A method, system and device for identifying a PS-QAM signal modulation format

Abstract

The application provides a PS-QAM signal modulation format identification method, system and device, and relates to the field of communication.The method comprises the following steps: processing historical PS-QAM signals of a communication system under different modulation formats, shaping entropies and signal-to-noise ratios to generate historical PS-QAM identification signals; calculating historical scaling factors and historical probability medians of the historical PS-QAM identification signals; training a support vector machine according to the historical scaling factors and the historical probability medians; processing actual PS-QAM signals to be identified to generate actual PS-QAM identification signals; calculating actual scaling factors and actual probability medians of the actual PS-QAM identification signals; and identifying the modulation format and the shaping entropy of the actual PS-QAM signals according to the trained support vector machine based on the actual scaling factors and the actual probability medians.The application can accurately identify the modulation format and finely identify the shaping entropy of the signals with low complexity, thereby improving the flexibility of the communication system.

Description

Technical Field

[0001] This invention relates to the field of communications, and in particular to a method, system, and device for identifying PS-QAM signal modulation formats. Background Technology

[0002] Due to the rapid growth of global IP traffic and the demand for various data services such as the Internet of Things, big data, cloud computing, and streaming video, communication networks need to adaptively adjust bandwidth, modulation format, and transmission rate according to communication link conditions and user traffic demands. This is to improve user service quality at a lower cost while increasing spectrum utilization. To adapt to dynamic communication needs, Probabilistic Shaping-Quadrature Amplitude Modulation (PS-QAM) technology and Modulation Format Identification (MFI) algorithms have begun to attract attention. Since the digital signal processing algorithms at the receiver end of a communication system require the modulation format and shaping parameters of the received signal as prior information to correctly reconstruct the transmission sequence, and the flexibility and real-time requirements of the communication network limit the complexity of the receiver, low-complexity identification algorithms capable of handling different shaping entropies and modulation formats are essential.

[0003] Currently, researchers have proposed various modulation format identification schemes for PS-QAM signals. Among them, non-data-aided schemes identify modulation formats by extracting features such as the cumulative amplitude distribution, signal power distribution, and peak-to-average power ratio (PAPR) of the input signal. However, with the development of error correction coding schemes, the signal-to-noise ratio (SNR) requirements of communication systems are gradually decreasing, and the performance of non-data-aided schemes at low SNR conditions does not meet practical needs. Deep neural networks and convolutional neural networks, leveraging machine learning algorithms, can achieve high MFI performance; however, the training cost of these schemes increases exponentially with the increase in shaping entropy granularity, resulting in high complexity and requiring significant computational resources. Furthermore, some schemes propose shaping signal identification based on frequency offset loading technology; however, the method of adding pilots in this scheme increases system complexity, making deployment in practical communication systems difficult and consuming spectrum resources to some extent.

[0004] In summary, current PS-QAM signal modulation format recognition schemes suffer from significant performance conflicts with system complexity and cannot achieve fine-grained shaping entropy recognition. Therefore, there is an urgent need for a low-complexity, high-performance PS-QAM signal modulation format recognition scheme that is compatible with fine-grained shaping. Summary of the Invention

[0005] The purpose of this invention is to provide a PS-QAM signal modulation format identification method, system, and device that can accurately identify the modulation format and, with low complexity and fine granularity, identify the signal shaping entropy, thereby improving the flexibility of the communication system.

[0006] To achieve the above objectives, the present invention provides the following solution:

[0007] A method for identifying PS-QAM signal modulation format includes:

[0008] The historical PS-QAM signals of the communication system are processed under different modulation formats, shaping entropy and signal-to-noise ratios to generate historical PS-QAM identification signals;

[0009] Calculate the historical scaling factor and the historical probability median of the historical PS-QAM identification signal;

[0010] Train a support vector machine based on the historical scaling factor and the historical probability median;

[0011] The actual PS-QAM signal to be identified is processed to generate the actual PS-QAM identification signal;

[0012] Calculate the actual scaling factor and the actual probability median of the actual PS-QAM recognition signal;

[0013] Based on the actual scaling factor and the actual probability median, the modulation format and shaping entropy of the actual PS-QAM signal are identified using the trained support vector machine.

