Channel-robust online modulation identification method and system
By employing a channel-robust online modulation identification method to perform blind synchronization and channel blind equalization on the received signal, the problem of low modulation type identification accuracy under multipath fading channels is solved, and high-precision online modulation identification is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA ELECTRONICS TECH GRP NO 7 RES INST
- Filing Date
- 2026-02-12
- Publication Date
- 2026-05-12
AI Technical Summary
Existing technologies have low accuracy in identifying the modulation type of wireless communication signals in multipath fading channel environments. In particular, when the modulation type of the signal is unknown, it is difficult to effectively utilize the statistical characteristics of the signal to perform blind equalization to eliminate the influence of multipath channels.
A channel-robust online modulation identification method is adopted. The original received signal is preprocessed with modulation-independent blind synchronization. The parameters are updated by training with a channel blind equalization module and a quantization identification module to obtain high-precision modulation type estimation. The method includes blind synchronization, channel blind equalization and quantization identification steps.
High-precision online identification of wireless communication signal modulation type was achieved in multipath fading channel environment. It can directly learn and track channel changes on test sample data, thus improving the identification accuracy.
Smart Images
Figure CN122027412A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of wireless communication technology, and more specifically, to a channel-robust online modulation identification method and system. Background Technology
[0002] In radio fields such as cognitive radio, spectrum sensing, and electronic reconnaissance, it is necessary to quickly and adaptively identify the modulation type of wireless communication signals to support applications such as cognitive communication and detection decoding. Currently, modulation identification is mainly achieved through likelihood-based and feature-based methods. For likelihood-based algorithms, the probability density function of the received signal is typically used to evaluate the likelihood of each possible modulation pattern hypothesis, and the identification result is obtained by maximizing the likelihood function. Feature-based algorithms, on the other hand, extract signal features based on domain expert knowledge or deep networks, and classify and identify them according to the differences in signal features between different modulation types.
[0003] However, when applied to real-world multipath fading channel environments, the above methods suffer severe degradation in recognition accuracy due to unwanted channel effects in the signal (such as inter-symbol interference, amplitude fading, frequency offset, and phase offset). Existing methods based on deep learning are mainly designed for modulation recognition tasks under simplified channels (such as ideal channels or additive white Gaussian noise channels). Due to the influence of multipath fading channels, these methods have low recognition accuracy in real-world open environments. Furthermore, when the signal modulation type is unknown, it is often difficult to effectively utilize the statistical characteristics of the signal (such as cyclostationarity, constant mode characteristics, constellation diagrams, etc.) to perform blind equalization in advance to eliminate the influence of multipath channels.
[0004] Furthermore, modulation recognition methods based on deep transfer learning can adapt to the impact of uncertainties in multipath fading channels in real-world scenarios to some extent, but their recognition accuracy is somewhat inferior to that of current deep learning modulation recognition methods trained offline on complete datasets. Summary of the Invention
[0005] To address the problem of low accuracy in identifying modulation types of wireless communication signals in existing technologies, this invention proposes a channel-robust online modulation identification method and system to achieve high-precision, real-time modulation identification of wireless communication signal modulation types in multipath fading channel environments.
[0006] To achieve the above-mentioned technical effects, the technical solution of the present invention is as follows:
[0007] This application proposes a channel-robust online modulation identification method, comprising the following steps: S1: Perform modulation-independent blind synchronization preprocessing on the original received signal to obtain the blind synchronized received signal; S2: Based on the received signal after blind synchronization and the constructed channel blind equalizer, the preset channel blind equalization module is used to obtain the transmitted signal estimate with unknown modulation type for blind equalization estimation. S3: Based on the transmitted signal estimation with unknown modulation type and the standard constellation set of all possible modulation types, with the goal of minimizing the sum of the constellation distance statistical error and the constellation probability distribution entropy under different modulation type assumptions, the modulation type estimation and quantization estimation of the transmitted signal are obtained by using a preset quantization identification module. At the same time, the quantization identification module and the channel blind equalization module are trained and the parameters of the quantization identification module and the channel blind equalization module are updated. S4: Determine whether the parameters of the quantization identification module and the channel blind equalization module meet the iteration update termination condition. If yes, output the modulation type estimate and quantization estimate of the transmitted signal; otherwise, return to step S2.
