A method for adaptive identification of signal modulation parameters

Abstract

The present invention discloses a signal modulation parameter adaptive identification method, including an automatic adjustment analysis bandwidth process and a signal modulation parameter identification process based on a node-integrated classifier. Through the automatic adjustment analysis bandwidth process, the present invention can obtain signal characteristics under different analysis bandwidth settings without manually setting the analysis bandwidth, thereby providing a data basis for subsequent adaptive identification modules. Through the signal modulation parameter identification process based on a node-integrated classifier, the present invention can extract high-value signals from complex signal characteristics, exclude redundant information and interference information, and realize adaptive and stable identification of signal modulation parameters. The present method does not require manual setting of the analysis bandwidth, and can realize adaptive identification of signals of different bandwidths, thereby reducing the operating burden of spectrum monitoring personnel and greatly improving signal analysis efficiency. The present method has good adaptability to signals of different bandwidths, and has good identification accuracy and stability.

Description

Technical Field

[0001] The invention relates to the field of signal modulation parameter recognition of spectrum monitoring instruments, and in particular to a signal modulation parameter adaptive recognition method. Background Art

[0002] Signal modulation parameter identification is an important function in spectrum monitoring instruments and is the basis for subsequent signal demodulation and analysis. However, signal modulation types are complex and difficult to identify. The signal modulation parameter identification function often becomes a shortcoming and bottleneck in the analysis capability of signal monitoring equipment, which has a great impact on the overall performance of the entire signal monitoring instrument. The modulation parameter identification method of existing instruments requires the operator to manually set the analysis bandwidth and does not have bandwidth adaptive capabilities. When the actual bandwidth of the measured signal does not match the manually set analysis bandwidth, the recognition accuracy will be significantly reduced.

[0003] The disadvantages of the prior art are:

[0004] (1) It cannot automatically adapt to the bandwidth of the measured signal, and the operator needs to manually set the analysis bandwidth, which increases the operator's operating burden and reduces the efficiency of spectrum monitoring.

[0005] (2) When the set analysis bandwidth is inconsistent with the actual signal bandwidth, the recognition accuracy will be seriously reduced. Summary of the invention

[0006] In view of the problem that the existing signal modulation parameter identification method cannot automatically adapt to the bandwidth of the measured signal, the present invention proposes a signal modulation parameter adaptive identification method which does not require manual setting of the analysis bandwidth and has good identification accuracy and stability.

[0007] The present invention adopts the following technical solutions:

[0008] A method for adaptively identifying signal modulation parameters comprises the following steps:

[0009] Step 1: Automatically adjust the analysis bandwidth: After the user selects the location of the signal to be identified, the frequency at which the maximum value of the power spectrum mean in the local range is located is selected as the signal center frequency f c , select the signal center frequency f c The 3dB bandwidth is taken as the initial analysis bandwidth B1, and the analysis bandwidth is gradually increased with k*B1 as the step size to form an analysis bandwidth set {B1, B2, ..., B N}, where k is the step size coefficient;

[0010] Step 2: Construct a training sample set: Construct a training sample set for signals with different modulation modes; collect N different samples for each signal according to the different bandwidths in the analysis bandwidth set, denoted as s (c,i,b), all samples constitute a training sample set, where c represents the modulation parameter type of the sample, i represents the i-th signal under the c-th modulation parameter type, and b represents the sample collected by the i-th signal sample under the c-th modulation parameter type according to the b-th analysis bandwidth;

[0011] Step 3: Node classifier training: The training sample set is divided into N training subsets according to the analysis bandwidth, and N modulation parameter recognition classifiers, i.e., node classifiers, are trained respectively;

[0012] Step 4: Integrated classifier training: After the node classifier training is completed, the recognition result vector is obtained, and the recognition result vector is used to form an integrated training sample set, and the integrated training sample set is used to train the integrated classifier;

[0013] Step 5: Combine the trained node classifier and ensemble classifier into a whole, that is, the results of the node classifier are concentrated and input into the ensemble classifier, and the ensemble classifier outputs the final recognition result.

[0014] The present invention has the following beneficial effects:

[0015] The process of automatically adjusting the analysis bandwidth, through which the present invention can obtain the signal characteristics under different analysis bandwidth settings without manually setting the analysis bandwidth, and provide a data basis for the subsequent adaptive recognition module. The process of signal modulation parameter recognition based on the node-ensemble classifier, through which the present invention can extract high-value signals from complex signal characteristics, exclude redundant information and interference information, and realize adaptive and stable recognition of signal modulation parameters.