[0014] Optionally, the historical scaling factor and historical probability median of the historical PS-QAM identification signal are calculated, specifically including:

[0015] The historical PS-QAM identification signal is normalized to generate a normalized historical PS-QAM identification signal;

[0016] Based on amplitude clustering, the amplitude of the normalized historical PS-QAM identification signal is truncated to determine the amplitude range of the historical PS-QAM identification signal while retaining the constellation point signal with the smallest amplitude.

[0017] Based on the reduced historical amplitude range, the amplitude distribution of the historical PS-QAM identification signal after amplitude truncation is statistically analyzed, the equally spaced amplitude segments and the number of data points of the historical PS-QAM identification signal corresponding to the amplitude segments are determined, and a data point sequence is generated.

[0018] The data sequence is subjected to differential differentiation and Gaussian smoothing to determine the derivative sequence representing the rate of change of the probability distribution of the truncated signal;

[0019] The historical scaling factor is determined based on the derivative sequence;

[0020] For the historical PS-QAM identification signal, artificial Gaussian white noise is introduced to generate a historical PS-QAM identification signal with continuous amplitude.

[0021] The historical probability median is determined based on the historical PS-QAM identification signal with continuous amplitude.

[0022] Optionally, the historical scaling factor is determined based on the derivative sequence, specifically including:

[0023] Determine whether the first minimum point of the derivative sequence is close to zero;

[0024] If so, the center of the amplitude segment corresponding to the minimum point is determined as the amplitude of the minimum amplitude constellation point signal in the derivative sequence;

[0025] If not, the center of the amplitude segment corresponding to the first zero point in the reciprocal sequence is the amplitude of the smallest amplitude constellation point signal in the derivative sequence;

[0026] The historical scaling factor is determined based on the amplitude of the minimum amplitude constellation point signal.

[0027] Optionally, the historical scaling factor is determined based on the amplitude of the minimum amplitude constellation point signal, specifically including:

[0028] Using formula Determine the historical scaling factor; where F is the historical scaling factor, and a min This represents the amplitude of the minimum amplitude constellation point signal.

[0029] Optionally, determining the historical probability median based on the historical PS-QAM identification signal with continuous amplitude specifically includes:

[0030] Using the formula M=meadian(a noise ) / max(a noise Determine the historical probability median; where M is the historical probability median, a noise For a historical PS-QAM identification signal with continuous amplitude, meadian(·) is used to find the median, and max(·) is used to find the maximum value.

[0031] Optionally, training a support vector machine based on the historical scaling factor and the historical probability median specifically includes:

[0032] Based on the historical scaling factor and the historical probability median, establish a two-dimensional feature distribution under different modulation formats and integer entropy;

[0033] The modulation format and integer entropy are used as labels, and the two-dimensional feature distribution is used as data to train a support vector machine; the support vector machine is a soft-margin support vector machine based on Gaussian radial basis functions.

[0034] Optionally, the actual scaling factor and the actual probability median of the actual PS-QAM recognition signal are calculated, specifically including:

[0035] The actual PS-QAM recognition signal is normalized to generate a normalized actual PS-QAM recognition signal;

[0036] Based on amplitude clustering, the amplitude of the normalized actual PS-QAM identification signal is truncated to determine the amplitude range of the actual PS-QAM identification signal while retaining the constellation point signal with the smallest amplitude.

[0037] Based on the reduced historical amplitude range, the amplitude distribution of the actual PS-QAM identification signal after amplitude truncation is statistically analyzed, the equally spaced amplitude segments and the number of data points of the actual PS-QAM identification signal corresponding to the amplitude segments are determined, and a data point sequence is generated.

[0038] The data sequence is subjected to differential differentiation and Gaussian smoothing to determine the derivative sequence representing the rate of change of the probability distribution of the truncated signal;

[0039] The actual scaling factor is determined based on the derivative sequence;

[0040] For the actual PS-QAM recognition signal, artificial Gaussian white noise is introduced to generate an actual PS-QAM recognition signal with continuous amplitude.

[0041] The historical probability median is determined based on the actual PS-QAM identification signal with continuous amplitude.

[0042] The historical PS-QAM identification signal generation module is used to process the historical PS-QAM signals of the communication system under different modulation formats, shaping entropy and signal-to-noise ratios to generate historical PS-QAM identification signals.

[0043] The historical scaling factor and historical probability median calculation module is used to calculate the historical scaling factor and historical probability median of the historical PS-QAM identification signal.

[0044] The training module is used to train a support vector machine based on the historical scaling factor and the historical probability median.