[0008] Preferably, in S1, the original received signal undergoes modulation-independent blind synchronization preprocessing to obtain the blind-synchronized received signal. The process is as follows: S11: The carrier frequency offset of the original received signal is estimated using a modulation-independent blind frequency offset estimation algorithm, and frequency offset compensation is performed. S12: Estimate the symbol rate of the signal processed in step S11 using a modulation-independent blind symbol rate estimation algorithm, and perform symbol sampling synchronization; S13: Use a matched filter to perform matched filtering on the signal processed in step S12 to obtain the matched filtered signal; S14: Based on the matched-filtered signal, the sampling deviation and sampling frequency difference are corrected by the modulation-independent blind symbol timing synchronization algorithm to obtain the timing-synchronized signal; S15: Based on the signal after timing synchronization, the carrier phase rotation deviation is corrected by using a modulation-independent blind carrier phase estimation method, and the carrier phase is locked to obtain the received signal after blind synchronization.
[0009] Preferably, in S2, based on the received signal after blind synchronization and the constructed channel blind equalizer, a preset channel blind equalization module is used to obtain an estimate of the transmitted signal with an unknown modulation type for blind equalization estimation. The process is as follows: Obtain the residual vector of the channel blind equalizer tap vector, the expression of which is:
[0010] in, For channel blind equalizer Initial estimation; Based on the received signal and residual vector after blind synchronization, an estimate of the transmitted signal with an unknown modulation type is obtained. Satisfies the expression:
[0011] in, This represents the received signal after blind synchronization. It has The transmitted signal vector of each independent and identically distributed modulation symbol. This represents the transpose of a vector. Taken from modulation type Standard constellation set Modulation type The set of all possible modulation types ; It is the equivalent channel vector. It is the total number of taps in the equivalent channel vector; This is the noise vector; For fixed phase bias; These are the first weighting factor and the second weighting factor, respectively. This represents the convolution operation. Represents the complex field.
[0012] Preferably, the constructed channel blind equalizer have A tap vector, expressed as: ,and .
[0013] Preferably, in S3, based on the transmitted signal estimation with unknown modulation type and the standard constellation set of all possible modulation types, with the goal of minimizing the sum of the constellation distance statistical error and the constellation probability distribution entropy under different modulation type assumptions, the modulation type estimate and quantization estimate of the transmitted signal are obtained using a preset quantization identification module. Simultaneously, the quantization identification module and the channel blind equalization module are trained, and the parameters of the quantization identification module and the channel blind equalization module are updated. The process is as follows: S31: Estimation based on the transmitted signal with the unknown modulation type and the standard constellation set of all possible modulation types By minimizing the statistical error of constellation distances Entropy of constellation probability distribution The sum of these values yields a quantization estimate of the transmitted signal. Modulation type estimation Satisfies the expression:
[0014] in, To utilize the channel blind equalization module for Blind equalization is performed to obtain the transmitted signal. The estimated posterior distribution, This represents calculating information entropy; Represents the standard constellation set, Indicates modulation type The corresponding standard constellation set, Modulation type The number of elements in the standard constellation set. express The symbol points Equal to modulation type The constellation points after phase correction The probability of; S32: Construct the loss function, based on the received signal after blind synchronization and the transmitted signal with unknown modulation type for estimation. Quantization estimation of transmitted signals Calculate the loss function; S33: Update the learning rate according to the loss function, and iteratively update the parameters of the quantization recognition module and the channel blind equalization module based on the learning rate.
[0015] Preferably, the expression for the loss function is:
[0016] in, The weighting factor of the loss function. To send a signal about The true posterior distribution, To utilize the channel blind equalization module for Blind equalization is performed to obtain the transmitted signal. The estimated posterior distribution, express and The KL divergence.
[0017] Preferably, and KL divergence The expression is:
[0018] in, Let be the expectation of the variational approximation.
[0019] This application proposes an electronic device, comprising: a memory, a processor, and a program stored in the memory and executable on the processor; characterized in that the processor is configured to read the program in the memory to implement the steps in the channel robust online modulation identification method.
[0020] This application proposes a computer storage medium for storing a program, characterized in that the program, when executed by a processor, implements the steps in the channel-robust online modulation identification method.