[0016] This method does not require manual setting of the analysis bandwidth, and can realize adaptive recognition of signals of different bandwidths, thereby reducing the operational burden of spectrum monitoring personnel and greatly improving the efficiency of signal analysis. This method has good adaptability to signals of different bandwidths and has good recognition accuracy and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 Schematic diagram of automatic adjustment of analysis bandwidth.

[0018] Figure 2 It is a schematic diagram of the process of the present invention.

[0019] Figure 3 Schematic diagram of node classifier and ensemble classifier training.

[0020] Figure 4 It is a schematic diagram of the overall flow of the modulation parameter adaptive identification of the present invention. DETAILED DESCRIPTION

[0021] The specific implementation of the present invention is further described below with reference to the accompanying drawings and specific examples:

[0022] Combination Figures 1 to 4 , a signal modulation parameter adaptive identification method, comprising the following steps:

[0023] Step 1: Automatically adjust the analysis bandwidth: After the user selects the location of the signal to be identified, the frequency at which the maximum value of the power spectrum mean in the local range is located is selected as the signal center frequency f c , select the signal center frequency f c The 3dB bandwidth is taken as the initial analysis bandwidth B1, and the analysis bandwidth is gradually increased with k*B1 as the step size to form an analysis bandwidth set {B1, B2, ..., B N}, where k is the step size coefficient. Figure 1 shown.

[0024] Step 2: Construct a training sample set: For L modulation modes, collect M signals for each modulation mode, and each signal is analyzed according to the bandwidth set {B1, B2, ..., B N} Collect N signal samples, then form a total training sample set of L*M*N samples. Each signal sample is denoted as s (c,i,b) , c=1,2,...,L,i=1,2,...,M,b=1,2,...,N,where c represents the modulation parameter type of the sample, i represents the ith signal under the cth modulation parameter type, and b represents the sample collected by the ith signal under the cth modulation parameter type according to the bth analysis bandwidth. Figure 2 shown.

[0025] Step 3: Node classifier training: Divide the training sample set into N training subsets according to the analysis bandwidth, and train N modulation parameter recognition classifiers, namely, node classifiers. Figure 3 shown.

[0026] The purpose of using a node classifier is to obtain the best modulation parameter classification capability under a specific analysis bandwidth. The node classifier can be freely selected according to the specific task scenario, such as decision tree, neural network, support vector machine, etc.

[0027] Step 4: Ensemble classifier training: After the node classifier training is completed, the recognition result vector is obtained. The recognition result vector is used to form an ensemble training sample set, and the ensemble classifier is trained using the ensemble training sample set. The ensemble classifier can be freely selected according to the specific classification scenario, such as adaboost. Figure 3 shown.

[0028] Step 5: The overall process of the modulation parameter adaptive identification of the present invention is as follows: Figure 4As shown in the figure. For the signal to be identified, the identification result of the node classifier is used as the input of the integrated classifier, and the integrated classifier gives the final modulation parameter identification result. The node classifier and the integrated classifier together constitute the modulation parameter identification scheme. The role of the integrated classifier is to integrate the processing results of each node classifier under different analysis bandwidths, and overcome the identification instability problem caused by setting a single analysis bandwidth in the traditional method.

[0029] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by technicians in this technical field within the essential scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A method for adaptively identifying signal modulation parameters, characterized in that: The following steps are involved: Step 1: Automatically adjust the analysis bandwidth: After the user selects the location of the signal to be identified, the frequency at which the maximum value of the power spectrum mean in the local range is located is selected as the signal center frequency f c , select the signal center frequency f c The 3dB bandwidth is taken as the initial analysis bandwidth B1, and the analysis bandwidth is gradually increased with k*B1 as the step size to form an analysis bandwidth set {B1, B2, ..., B N }, where k is the step size coefficient; Step 2: Construct a training sample set: Construct a training sample set for signals with different modulation modes; collect N different samples for each signal according to the different bandwidths in the analysis bandwidth set, denoted as s (c,i,b) , all samples constitute a training sample set, where c represents the modulation parameter type of the sample, i represents the i-th signal under the c-th modulation parameter type, and b represents the sample collected by the i-th signal sample under the c-th modulation parameter type according to the b-th analysis bandwidth; Step 3: Node classifier training: The training sample set is divided into N training subsets according to the analysis bandwidth, and N modulation parameter recognition classifiers, i.e., node classifiers, are trained respectively; Step 4: Integrated classifier training: After the node classifier training is completed, the recognition result vector is obtained, and the recognition result vector is used to form an integrated training sample set, and the integrated training sample set is used to train the integrated classifier; Step 5: Combine the trained node classifier and ensemble classifier into a whole, that is, the results of the node classifier are concentrated and input into the ensemble classifier, and the ensemble classifier outputs the final recognition result.

Images