[0045] The actual PS-QAM recognition signal generation module is used to process the actual PS-QAM signal to be recognized and generate the actual PS-QAM recognition signal.

[0046] The actual scaling factor and actual probability median calculation module is used to calculate the actual scaling factor and actual probability median of the actual PS-QAM recognition signal.

[0047] The identification module is used to identify the modulation format and shaping entropy of the actual PS-QAM signal based on the actual scaling factor and the actual probability median, according to the trained support vector machine.

[0048] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor runs the computer program to enable the electronic device to perform the PS-QAM signal modulation format recognition method described above.

[0049] Optionally, the memory is a non-transitory computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described PS-QAM signal modulation format recognition method.

[0050] According to specific embodiments provided by the present invention, the following technical effects are disclosed: Under different modulation formats, shaping entropy, and signal-to-noise ratios, the historical PS-QAM signals of a communication system are processed to generate historical PS-QAM identification signals. The historical scaling factor and historical probability median of the historical PS-QAM identification signals are calculated to train a support vector machine (SVM). Based on the actual scaling factor and actual probability median corresponding to the actual PS-QAM signal to be identified, the trained SVM is used to identify the modulation format and shaping entropy of the actual PS-QAM identification signal. The present invention can achieve high-precision fine-grained identification by calculating the scaling factor and probability median, two features with significant discriminative power and insensitivity to noise. The present invention itself is computationally simple, and based on the SVM, it has low training costs, avoiding the large datasets and training costs of deep learning and other solutions, and exhibiting low complexity compared to neural networks. Through low-complexity, high-precision identification of the modulation format and fine-grained shaping entropy of PS-QAM signals, the flexibility of the communication system can be significantly improved. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 The flowchart of the PS-QAM signal modulation format recognition method provided by the present invention is shown below;

[0053] Figure 2 This is a schematic diagram of the receiving end of the communication system provided by the present invention;

[0054] Figure 3 The following is a flowchart of the scaling factor calculation provided by the present invention;

[0055] Figure 4 The distribution diagram of scaling factor and probability median under different modulation formats, shaping entropy, and signal-to-noise ratios provided by this invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] The purpose of this invention is to provide a PS-QAM signal modulation format identification method, system, and device that can realize the PS-QAM signal modulation format and fine-grained shaping entropy with low complexity and high accuracy, thereby significantly improving the flexibility of the communication system.

[0058] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0059] Example 1

[0060] like Figure 1 As shown, the present invention provides a PS-QAM signal modulation format identification method, comprising:

[0061] Step 101: Process the historical PS-QAM signal of the communication system under different modulation formats, shaping entropy and signal-to-noise ratios to generate historical PS-QAM identification signals.

[0062] Step 102: Calculate the historical scaling factor and the historical probability median of the historical PS-QAM identification signal.

[0063] In practical applications, step 102 specifically includes: normalizing the historical PS-QAM identification signal to generate a normalized historical PS-QAM identification signal; truncating the amplitude of the normalized historical PS-QAM identification signal based on amplitude clustering to determine the amplitude range of the historical PS-QAM identification signal while retaining the signal with the smallest amplitude constellation point; statistically analyzing the amplitude distribution of the amplitude-truncated historical PS-QAM identification signal based on the reduced historical amplitude range, determining equally spaced amplitude segments and the number of data points of the historical PS-QAM identification signal corresponding to the amplitude segments, and generating a data point sequence; performing differential differentiation and Gaussian smoothing on the data point sequence to determine a derivative sequence representing the rate of change of the probability distribution of the truncated signal; determining a historical scaling factor based on the derivative sequence; introducing artificial Gaussian white noise into the historical PS-QAM identification signal to generate a historical PS-QAM identification signal with continuous amplitude; and determining the historical probability median based on the historical PS-QAM identification signal with continuous amplitude.

[0064] In practical applications, when performing amplitude clustering, the number of centers is equal to the number N of the lowest order amplitude values ​​of the identified signal. A The initial cluster centers are set to C. int The normalized PS-QAM recognition signal set is then subjected to K-means clustering based on the difference between the absolute value of the amplitude and the cluster center, where C int This is a set of amplitude values. After clustering, the amplitude of the smallest centroid is used as a reference to truncate the signal to be identified, retaining only signals whose absolute amplitude is a few times smaller than the amplitude of the smallest centroid, with a reference multiple of 2.

[0065] In practical applications, determining the historical scaling factor based on the derivative sequence specifically includes: determining whether the first minimum point of the derivative sequence is close to zero; if so, determining the center of the amplitude segment corresponding to the minimum point as the amplitude of the minimum amplitude constellation point signal in the derivative sequence; if not, determining the center of the amplitude segment corresponding to the first zero point in the reciprocal sequence as the amplitude of the minimum amplitude constellation point signal in the derivative sequence; and determining the historical scaling factor based on the amplitude of the minimum amplitude constellation point signal.

[0066] In practical applications, the historical scaling factor is determined based on the amplitude of the minimum amplitude constellation point signal, specifically including: using the formula Determine the historical scaling factor; where F is the historical scaling factor, and a min This represents the amplitude of the minimum amplitude constellation point signal.

[0067] In practical applications, determining the historical probability median based on the historical PS-QAM identification signal with continuous amplitude specifically includes: using the formula M = median(anoise ) / max(a noise Determine the historical probability median; where M is the historical probability median, a noise For a historical PS-QAM identification signal with continuous amplitude, meadian(·) is used to find the median, and max(·) is used to find the maximum value.

[0068] Step 103: Train a support vector machine based on the historical scaling factor and the historical probability median.

[0069] In practical applications, step 103 specifically includes: establishing two-dimensional feature distributions under different modulation formats and integer entropy based on the historical scaling factor and the historical probability median; using the modulation format and integer entropy as labels, and using the two-dimensional feature distributions as data to train a support vector machine; the support vector machine is a soft-margin support vector machine based on Gaussian radial basis functions.

[0070] In practical applications, the modulation format and integer entropy are used as labels, and the scaling factor and median probability are used as data, for a quantity of N. m ×N ν A soft-margin support vector machine based on Gaussian radial basis functions (N-1) is trained to obtain hyperplane parameters for classifying modulation formats and integer entropy using scaling factors and median probabilities. Where N-1... m For modulation format types, N ν This refers to the types of integer entropy under each modulation format.

[0071] Step 104: Process the actual PS-QAM signal to be identified to generate the actual PS-QAM identification signal.

[0072] In practical applications, both the historical PS-QAM signal and the actual PS-QAM signal to be identified undergo digital signal processing with transparent modulation format.

[0073] Step 105: Calculate the actual scaling factor and the actual probability median of the actual PS-QAM recognition signal.

[0074] In practical applications, step 105 specifically includes: normalizing the actual PS-QAM recognition signal to generate a normalized actual PS-QAM recognition signal; truncating the amplitude of the normalized actual PS-QAM recognition signal based on amplitude clustering to determine the amplitude range of the actual PS-QAM recognition signal while retaining the signal with the smallest amplitude constellation point; statistically analyzing the amplitude distribution of the amplitude-truncated actual PS-QAM recognition signal based on the reduced historical amplitude range, determining equally spaced amplitude segments and the number of data points of the actual PS-QAM recognition signal corresponding to the amplitude segments, and generating a data point sequence; performing differential differentiation and Gaussian smoothing on the data point sequence to determine a derivative sequence representing the rate of change of the probability distribution of the truncated signal; determining the actual scaling factor based on the derivative sequence; introducing artificial Gaussian white noise into the actual PS-QAM recognition signal to generate an amplitude-continuous actual PS-QAM recognition signal; and determining the historical probability median based on the amplitude-continuous actual PS-QAM recognition signal.

[0075] In practical applications, the calculation process for the actual scaling factor and the median probability of the actual PS-QAM recognition signal is the same as that for the historical scaling factor and the median probability of the historical PS-QAM recognition signal.

[0076] Step 106: Based on the actual scaling factor and the actual probability median, identify the modulation format and shaping entropy of the actual PS-QAM signal according to the trained support vector machine.

[0077] In practical applications, the reference value for artificial Gaussian white noise is 10 dB.

[0078] Example 2

[0079] The specific implementation of this invention will be described using the identification of modulation format and shaped entropy in a communication system with an entropy value interval of 0.5 bits / symbol and an entropy value range of 3-4 bits / symbol, a PS-32 PS-QAM signal with an entropy value range of 4-5 bits / symbol, and a PS-64 PS-QAM signal with an entropy value range of 5-6 bits / symbol as an example.

[0080] Step 1: Referring to the signal-to-noise ratio (SNR) limits of different modulation formats under the 7% forward error correction limit, for 20 groups of shaped entropy H with an entropy value interval of 0.5 bits / symbol, the entropy value range of shaped entropy H is 3-4 bits / symbol, and the SNR range is 18-30dB for PS-16PS-QAM signals; the entropy value range of shaped entropy H is 4-5 bits / symbol, and the SNR range is 22-30dB for PS-32PS-QAM signals; and the entropy value range of shaped entropy H is 5-6 bits / symbol, and the SNR range is 24-30dB for PS-64PS-QAM signals, in the following... Figure 2 The receiving end of the communication system shown obtains the signal to be identified after digital signal processing with transparent modulation format. The scaling factor and probability median of the signal to be identified are calculated. Based on the above two features, a two-dimensional feature distribution with different modulation formats and shaped entropy is established, and a support vector machine is trained. Figure 2 It includes a modulation format transparent digital signal processing module, a probability-shaping orthogonal amplitude modulation signal modulation format recognition module, and a modulation format related digital signal processing module.

[0081] During the training process of the support vector machine, the signal to be identified is the historical PS-QAM recognition signal; during the recognition process using the support vector machine, the signal to be identified is the actual PS-QAM recognition signal.

[0082] Step 1.1: According to... Figure 3 The scaling factor calculation method shown calculates the scaling factor of the signal to be identified, and firstly, the signal to be identified is normalized.

[0083] Subsequently, amplitude-based clustering was performed on the normalized signal to truncate its amplitude, narrowing the amplitude range of the identified signal while retaining the signal with the smallest amplitude constellation point. For amplitude clustering, three centers were used, with initial center values ​​of 0.2143, 0.4286, and 0.6429. K-means clustering was performed on the signal amplitude based on the difference between the absolute amplitude value and the cluster center. After clustering, the signal to be identified was truncated based on the amplitude of the smallest center point, retaining only signals whose absolute amplitude value was less than twice the amplitude of the smallest center point.

[0084] By statistically analyzing the amplitude distribution of the truncated signal, we can obtain equally spaced amplitude segments and the number of data points corresponding to each amplitude segment.

[0085] By performing differential differentiation and Gaussian smoothing on the data sequence, a smooth derivative sequence representing the rate of change of the probability distribution of the truncated signal is obtained.

[0086] For a derivative sequence, first determine whether the first minimum point of the sequence is close to zero. If it is close to zero, the center of the amplitude segment corresponding to the minimum point is the amplitude of the minimum amplitude constellation point of the sequence to be identified. If there is no first near-zero minimum point, the center of the amplitude segment corresponding to the first zero point of the derivative sequence is the amplitude of the minimum amplitude constellation point of the sequence to be identified.

[0087] The amplitude a of the obtained minimum amplitude constellation point min The following processing is performed to obtain the specific scaling factor F:

[0088] Step 1.2: For the signal to be processed, introduce artificial white Gaussian noise with a signal-to-noise ratio of 10dB to obtain a signal a with continuous amplitude. noise And calculate the median probability M according to the following formula: M = meadian(a noise ) / max(a noise ); where meadian(·) means to find the median, and max(·) means to find the maximum value.

[0089] Step 1.3: Use the modulation format and integer entropy as tags, such as... Figure 4 The scaling factor and median probability shown are used as data to train an 8-scale soft-margin support vector machine based on Gaussian radial basis functions, resulting in hyperplane parameters used to classify modulation formats and integer entropy using the scaling factor and median probability.

[0090] Step 2: When transmitting actual signals in the communication system, the receiving end obtains the signal to be identified after digital signal processing with transparent modulation format. The scaling factor and probability median of the signal to be identified are calculated. Based on the support vector machine trained in Step 1, the modulation format and shaping entropy of the signal are identified.

[0091] Step 2.1: According to... Figure 3 The scaling factor calculation method shown calculates the scaling factor of the signal to be identified, and firstly, the signal to be identified is normalized.

[0092] Amplitude-based clustering is used to truncate the amplitude of the normalized signal, narrowing the amplitude range of the identified signal while retaining signals with the smallest amplitude constellation points. During amplitude clustering, the number of centers is three, equal to the number of lowest-order amplitude values ​​in the identified signal. Initial centers are set to 0.2143, 0.4286, and 0.6429. K-means clustering is performed on the signal amplitude based on the difference between the absolute amplitude value and the cluster center. After clustering, the signal to be identified is truncated based on the amplitude of the smallest center point, retaining only signals whose absolute amplitude value is less than twice the amplitude of the smallest center point.

[0093] By statistically analyzing the amplitude distribution of the truncated signal, we can obtain equally spaced amplitude segments and the number of data points corresponding to each amplitude segment.

[0094] By performing differential differentiation and Gaussian smoothing on the data sequence, a smooth derivative sequence representing the rate of change of the probability distribution of the truncated signal is obtained.

[0095] For a derivative sequence, first determine whether the first minimum point of the sequence is close to zero. If it is close to zero, the center of the amplitude segment corresponding to the minimum point is the amplitude of the minimum amplitude constellation point of the sequence to be identified. If there is no first near-zero minimum point, the center of the amplitude segment corresponding to the first zero point of the derivative sequence is the amplitude of the minimum amplitude constellation point of the sequence to be identified.

[0096] The amplitude a of the obtained minimum amplitude constellation point min The following processing is performed to obtain the specific scaling factor F:

[0097] Step 2.2: For the signal to be processed, introduce artificial white Gaussian noise with a signal-to-noise ratio of 10dB to obtain a signal a with continuous amplitude. noise And calculate the median probability M according to the following formula: M = meadian(a noise ) / max(a noise ); where meadian(·) means to find the median, and max(·) means to find the maximum value.

[0098] Step 2.3: Classify the signal to be identified based on the signal modulation format, shaping entropy, and the trained support vector machine.

[0099] Example 3

[0100] In order to implement the method corresponding to Embodiment 1 above and achieve the corresponding functions and technical effects, a PS-QAM signal modulation format recognition system is provided below.

[0101] The historical PS-QAM identification signal generation module is used to process the historical PS-QAM signals of the communication system under different modulation formats, shaping entropy and signal-to-noise ratios to generate historical PS-QAM identification signals.

[0102] The historical scaling factor and historical probability median calculation module is used to calculate the historical scaling factor and historical probability median of the historical PS-QAM identification signal.

[0103] The training module is used to train a support vector machine based on the historical scaling factor and the historical probability median.

[0104] The actual PS-QAM recognition signal generation module is used to process the actual PS-QAM signal to be recognized and generate the actual PS-QAM recognition signal.

[0105] The module for calculating the actual scaling factor and the actual probability median is used to calculate the actual scaling factor and the actual probability median of the actual PS-QAM recognition signal.

[0106] The identification module is used to identify the modulation format and shaping entropy of the actual PS-QAM signal based on the actual scaling factor and the actual probability median, according to the trained support vector machine.

[0107] Example 4

[0108] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor runs the computer program to enable the electronic device to perform the PS-QAM signal modulation format recognition method described above.

[0109] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the PS-QAM signal modulation format recognition method described above.

[0110] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0111] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for identifying PS-QAM signal modulation format, characterized in that, include: The historical PS-QAM signals of the communication system are processed under different modulation formats, shaping entropy and signal-to-noise ratios to generate historical PS-QAM identification signals; Calculate the historical scaling factor and the historical probability median of the historical PS-QAM identification signal; Train a support vector machine based on the historical scaling factor and the historical probability median; The actual PS-QAM signal to be identified is processed to generate the actual PS-QAM identification signal; Calculate the actual scaling factor and the actual probability median of the actual PS-QAM recognition signal; Based on the actual scaling factor and the actual probability median, the modulation format and shaping entropy of the actual PS-QAM signal are identified using the trained support vector machine.

2. The PS-QAM signal modulation format recognition method according to claim 1, characterized in that, Calculating the historical scaling factor and historical probability median of the historical PS-QAM identification signal specifically includes: The historical PS-QAM identification signal is normalized to generate a normalized historical PS-QAM identification signal; Based on amplitude clustering, the amplitude of the normalized historical PS-QAM identification signal is truncated to determine the amplitude range of the historical PS-QAM identification signal while retaining the constellation point signal with the smallest amplitude. Based on the reduced historical amplitude range, the amplitude distribution of the historical PS-QAM identification signal after amplitude truncation is statistically analyzed, the equally spaced amplitude segments and the number of data points of the historical PS-QAM identification signal corresponding to the amplitude segments are determined, and a data point sequence is generated. The data sequence is subjected to differential differentiation and Gaussian smoothing to determine the derivative sequence representing the rate of change of the probability distribution of the truncated signal; The historical scaling factor is determined based on the derivative sequence; For the historical PS-QAM identification signal, artificial Gaussian white noise is introduced to generate a historical PS-QAM identification signal with continuous amplitude. The historical probability median is determined based on the historical PS-QAM identification signal with continuous amplitude.

3. The PS-QAM signal modulation format recognition method according to claim 2, characterized in that, Determining the historical scaling factor based on the derivative sequence specifically includes: Determine whether the first minimum point of the derivative sequence is close to zero; If so, determine the center of the amplitude segment corresponding to the minimum point as the amplitude of the minimum amplitude constellation point signal in the derivative sequence; If not, the center of the amplitude segment corresponding to the first zero in the derivative sequence is the amplitude of the smallest amplitude constellation point signal in the derivative sequence; The historical scaling factor is determined based on the amplitude of the minimum amplitude constellation point signal.

4. The PS-QAM signal modulation format recognition method according to claim 3, characterized in that, The historical scaling factor is determined based on the amplitude of the minimum amplitude constellation point signal, specifically including: Using formula Determine the historical scaling factor; where, F As a historical scaling factor, This represents the amplitude of the minimum amplitude constellation point signal.

5. The PS-QAM signal modulation format recognition method according to claim 2, characterized in that, Determining the historical probability median based on the historical PS-QAM identification signal with continuous amplitude specifically includes: Using formula Determine the median of historical probabilities; where, M This is the historical probability median. For historical PS-QAM identification signals with continuous amplitude, To find the median, To find the maximum value.

6. The PS-QAM signal modulation format recognition method according to claim 1, characterized in that, Training a support vector machine based on the historical scaling factor and the historical median probability specifically includes: Based on the historical scaling factor and the historical probability median, establish a two-dimensional feature distribution under different modulation formats and integer entropy; The modulation format and integer entropy are used as labels, and the two-dimensional feature distribution is used as data to train a support vector machine; the support vector machine is a soft-margin support vector machine based on Gaussian radial basis functions.

7. The PS-QAM signal modulation format recognition method according to claim 1, characterized in that, The calculation of the actual scaling factor and the actual probability median of the actual PS-QAM recognition signal specifically includes: The actual PS-QAM recognition signal is normalized to generate a normalized actual PS-QAM recognition signal; Based on amplitude clustering, the amplitude of the normalized actual PS-QAM identification signal is truncated to determine the amplitude range of the actual PS-QAM identification signal while retaining the constellation point signal with the smallest amplitude. Based on the reduced actual amplitude range, the amplitude distribution of the actual PS-QAM recognition signal after amplitude truncation is statistically analyzed, the equally spaced amplitude segments and the number of data points of the actual PS-QAM recognition signal corresponding to the amplitude segments are determined, and a data point sequence is generated. The data sequence is subjected to differential differentiation and Gaussian smoothing to determine the derivative sequence representing the rate of change of the probability distribution of the truncated signal; The actual scaling factor is determined based on the derivative sequence; For the actual PS-QAM recognition signal, artificial Gaussian white noise is introduced to generate an actual PS-QAM recognition signal with continuous amplitude. The actual probability median is determined based on the actual PS-QAM recognition signal with continuous amplitude.

8. A PS-QAM signal modulation format recognition system, characterized in that, include: The historical PS-QAM identification signal generation module is used to process the historical PS-QAM signals of the communication system under different modulation formats, shaping entropy and signal-to-noise ratios to generate historical PS-QAM identification signals. The historical scaling factor and historical probability median calculation module is used to calculate the historical scaling factor and historical probability median of the historical PS-QAM identification signal. The training module is used to train a support vector machine based on the historical scaling factor and the historical probability median. The actual PS-QAM recognition signal generation module is used to process the actual PS-QAM signal to be recognized and generate the actual PS-QAM recognition signal. The actual scaling factor and actual probability median calculation module is used to calculate the actual scaling factor and actual probability median of the actual PS-QAM recognition signal. The identification module is used to identify the modulation format and shaping entropy of the actual PS-QAM signal based on the actual scaling factor and the actual probability median, according to the trained support vector machine.

9. An electronic device, characterized in that, The device includes a memory and a processor, the memory being used to store a computer program, and the processor running the computer program to cause the electronic device to perform the PS-QAM signal modulation format recognition method as described in any one of claims 1-7.

10. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium is a memory, and the non-transitory computer-readable storage medium stores a computer program, which, when executed by a processor, implements the PS-QAM signal modulation format recognition method as described in any one of claims 1-7.

Images