[0021] This application also proposes a channel-robust online modulation identification system for implementing the method, comprising: The preprocessing unit is used to perform modulation-independent blind synchronization preprocessing on the original received signal to obtain the blind synchronized received signal. The channel blind equalization unit is used to obtain the transmitted signal estimate with unknown modulation type based on the received signal after blind synchronization and the constructed channel blind equalizer, using a preset channel blind equalization module. The quantization identification and module training unit is used to estimate the modulation type of the transmitted signal based on the unknown modulation type and the standard constellation set of all possible modulation types. With the goal of minimizing the sum of the constellation distance statistical error and the constellation probability distribution entropy under different modulation type assumptions, it uses a preset quantization identification module to obtain the modulation type estimate and quantization estimate of the transmitted signal. At the same time, it trains the quantization identification module and the channel blind equalization module and updates the parameters of the quantization identification module and the channel blind equalization module. The loop unit is used to determine whether the parameters of the quantization identification module and the channel blind equalization module meet the iteration update termination condition. If so, it outputs the modulation type estimate and quantization estimate of the transmitted signal; otherwise, it returns to the channel blind equalization unit.
[0022] Compared with the prior art, the beneficial effects of the technical solution of the present invention are: This invention proposes a channel-robust online modulation identification method and system. The method involves performing modulation-independent blind synchronization preprocessing on the original received signal to obtain a blind-synchronized received signal. Based on the blind-synchronized received signal and a pre-constructed channel blind equalizer, a channel blind equalization module is used to obtain an estimate of the transmitted signal with an unknown modulation type. Based on the unknown modulation type transmitted signal estimate and a standard constellation set of all possible modulation types, with the objective of minimizing the sum of the constellation distance statistical error and the constellation probability distribution entropy under different modulation type assumptions, a quantization identification module is used to obtain the modulation type estimate and quantization estimate of the transmitted signal. Simultaneously, the quantization identification module and the channel blind equalization module are trained and their parameters are updated until the iteration update termination condition is met. The proposed method operates without the need for pre-trained deep models, enabling direct learning and tracking of channel changes on test sample data. This allows for high-precision online identification of the modulation type of wireless communication signals in multipath fading channel environments. Attached Figure Description
[0023] Figure 1 A flowchart illustrating the channel-robust online modulation identification method proposed in this embodiment of the invention; Figure 2 This is a comparison chart showing the average recognition accuracy proposed in the embodiments of the present invention; Figure 3 This represents the modulation identification and confusion matrix diagram proposed in the embodiments of the present invention; Figure 4 This diagram illustrates the structure of the electronic device proposed in the embodiments of the present invention. Figure 5 This diagram illustrates the composition of the channel-robust online modulation identification system proposed in this embodiment of the invention. Detailed Implementation
[0024] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this patent. To better illustrate this embodiment, some parts of the accompanying drawings may be omitted, enlarged, or reduced, and do not represent the actual dimensions; It is understandable to those skilled in the art that some well-known details may be omitted from the accompanying drawings.
[0025] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments; The positional relationships depicted in the accompanying drawings are for illustrative purposes only and should not be construed as limiting this patent.
[0026] Example 1 This embodiment provides a channel-robust online modulation identification method, the flowchart of which can be found in [link to flowchart]. Figure 1 This includes the following steps: S1: Perform modulation-independent blind synchronization preprocessing on the original received signal to obtain the blind synchronized received signal; S2: Based on the received signal after blind synchronization and the constructed channel blind equalizer, the preset channel blind equalization module is used to obtain the transmitted signal estimate with unknown modulation type for blind equalization estimation. S3: Based on the transmitted signal estimation with unknown modulation type and the standard constellation set of all possible modulation types, with the goal of minimizing the sum of the constellation distance statistical error and the constellation probability distribution entropy under different modulation type assumptions, the modulation type estimation and quantization estimation of the transmitted signal are obtained by using a preset quantization identification module. At the same time, the quantization identification module and the channel blind equalization module are trained and the parameters of the quantization identification module and the channel blind equalization module are updated. S4: Determine whether the parameters of the quantization identification module and the channel blind equalization module meet the iteration update termination condition. If yes, output the modulation type estimate and quantization estimate of the transmitted signal; otherwise, return to step S2.
[0027] In this embodiment, a blindly synchronized received signal is obtained by performing modulation-independent blind synchronization preprocessing on the original received signal. Based on the blindly synchronized received signal and the constructed channel blind equalizer, the channel blind equalization module obtains an estimate of the transmitted signal with an unknown modulation type. Based on the transmitted signal estimate with an unknown modulation type and a set of standard constellations for all possible modulation types, with the goal of minimizing the sum of the constellation distance statistical error and the constellation probability distribution entropy under different modulation type assumptions, the quantization identification module obtains the modulation type estimate and quantization estimate of the transmitted signal. Simultaneously, the quantization identification module and the channel blind equalization module are trained and their parameters are updated until the parameters of the quantization identification module and the channel blind equalization module meet the iteration update termination condition. The method proposed in this invention operates without the need for pre-trained deep models and can directly learn and track channel changes on test sample data, thereby achieving high-precision online identification of robust modulation types in multipath fading channel environments.
[0028] Example 2 In this embodiment, in S1, the original received signal undergoes modulation-independent blind synchronization preprocessing to obtain the blind-synchronized received signal. The process is as follows: S11: The carrier frequency offset of the original received signal is estimated using a modulation-independent blind frequency offset estimation algorithm, and frequency offset compensation is performed. S12: Estimate the symbol rate of the signal processed in step S11 using a modulation-independent blind symbol rate estimation algorithm, and perform symbol sampling synchronization; S13: Use a matched filter to perform matched filtering on the signal processed in step S12 to obtain the matched filtered signal; S14: Based on the matched-filtered signal, the sampling deviation and sampling frequency difference are corrected by the modulation-independent blind symbol timing synchronization algorithm to obtain the timing-synchronized signal; S15: Based on the signal after timing synchronization, the carrier phase rotation deviation is corrected by using a modulation-independent blind carrier phase estimation method, and the carrier phase is locked to obtain the received signal after blind synchronization.
[0029] Specifically, in S11, the modulation-independent blind frequency offset estimation algorithm employs blind frequency offset estimation based on joint high-order cyclic cumulants; in S12, the modulation-independent blind symbol rate estimation algorithm can employ spectral peak detection based on signal envelope or symbol rate blind estimation based on cyclic spectrum, and the oversampling rate of the processed signal is a positive integer; in S13, the matched filter employs a root-raised cosine filter with a roll-off factor of 0.5 and a filter length span of 6 to improve the signal-to-noise ratio of the received signal output; in S14, the modulation-independent blind symbol timing synchronization algorithm employs Gardner-based blind symbol timing synchronization; in S15, the modulation-independent blind carrier phase estimation method employs carrier phase offset correction based on nonlinear least mean square estimation to correct carrier phase rotation deviation, lock the carrier phase of the received signal, and finally obtain the blind-synchronized received signal. .
[0030] In this embodiment, under the condition that the modulation type and channel state are unknown, the modulation type is estimated while mitigating the impact of inter-symbol interference. And transmitted signal estimation To achieve the goal, a joint blind equalization and modulation identification problem model is constructed, satisfying the expression:
[0031] in, The non-convex cost function depends on the specific algorithm; transmitted signal estimation , representing the estimated value. A channel blind equalization module and a quantization identification module are constructed, and the joint blind equalization and modulation identification problem model is solved based on online learning.
[0032] Preferably, in S2, based on the received signal after blind synchronization and the constructed channel blind equalizer, a preset channel blind equalization module is used to obtain an estimate of the transmitted signal with an unknown modulation type for blind equalization estimation. The process is as follows: Obtain the residual vector of the channel blind equalizer tap vector, the expression of which is:
[0033] in, For channel blind equalizer Initial estimation; Based on the received signal and residual vector after blind synchronization, an estimate of the transmitted signal with an unknown modulation type is obtained. Satisfies the expression:
[0034] in, This represents the received signal after blind synchronization. It has The transmitted signal vector of each independent and identically distributed modulation symbol. This represents the transpose of a vector. Taken from modulation type Standard constellation set Modulation type The set of all possible modulation types ; It is the equivalent channel vector. It is the total number of taps in the equivalent channel vector; This is the noise vector; For fixed phase bias; These are the first weighting factor and the second weighting factor, respectively. This represents the convolution operation. Represents the complex field.
[0035] Specifically, when sending signal vector When it is BPSK modulation, it is denoted as ,at this time The equivalent channel vector is the equivalent impulse response of the transmitter shaping filter, multipath channel, and receiver matched filter, etc., while the noise vector... With a mean of 0 and a variance of Additive white Gaussian noise, phase bias It is introduced during the signal preprocessing stage and is related to the signal modulation type. Related fixed phase offset. Different modulation types are shown in Table 1. Phase offset of the signal after blind synchronization through signal preprocessing surface.
[0036] Table 1
[0037] In practice, phase deviation The results may differ from those in Table 1 depending on the specific blind symbol timing synchronization algorithm and blind carrier phase locking algorithm used.
[0038] Preferably, the constructed channel blind equalizer have A tap vector, expressed as: ,and .
[0039] Preferably, in S3, based on the transmitted signal estimation with unknown modulation type and the standard constellation set of all possible modulation types, with the goal of minimizing the sum of the constellation distance statistical error and the constellation probability distribution entropy under different modulation type assumptions, the modulation type estimate and quantization estimate of the transmitted signal are obtained using a preset quantization identification module. Simultaneously, the quantization identification module and the channel blind equalization module are trained, and the parameters of the quantization identification module and the channel blind equalization module are updated. The process is as follows: S31: Estimation based on the transmitted signal with the unknown modulation type and the standard constellation set of all possible modulation types By minimizing the statistical error of constellation distances Entropy of constellation probability distribution The sum of these values yields a quantization estimate of the transmitted signal. Modulation type estimation Satisfies the expression:
[0040] in, To utilize the channel blind equalization module for Blind equalization is performed to obtain the transmitted signal. The estimated posterior distribution, This represents calculating information entropy; Represents the standard constellation set, Indicates modulation type The corresponding standard constellation set, Modulation type The number of elements in the standard constellation set. express The symbol points Equal to modulation type The constellation points after phase correction The probability of; S32: Construct the loss function, based on the received signal after blind synchronization and the transmitted signal with unknown modulation type for estimation. Quantization estimation of transmitted signals Calculate the loss function; S33: Update the learning rate according to the loss function, and iteratively update the parameters of the quantization recognition module and the channel blind equalization module based on the learning rate.
[0041] Specifically, in S31, probability It can be calculated using the following formula:
[0042] In S33, the learning rate is updated according to the loss function, and the process is as follows: like Then update the learning rate. ;in, Adjusting the threshold for the learning rate, superscript on the variable This indicates the number of online learning iterations; learning updates are based on the learning rate. .
[0043] In step S4, it is determined whether the parameters of the quantization identification module and the channel blind equalization module meet the iteration update termination condition. If so, the modulation type estimate and quantization estimate of the transmitted signal are output; otherwise, the process returns to step S2. like Then output the quantized estimate of the transmitted signal obtained in this iteration round. Modulation type estimation Otherwise, return to step S2; where, The termination threshold is used for iterative optimization.
[0044] Preferably, the expression for the loss function is:
[0045] in, The weighting factor of the loss function. To send a signal about The true posterior distribution, To utilize the channel blind equalization module for Blind equalization is performed to obtain the transmitted signal. The estimated posterior distribution, express and The KL divergence.
[0046] Preferably, and KL divergence The expression is:
[0047] in, Let be the expectation of the variational approximation.
[0048] Specifically, the expression for calculating the first term on the right side of the equation is: ; The expression for calculating the second term on the right side of the equation is:
[0049] in, , ;and , .
[0050] In this embodiment, the quantization identification module and the channel blind equalization module are trained based on online learning, and the modulation type estimate and quantization estimate of the transmitted signal are obtained. The process is as follows: C1: Initialize the learning rate Learning rate adjustment threshold Iterative optimization termination threshold Initial estimation of blind equalizer tap vectors Equivalent channel vector The residual vector of the blind equalizer tap vector First weighting factor Second weighting factor Loss function weighting factor .
[0051] C2: Based on the received signal after blind synchronization and the constructed channel blind equalizer, the preset channel blind equalization module is used to obtain the transmitted signal estimate with unknown modulation type for blind equalization estimation. .
[0052] C3: Estimation of the transmitted signal based on the unknown modulation type and the standard constellation set of all possible modulation types With the goal of minimizing the sum of the constellation distance statistical error and the constellation probability distribution entropy under different modulation type assumptions, the modulation type estimate of the transmitted signal is obtained using a pre-set quantization identification module. and quantitative estimation .
[0053] C4: Construct the loss function, based on the received signal after blind synchronization and the transmitted signal with unknown modulation type for estimation. Quantization estimation of transmitted signals Calculate the loss function .
[0054] C5: If Then update the learning rate. .
[0055] C6: If Then output the quantized estimate of the transmitted signal obtained in this iteration round. Modulation type estimation Otherwise, return to step C2.
[0056] C7: Learning and updating based on learning rate. .
[0057] Among them, variable superscript This indicates the number of iterations in the online learning process.
[0058] Example 3 In this embodiment, by simulating real information sources and typical multipath Rayleigh fading channel scenarios, the modulation recognition accuracy performance of the proposed method in this invention is compared with that of deep learning methods such as DA-CNN, CLDNN, and MCNet.
[0059] Specifically, ideal signals with different modulations were generated by simulating real information sources. The sample parameters of the modulation signals are shown in Table 2. Each modulation type has 50,000 samples, of which 40,000 are training samples used for training models based on offline deep learning methods such as DA-CNN, CLDNN, and MCNet; and 10,000 are test samples used to test the performance of each method.
[0060] The simulation simulates a multipath Rayleigh fading channel scenario, which adds distortion / error to the ideal signal due to different channel fading factors in actual communication, including different SNR, time delay, frequency offset and phase offset, frequency / phase jitter, IQ quadrature modulation error, and multipath channel fading, etc. The specific channel influencing factors and parameter values are shown in Table 3.
[0061] Table 2
[0062] Table 3
[0063] In simulating real-world signal sources and multipath Rayleigh channel fading scenarios, the method proposed in this invention is compared with the average recognition accuracy of DA-CNN, CLDNN, and MCNet under multipath Rayleigh channel fading conditions with different SNRs. Figure 2 As shown.
[0064] Depend on Figure 2 It is known that typical supervised deep learning algorithms (CLDNN, MCNet) cannot adapt to signal distortion / errors in simulated real multipath Rayleigh channel environments, resulting in low recognition accuracy. The DA-CNN method based on domain adaptive transfer learning shows a significant improvement in recognition accuracy compared to typical deep learning methods, indicating that it can adapt to the signal distortion caused by channel fading to a certain extent. Furthermore, the method proposed in this invention operates without the need for pre-trained deep models and consistently maintains the optimal average recognition accuracy. This is because this method not only achieves unsupervised online recognition of modulation types but also benefits from blind signal synchronization preprocessing and loss function constraint design. Through the quantization recognition module, the channel blind equalization module adaptively matches the constellation characteristics of the modulated signal, "approaching the blind equalization variational approximation" while "approaching the constellation points as uniformly as possible," effectively reducing the degradation of modulation recognition performance caused by various channel effects such as ISI.
[0065] Figure 3 Under multipath Rayleigh channel fading conditions with SNR=4dB, the method proposed in this invention was used to identify the confusion matrix of test samples of 12 different modulated signals, achieving an average identification accuracy of 97.9%. Figure 3 The proposed method demonstrates excellent performance in a multipath Rayleigh channel fading scenario with an SNR of 4dB, exhibiting high discriminative power in identifying 12 modulation signals. The accuracy rate for most signal categories approaches or reaches 100%, with only slight confusion between a few signals. For example, there are very few misclassifications between 16PSK and 16QAM, and between 16QAM and 64QAM. Analysis suggests this is due to the similarity in signal constellation structures and feature overlap caused by channel fading, but the misclassification rate is consistently below 3%. Overall, this method effectively resists multipath fading and low SNR interference, demonstrating strong channel robustness and providing a highly reliable solution for modulation identification in complex channel environments.
[0066] Example 4 This application provides an electronic device, including a processor, a memory, and a program stored in the memory and executable on the processor. When the program is executed by the processor, it implements the various processes of the above-described channel robust online modulation identification method embodiment and achieves the same technical effect. To avoid repetition, it will not be described again here.
[0067] For details, see Figure 4 This application also provides an electronic device, including a bus 401, a transceiver 402, an antenna 403, a bus interface 404, a processor 405, and a memory 406.
[0068] The transceiver 402 is used to acquire at least one of the standard constellation set of all possible modulation types and the original received signal; The processor 405 is used to perform modulation-independent blind synchronization preprocessing on the original received signal to obtain a blind-synchronized received signal; based on the blind-synchronized received signal and the constructed channel blind equalizer, a preset channel blind equalization module is used to obtain an estimate of the transmitted signal with an unknown modulation type; based on the estimate of the transmitted signal with an unknown modulation type and a set of standard constellations for all possible modulation types, with the goal of minimizing the sum of the statistical error of constellation distance and the entropy of constellation probability distribution under different modulation type assumptions, a preset quantization identification module is used to obtain the modulation type estimate and quantization estimate of the transmitted signal; simultaneously, the quantization identification module and the channel blind equalization module are trained using an online learning method, and the parameters of the quantization identification module and the channel blind equalization module are updated; until the parameters of the quantization identification module and the channel blind equalization module meet the iteration update termination condition.
[0069] exist Figure 4In this context, a bus architecture (represented by bus 401) is used. Bus 401 can include any number of interconnected buses and bridges, linking various circuits including one or more processors represented by processor 405 and memory represented by memory 406. Bus 401 can also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. Bus interface 404 provides an interface between bus 401 and transceiver 402. Transceiver 402 can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by processor 405 is transmitted over a wireless medium via antenna 403, which further receives data and transmits data to processor 405.
[0070] Processor 405 is responsible for managing bus 401 and general processing, and can also provide various functions, including timing, peripheral interface, voltage regulation, power management, and other control functions. Memory 406 can be used to store data used by processor 405 during operation.
[0071] Optionally, the processor 405 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a complex programmable logic device (CPLD).
[0072] This application also provides a computer-readable storage medium storing a computer program. When executed by a processor, this computer program implements the various processes of the above-described channel-robust online modulation identification method embodiments and achieves the same technical effects. To avoid repetition, it will not be described again here. The computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.
[0073] This embodiment also provides a channel-robust online modulation identification system, see [link to documentation]. Figure 5 This system is used to implement a channel-robust online modulation identification method, including: The preprocessing unit is used to perform modulation-independent blind synchronization preprocessing on the original received signal to obtain the blind synchronized received signal. The channel blind equalization unit is used to obtain the transmitted signal estimate with unknown modulation type based on the received signal after blind synchronization and the constructed channel blind equalizer, using a preset channel blind equalization module. The quantization identification and module training unit is used to estimate the modulation type of the transmitted signal based on the unknown modulation type and the standard constellation set of all possible modulation types. With the goal of minimizing the sum of the constellation distance statistical error and the constellation probability distribution entropy under different modulation type assumptions, it uses a preset quantization identification module to obtain the modulation type estimate and quantization estimate of the transmitted signal. At the same time, it trains the quantization identification module and the channel blind equalization module and updates the parameters of the quantization identification module and the channel blind equalization module. The loop unit is used to determine whether the parameters of the quantization identification module and the channel blind equalization module meet the iteration update termination condition. If so, it outputs the modulation type estimate and quantization estimate of the transmitted signal; otherwise, it returns to the channel blind equalization unit.
[0074] The same or similar labels correspond to the same or similar parts; The terms used to describe positional relationships in the accompanying drawings are for illustrative purposes only and should not be construed as limiting the invention. Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.
Claims
1. A channel-robust online modulation identification method, characterized in that, Includes the following steps: S1: Perform modulation-independent blind synchronization preprocessing on the original received signal to obtain the blind synchronized received signal; S2: Based on the received signal after blind synchronization and the constructed channel blind equalizer, the preset channel blind equalization module is used to obtain the transmitted signal estimate with unknown modulation type for blind equalization estimation. S3: Based on the transmitted signal estimation with unknown modulation type and the standard constellation set of all possible modulation types, with the goal of minimizing the sum of the constellation distance statistical error and the constellation probability distribution entropy under different modulation type assumptions, the modulation type estimation and quantization estimation of the transmitted signal are obtained by using a preset quantization identification module. At the same time, the quantization identification module and the channel blind equalization module are trained and the parameters of the quantization identification module and the channel blind equalization module are updated. S4: Determine whether the parameters of the quantization identification module and the channel blind equalization module meet the iteration update termination condition. If yes, output the modulation type estimate and quantization estimate of the transmitted signal; otherwise, return to step S2.
2. The channel-robust online modulation identification method according to claim 1, characterized in that, In S1, the original received signal undergoes modulation-independent blind synchronization preprocessing to obtain the blind-synchronized received signal. The process is as follows: S11: The carrier frequency offset of the original received signal is estimated using a modulation-independent blind frequency offset estimation algorithm, and frequency offset compensation is performed. S12: Estimate the symbol rate of the signal processed in step S11 using a modulation-independent blind symbol rate estimation algorithm, and perform symbol sampling synchronization; S13: Use a matched filter to perform matched filtering on the signal processed in step S12 to obtain the matched filtered signal; S14: Based on the matched-filtered signal, the sampling deviation and sampling frequency difference are corrected by the modulation-independent blind symbol timing synchronization algorithm to obtain the timing-synchronized signal; S15: Based on the signal after timing synchronization, the carrier phase rotation deviation is corrected by using a modulation-independent blind carrier phase estimation method, and the carrier phase is locked to obtain the received signal after blind synchronization.
3. The channel-robust online modulation identification method according to claim 1, characterized in that, In S2, based on the received signal after blind synchronization and the constructed channel blind equalizer, the estimated transmitted signal with an unknown modulation type is obtained using a preset channel blind equalization module. The process is as follows: Obtain the residual vector of the channel blind equalizer tap vector, the expression of which is: in, For channel blind equalizer Initial estimation; Based on the received signal and residual vector after blind synchronization, an estimate of the transmitted signal with an unknown modulation type is obtained. Satisfies the expression: in, This represents the received signal after blind synchronization. It has The transmitted signal vector of each independent and identically distributed modulation symbol. This represents the transpose of a vector. Taken from modulation type Standard constellation set Modulation type The set of all possible modulation types ; It is the equivalent channel vector. It is the total number of taps in the equivalent channel vector; This is the noise vector; For fixed phase bias; These are the first weighting factor and the second weighting factor, respectively. This represents the convolution operation. Represents the complex field.
4. The channel-robust online modulation identification method according to claim 3, characterized in that, The constructed channel blind equalizer have A tap vector, expressed as: ,and .
5. The channel-robust online modulation identification method according to claim 3, characterized in that, In S3, based on the transmitted signal estimation with unknown modulation type and the standard constellation set of all possible modulation types, with the objective of minimizing the sum of the constellation distance statistical error and the constellation probability distribution entropy under different modulation type assumptions, the modulation type estimate and quantization estimate of the transmitted signal are obtained using a preset quantization identification module. Simultaneously, the quantization identification module and the channel blind equalization module are trained and their parameters are updated. The process is as follows: S31: Estimation based on the transmitted signal with the unknown modulation type and the standard constellation set of all possible modulation types By minimizing the statistical error of constellation distances Entropy of constellation probability distribution The sum of these values yields a quantization estimate of the transmitted signal. Modulation type estimation Satisfies the expression: in, To utilize the channel blind equalization module for Blind equalization is performed to obtain the transmitted signal. The estimated posterior distribution, This represents calculating information entropy; Indicates modulation type The corresponding standard constellation set, Modulation type The number of elements in the standard constellation set. express The symbol points Equal to modulation type The constellation points after phase correction The probability of; S32: Construct the loss function, based on the received signal after blind synchronization and the transmitted signal with unknown modulation type for estimation. Quantization estimation of transmitted signals Calculate the loss function; S33: Update the learning rate according to the loss function, and iteratively update the parameters of the quantization recognition module and the channel blind equalization module based on the learning rate.
6. The channel-robust online modulation identification method according to claim 5, characterized in that, The expression for the loss function is: in, The weighting factor of the loss function. To send a signal about The true posterior distribution, To utilize the channel blind equalization module for Blind equalization is performed to obtain the transmitted signal. The estimated posterior distribution, express and The KL divergence.
7. The channel-robust online modulation identification method according to claim 6, characterized in that, and KL divergence The expression is: in, Let be the expectation of the variational approximation.
8. An electronic device, comprising: A memory, a processor, and a program stored in the memory and executable on the processor; characterized in that the processor is configured to read the program in the memory to implement the steps of the channel robust online modulation identification method as described in any one of claims 1 to 7.
9. A computer storage medium for storing programs, characterized in that, When the program is executed by the processor, it implements the steps of the channel robust online modulation identification method as described in any one of claims 1 to 7.
10. A channel-robust online modulation identification system, used to implement the channel-robust online modulation identification method according to any one of claims 1-7, characterized in that, include: The preprocessing unit is used to perform modulation-independent blind synchronization preprocessing on the original received signal to obtain the blind synchronized received signal. The channel blind equalization unit is used to obtain the transmitted signal estimate with unknown modulation type based on the received signal after blind synchronization and the constructed channel blind equalizer, using a preset channel blind equalization module. The quantization identification and module training unit is used to estimate the modulation type of the transmitted signal based on the unknown modulation type and the standard constellation set of all possible modulation types. With the goal of minimizing the sum of the constellation distance statistical error and the constellation probability distribution entropy under different modulation type assumptions, it uses a preset quantization identification module to obtain the modulation type estimate and quantization estimate of the transmitted signal. At the same time, it trains the quantization identification module and the channel blind equalization module and updates the parameters of the quantization identification module and the channel blind equalization module. The loop unit is used to determine whether the parameters of the quantization identification module and the channel blind equalization module meet the iteration update termination condition. If so, it outputs the modulation type estimate and quantization estimate of the transmitted signal; otherwise, it returns to the channel blind equalization unit.