low-complexity digital predistortion | |
Sheikhi et al. | Machine Learning Based Digital Pre-Distortion in Massive MIMO Systems: Complexity-Performance Trade-offs |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |