Method and apparatus for discriminating line-of-sight and non-line-of-sight signals, device, and storage medium

Abstract

The present invention provides a method and device, equipment and storage medium for discriminating line-of-sight and non-line-of-sight signals. The method stacks the real part, imaginary part and modulus data divided from the received original communication signal in the form of complex numbers according to the RGB color mode to form a signal image corresponding to the original communication signal; and inputs the signal image into a preset convolutional neural network so that the convolutional neural network outputs the signal type corresponding to the original communication signal. By stacking the real part, imaginary part and modulus data divided from the original communication signal in the form of complex numbers, the present invention forms a signal image that retains the amplitude and phase information of the original signal for signal classification through a convolutional neural network, which can not only improve the accuracy of feature extraction of the original communication signal, but also improve the generalization performance of signal recognition; thus, when ranging through communication signals, the problem of positioning errors caused by non-line-of-sight signals can be effectively alleviated.

Description

Technical Field

[0001] The present invention relates to the technical field, and in particular to a method and apparatus, device, and storage medium for distinguishing line-of-sight and non-line-of-sight signals. Background Art

[0002] With the advancement of communications technology, the high-precision ranging capabilities of positioning technologies such as fifth-generation mobile communications (5G) and ultra-wideband (UWB) have promoted a wider range of positioning applications and services. However, due to complex indoor environments, receivers are sometimes in a non-line-of-sight (NLOS) state with the transmitter, significantly impacting the performance of positioning technologies. NLOS conditions are generally considered the most significant factor affecting indoor ranging and positioning performance. This NLOS condition occurs when the signal from a transmitter cannot travel directly to the receiver in a straight line. Under these conditions, the transmitted signal propagates through obstacles or repeatedly reflects off the surfaces of obstacles before ultimately reaching the receiver. The transmission, refraction, and reflection that occur during this process not only depletes the transmitted signal's energy, reducing its received power upon reaching the receiver, but also, due to the longer transmission path, the timestamp estimated by the receiver upon receiving the signal is often overstated. Due to these errors, NLOS conditions can severely impact indoor ranging and positioning accuracy. Therefore, effectively mitigating NLOS errors is a critical issue. Before mitigating NLOS errors, identifying LOS / NLOS signals is crucial.

[0003] The Third Generation Partnership Project (3GPP) has discussed how to mitigate multipath and NLOS errors in enhanced indoor positioning at the R17 stage, and the meeting first reached a consensus on the establishment of a LOS / NLOS indicator to indicate whether the received signal is a LOS signal or a NLOS signal. In terms of identifying LOS and NLOS signals, many studies are based on CIR (channel impulse response) signals for detection. Currently, the processing methods for distinguishing LOS / NLOS signals mainly include: (1) Threshold-based NLOS identification method. This method is generally based on a single eigenvalue threshold judgment method. By extracting a certain eigenvalue from the CIR signal, a certain threshold is set based on experience, and the threshold is used as the boundary for distinguishing LOS / NLOS signals. This method was more common in the early days and has a certain accuracy, but it has certain limitations. When the application scenario switches, the changes in factors such as the location of obstacles and environmental noise will have a relatively large impact on the threshold, causing the performance of this method to drop significantly when applied to new scenarios. In addition, the process of manually setting the threshold is also relatively cumbersome, so the generalization of this method is not ideal. (2) NLOS identification methods based on statistical models. This method generally first examines the probability density distribution of a certain characteristic value of the LOS / NLOS signal and then performs NLOS identification based on the statistical characteristics of a large amount of data. This type of method usually has a high classification accuracy. However, since it is based on a probability model, it usually takes a longer time and therefore cannot perform NLOS detection in real time. (3) NLOS identification methods based on machine learning extract multiple characteristic values of the CIR signal and select the best feature combination as input to obtain the signal classification result. This method requires manual selection of the optimal classification combination from a large number of feature combinations, which is time-consuming and labor-intensive. Moreover, experiments have shown that when the application scenario changes, the optimal feature classification combination of the signal will also change. Therefore, this method cannot guarantee the robustness of the detection system. Summary of the invention

[0004] In view of this, embodiments of the present invention provide a method and apparatus, a device, and a storage medium for distinguishing line-of-sight and non-line-of-sight signals, so as to eliminate or improve one or more defects in the prior art.

[0005] One aspect of the present invention provides a method for distinguishing line-of-sight and non-line-of-sight signals, comprising:

[0006] The real part, imaginary part and modulus data of the received original communication signal divided into complex form are stacked according to the RGB color mode to form a signal image corresponding to the original communication signal; the signal image is input into a preset convolutional neural network so that the convolutional neural network outputs a signal type corresponding to the original communication signal, and the signal type includes: line-of-sight signal and non-line-of-sight signal.

[0007] In some embodiments of the present invention, stacking the real part, imaginary part, and modulus data obtained by dividing the received original communication signal in complex form according to the RGB color mode to form a signal image corresponding to the original communication signal includes:

[0008] Combining the real part, imaginary part, and modulus data obtained by dividing the discrete communication signal components received from each channel in complex form to form a real part row vector, an imaginary part row vector, and a modulus data row vector corresponding to an original communication signal for each of the communication signal components; reshaping the real part row vector, imaginary part row vector, and modulus data row vector to form a real part matrix, an imaginary part matrix, and a modulus data matrix of the same size; stacking the real part matrix, imaginary part matrix, and modulus data matrix together according to the RGB color mode to obtain a signal image corresponding to the original communication signal.

[0009] In some embodiments of the present invention, combining the real part, imaginary part, and modulus data obtained by dividing the discrete communication signal components received from each channel in complex form to form a real part row vector, an imaginary part row vector, and a modulus data row vector corresponding to an original communication signal for each of the communication signal components further includes: respectively preprocessing the real part row vector, imaginary part row vector, and modulus data row vector, scaling the element values in each row vector to a specific interval, and supplementing the number of elements in each row vector to a predetermined size.

[0010] In some embodiments of the present invention, scaling the element values in each row vector to a specific interval includes:

[0011] Taking the absolute value of each element in the real part row vector, imaginary part row vector, and modulus data row vector respectively, subtracting the minimum value in each row vector from the absolute value of the elements in each row vector, and then dividing by the maximum value in each vector to limit the element values in each row vector to [0, 1]; then multiplying the element values in each row vector limited to [0, 1] by a specific threshold of the element values to limit the element values in each row vector to a specific interval.

[0012] In some embodiments of the present invention, supplementing the element data in each row vector to a predetermined size includes: according to the size of the target matrix after reshaping the row vector, padding the elements in each row vector with 0 elements so that the number of elements in each row vector is the same as the number of elements in the target matrix.

[0013] In some embodiments of the present invention, stacking the real part matrix, imaginary part matrix, and modulus data matrix together according to the RGB color mode to obtain a signal image corresponding to the original communication signal includes: stacking the real part matrix, the imaginary part matrix, and the modulus data matrix together through three color channels of red, green, and blue to form a three-channel color image corresponding to the original communication signal.

[0014] In some embodiments of the present invention, it further includes: using the signal image corresponding to the historical communication signal and the signal classification label as the training data in the training set of the convolutional neural network to train the convolutional neural network, so as to obtain a convolutional application network for correspondingly outputting the signal type of the communication signal according to the signal image of the communication signal, wherein the signal classification label is used to identify the signal type of its uniquely corresponding historical communication signal.

[0015] Another aspect of the present invention provides a device for discriminating line-of-sight and non-line-of-sight signals, including:

[0016] A signal image generation module, configured to stack the real part, imaginary part, and modulus data divided from the received original communication signal in complex form according to the RGB color mode to form a signal image corresponding to the original communication signal; a signal type determination module, configured to input the signal image into a preset convolutional neural network, so that the convolutional neural network correspondingly outputs the signal type corresponding to the original communication signal, and the signal type includes: line-of-sight signal and non-line-of-sight signal.

[0017] Another aspect of the present invention provides an electronic device, including a processor and a memory, wherein computer instructions are stored in the memory, and the processor is configured to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method for discriminating line-of-sight and non-line-of-sight signals as described above.

[0018] Another aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the steps of the method for discriminating line-of-sight and non-line-of-sight signals as described above.

[0019] The method, device, equipment, and storage medium for discriminating line-of-sight and non-line-of-sight signals of the present invention stack the real part, imaginary part, and modulus data divided from the original communication signal in complex form to form a signal image that retains the amplitude and phase information of the original signal for signal classification through a convolutional neural network, enabling the convolutional neural network to simultaneously extract the amplitude features and phase features of the communication signal; thereby improving both the accuracy of discriminating signal types and the generalization performance of signal recognition; and further effectively alleviating the positioning error problem caused by non-line-of-sight signals when ranging through communication signals.

[0020] The additional advantages, objects, and features of the present invention will be partially described below and will become partially apparent to those of ordinary skill in the art after studying the following text, or can be learned from the practice of the present invention. The objects and other advantages of the present invention can be achieved and obtained through the structures specifically pointed out in the specification and the drawings.

[0021] Those skilled in the art will understand that the objectives and advantages achievable by the present invention are not limited to those specifically described above, and the above and other objectives achievable by the present invention will be more clearly understood from the following detailed description. Description of the Drawings

[0022] The drawings described herein are used to provide a further understanding of the present invention, form a part of this application, and do not limit the present invention.

[0023] Figure 1 It is a discriminant flow chart for line-of-sight and non-line-of-sight signals.

[0024] Figure 2 It is an amplitude curve graph of the received communication signal.

[0025] Figure 3 It is a data processing structure diagram for the received communication signal in an embodiment of the present invention.

[0026] Figure 4 It is a data processing structure diagram of a convolutional neural network in an embodiment of the present invention. Detailed Embodiments

[0027] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in combination with the embodiments and the drawings. Herein, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but do not limit the present invention.

[0028] Herein, it also needs to be noted that in order to avoid obscuring the present invention due to unnecessary details, only the structures and / or processing steps closely related to the solution according to the present invention are shown in the drawings, while other details less related to the present invention are omitted.

[0029] It should be emphasized that the term "comprising / including" when used herein refers to the presence of features, elements, steps, or components, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0030] Herein, it also needs to be noted that if not otherwise specified, the term "connection" in this article can not only refer to a direct connection, but also represent an indirect connection with an intermediate.

[0031] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0032] In the application process of the communication positioning technology mentioned in the background art, due to the complex external environment, the non-line-of-sight state is likely to occur between the signal transmitter and the signal receiver, thus interfering with the accuracy of signal ranging or signal positioning. Therefore, it becomes particularly crucial to distinguish whether the signal received by the signal receiver is a line-of-sight signal or a non-line-of-sight signal to ensure the accuracy of ranging or positioning. When distinguishing whether the received communication signal is a line-of-sight signal or a non-line-of-sight signal, researchers often analyze the power or amplitude of the received communication signal, and use the power curve graph or the amplitude curve graph as shown in Figure 2 as the input data of the convolutional neural network, so as to distinguish whether the signal type of the received communication signal is a line-of-sight signal or a non-line-of-sight signal through the convolutional neural network. However, since the current signal images used as the input data of the convolutional neural network are only transformed based on the real part data of the received communication signal, when using it as the input data of the convolutional neural network, the influence of the non-line-of-sight state between the transmitting end and the receiving end and the multipath effect of the communication signal on the phase of the signal is not considered. Therefore, there are still certain errors when the convolutional neural network performs image recognition.

[0033] In view of the problems existing in the above signal type recognition technology, the present application provides a method for discriminating line-of-sight and non-line-of-sight signals, as shown in Figure 1 including steps S110 - S120:

[0034] Step S110: Stack the real part, imaginary part, and modulus data of the received original communication signal divided in complex form according to the RGB color mode to form a signal image corresponding to the original communication signal.

[0035] The communication signal for ranging is sent from the transmitter. Due to the uncertainty of the wireless channel, the communication signal undergoes various forms of propagation such as scattering, refraction, and / or reflection during the transmission process. After reaching the receiver, it will form multipath transmission, and finally the receiver receives the communication signal components from multiple different channels; the communication signal components received by each channel are in complex form: Z n = a + bi, where n = 1, 2, 3,... represents the serial number of the communication signal component, a represents the real part of the current communication signal component, b represents the imaginary part of the current communication signal component, represents the modulus data of the current communication signal component.

[0036] In the above step S110, the real part, imaginary part, and modulus data obtained by dividing the discrete communication signal components received from each channel in complex form are combined to form a real part row vector, an imaginary part row vector, and a modulus data row vector corresponding to each of the communication signal components; then, preprocessing is respectively performed on the real part row vector, the imaginary part row vector, and the modulus data row vector, the element values in each row vector are scaled to a specific interval, and the number of elements in each row vector is supplemented to a predetermined size, so that the elements in each row vector can be reshaped into a matrix of a specific size; the preprocessed real part row vector, imaginary part row vector, and modulus data row vector are reshaped to form a real part matrix, an imaginary part matrix, and a modulus data matrix of the same size; the real part matrix, the imaginary part matrix, and the modulus data matrix are stacked together in the RGB color mode to obtain a signal image corresponding to the original communication signal.

[0037] In one or more embodiments of the present application, the step of scaling the element values in each row vector to a specific interval includes: taking the absolute value of each element in the real part row vector, the imaginary part row vector, and the modulus data row vector respectively, subtracting the minimum value in each row vector from the absolute value of the elements in each row vector, and then dividing by the maximum value in each vector, so as to limit the element values in each row vector to [0, 1]; then, the element values in each row vector limited to [0, 1] are respectively multiplied by a specific threshold of the element values to limit the element values in each row vector to a specific interval.

[0038] In one or more embodiments of the present application, the step of supplementing the element data in each row vector to a predetermined size includes: according to the size of the target matrix after reshaping the row vector, padding the elements in each row vector with 0 elements, so that the number of elements in each row vector is the same as the number of elements in the target matrix.

[0039] In one or more embodiments of the present application, the real part matrix, the imaginary part matrix, and the modulus data matrix are stacked together through three color channels of red, green, and blue to form a three-channel color image (RGB image) corresponding to the original communication signal.

[0040] In one embodiment, a signal pulse response signal (CIR signal) sent by a base station is received at a fixed point indoors. Among them, the CIR signal received at the fixed point indoors includes discrete CIR signal components received from 1016 channels. Through the above step S110, the received CIR signal data is processed to obtain a signal image corresponding to the CIR signal for use as input data of a convolutional neural network, as Figure 3 shown, including steps S111 - S114:

[0041] Step S111: Separate the real part, imaginary part, and modulus data of the CIR signal components of the received 1016 channels in the plural form, and combine the real part, imaginary part, and modulus data in each CIR signal component according to the division method of the real part, imaginary part, and modulus data to form three row vectors with a length of 1016, including a real part row vector, an imaginary part row vector, and a modulus data row vector;

[0042] Step S112: Preprocess each row vector. Take the absolute value of each element in each row vector, subtract the minimum value in the row vector where the element is located from the absolute value of each element, and then divide the difference between the two by the maximum value in the row vector where the element is located, so as to limit the element values in each row vector within the interval [0, 1]. Then multiply the elements of each row vector limited within the interval [0, 1] by a preset specific threshold 255 required for subsequent convolution, so as to scale the value of each element in each row vector to within the interval [0, 255]; The formulas for preprocessing the three row vectors are respectively:

[0043]

[0044]

[0045]

[0046] Among them, Vctor real represents the real part row vector, Vctor Imaginary represents the imaginary part row vector, ctor Mod represents the modulus data row vector.

[0047] Step S113: Supplement 8 zeros after the last element of each preprocessed row vector, so that the length of each row vector becomes 1024; And reshape the three row vectors after padding into a 32×32 matrix, namely a real part matrix, an imaginary part matrix, and a modulus data matrix;

[0048] Step S114: Stack the three matrices together in the way of the RGB color mode to obtain a three-channel color image corresponding to the received CIR signal.

[0049] In this application, data preprocessing, padding, and reshaping are simultaneously performed on the real part, imaginary part, and modulus data in the received original communication signal to obtain a signal image corresponding to the original communication signal. This signal image retains the phase information and amplitude information of the original communication signal, can effectively improve the effectiveness and reliability of feature extraction, and further improve the discrimination accuracy of signal classification.

[0050] Step S120: Input the signal image into a preset convolutional neural network so that the convolutional neural network outputs the signal type corresponding to the original communication signal. The signal type includes: line-of-sight signal and non-line-of-sight signal.

[0051] The convolutional neural network is used to automatically extract data features from the input data and classify the input data according to the data features, and output whether the signal type of the input data is a line-of-sight signal or a non-line-of-sight signal. The convolutional neural network includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. Among them, the input layer is used to introduce the input data into the convolutional neural network; the convolutional layer is used to perform a convolution operation on the input data through a set of convolution kernels to extract the data features of the input data; the pooling layer is arranged after the convolutional layer and is used to reduce the spatial size of the output of the convolutional layer, thereby reducing the computational complexity of the subsequent layers and preventing overfitting; the fully connected layer is arranged after the pooling layer and is used to map the extracted data features to the corresponding classification labels, thereby converting the data features into corresponding classification probabilities; the output layer is arranged after the fully connected layer and usually uses the softmax function (normalized exponential function) to normalize the probabilities of the data features in each classification, and outputs the classification label obtained by predicting the input data according to the normalized result.

[0052] Before the convolutional neural network is actually applied, the signal image corresponding to the historical communication signal and the signal classification label are used as the training data in the training set of the convolutional neural network to train the convolutional neural network, so as to obtain a convolutional neural network for outputting the signal type of the communication signal according to the signal image of the communication signal. Among them, the signal classification label is used to identify the signal type of its uniquely corresponding historical communication signal. The training process of the above convolutional neural network is as follows: First, input the historical communication signal in the training data set into the input layer of the convolutional neural network, and through forward propagation, pass through the convolutional layer, pooling layer, and fully connected layer in sequence, and finally output the classification label corresponding to the historical communication signal from the output layer; then compare the classification label output by the output layer with the true classification label of the historical communication signal. The convolutional neural network updates the weights and biases in the convolutional neural network through backward propagation, passing through the fully connected layer, pooling layer, and convolutional layer in sequence. After training the convolutional neural network with multiple training data, the prediction result output by the convolutional neural network for the communication signal is made as close as possible to its corresponding true classification change, thereby completing the training of the convolutional neural network. After the convolutional neural network is trained, the test data in the test set can also be input into the input layer of the trained convolutional neural network, and through forward propagation, the classification label of the test data is output by the output layer. If the output classification label is the same as the true classification label of the test data, it means that the convolutional neural network has been trained. Otherwise, the convolutional neural network needs to be trained again.

[0053] In one embodiment, the process of using a convolutional neural network to extract features and classify features of a signal image corresponding to an original communication signal is as follows Figure 4 As shown, the signal image corresponding to the original communication signal is input into the input layer of the convolutional neural network. In this convolutional neural network, through forward propagation, it sequentially passes through the convolutional layer, pooling layer, and fully connected layer, and finally outputs the classification label corresponding to the signal image from the output layer to obtain the signal type corresponding to the signal image. Among them, in the input layer, according to the need to extract features from the signal image, convolutional kernels in the input layer are set. Each convolutional kernel can learn a specific feature, such as edges, corners, textures, colors, etc.; in the convolutional layer, the convolutional kernels slide on the signal image at a certain stride, and during each sliding process, the convolutional kernels are multiplied point by point with the corresponding regions in the signal image, and then the results of all point multiplications are added together to obtain a pixel value of the output image of the convolutional layer. In this way, the output image of the convolutional layer is obtained; in the pooling layer, the maximum or average value in the window is taken; in the fully connected layer, the feature vector is converted into a classification probability and the classification result of the signal image is given; the signal type of the signal image is output by the output layer.

[0054] Compared with the classification method of manually inputting image features and then using traditional machine learning algorithms such as support vector machines, the convolutional neural network adopted in this application can automatically extract the features of the signal image for high-dimensional signal image data, more conveniently and accurately extract the features in the signal image, and can effectively avoid the trouble of manually extracting features; the convolutional layer in the convolutional neural network has the characteristics of local connectivity and weight sharing, which can reduce the number of parameters, thus effectively preventing overfitting. While retaining the spatial signal of the signal image, the computational amount can be greatly reduced; in the convolutional neural network, more complex features can be learned by setting multiple convolutional layers and pooling layers, which can effectively improve the reliability and effectiveness of feature extraction. Therefore, using a convolutional neural network to extract features and classify the signal image in this application can effectively improve the generalization performance, especially having more prominent advantages for high-dimensional image data.

[0055] From the above analysis content, it can be seen that the method for discriminating line-of-sight and non-line-of-sight signals described in this application effectively improves the reliability of feature extraction and the accuracy of signal classification by retaining the phase information of the original communication signal; at the same time, by using a convolutional neural network to extract features and classify the signal image, the generalization performance of the method for discriminating line-of-sight and non-line-of-sight signals is effectively improved.

[0056] Correspondingly to the above method, the present invention further provides a discrimination device for line-of-sight and non-line-of-sight signals, including: a signal image generation module, configured to stack the real part, imaginary part, and phase data obtained by dividing the received original communication signal in complex form according to the RGB color mode to form a signal image corresponding to the original communication signal; a signal type determination module, configured to input the signal image into a preset convolutional neural network, so that the convolutional neural network outputs the signal type corresponding to the original communication signal, and the signal type includes: line-of-sight signals and non-line-of-sight signals.

[0057] In a specific embodiment, the signal image generation module may execute the processing flow described in step S110 of the above method for discriminating line-of-sight and non-line-of-sight signals, which will not be elaborated here and can refer to the detailed description of step S110 in the above method for discriminating line-of-sight and non-line-of-sight signals. In a specific embodiment, the signal type determination module may execute the processing flow described in step S120 of the above method for discriminating line-of-sight and non-line-of-sight signals, which will not be elaborated here and can refer to the detailed description of step S120 in the above method for discriminating line-of-sight and non-line-of-sight signals.

[0058] Correspondingly to the above method, the present invention further provides an electronic device, which includes a computer device. The computer device includes a processor and a memory. The memory stores computer instructions, and the processor is configured to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the electronic device implements the steps of the method for discriminating line-of-sight and non-line-of-sight signals as described above.

[0059] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the foregoing method for discriminating line-of-sight and non-line-of-sight signals are implemented. The computer-readable storage medium may be a tangible storage medium, such as a random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, register, floppy disk, hard disk, removable storage disk, CD-ROM, or any other form of storage medium known in the art.

[0060] Those of ordinary skill in the art should understand that the various exemplary components, systems, and methods described in connection with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Specifically, whether to implement it in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention. When implemented in hardware, it can be, for example, an electronic circuit, an application-specific integrated circuit (ASIC), appropriate firmware, a plug-in, a functional card, and so on. When implemented in software, the elements of the present invention are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave on a transmission medium or a communication link.

[0061] It should be clear that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order between steps after understanding the spirit of the present invention.

[0062] In the present invention, the features described and / or illustrated for one embodiment can be used in the same or similar manner in one or more other embodiments, and / or combined with the features of other embodiments or replace the features of other embodiments.

[0063] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, various changes and modifications can be made to the embodiments of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for discriminating line-of-sight and non-line-of-sight signals, characterized in that, Including: Stacking the real part, imaginary part, and modulus data obtained by dividing the received original communication signal in complex form according to the RGB color model to form a signal image corresponding to the original communication signal; Inputting the signal image into a preset convolutional neural network so that the convolutional neural network outputs the signal type corresponding to the original communication signal, and the signal type includes: line-of-sight signal and non-line-of-sight signal.

2. The method according to claim 1, wherein The step of stacking the real part, imaginary part, and modulus data obtained by dividing the received original communication signal in complex form according to the RGB color model to form a signal image corresponding to the original communication signal includes: Combining the real part, imaginary part, and modulus data obtained by dividing the discrete communication signal components received from each channel in complex form to form a real part row vector, an imaginary part row vector, and a modulus data row vector corresponding to each of the communication signal components for an original communication signal; Reshaping the real part row vector, imaginary part row vector, and modulus data row vector to form a real part matrix, an imaginary part matrix, and a modulus data matrix with the same size; Stacking the real part matrix, imaginary part matrix, and modulus data matrix together according to the RGB color model to obtain a signal image corresponding to the original communication signal.

3. The method according to claim 2, wherein The step of combining the real part, imaginary part, and modulus data obtained by dividing the discrete communication signal components received from each channel in complex form to form a real part row vector, an imaginary part row vector, and a modulus data row vector corresponding to each of the communication signal components for an original communication signal further includes: Preprocessing the real part row vector, imaginary part row vector, and modulus data row vector respectively, scaling the element values in each row vector to a specific interval, and supplementing the number of elements in each row vector to a predetermined size.

4. The method according to claim 3, characterized in that The step of scaling the element values in each row vector to a specific interval includes: Taking the absolute value of each element in the real part row vector, imaginary part row vector, and modulus data row vector respectively, subtracting the minimum value in each row vector from the absolute value of the elements in each row vector, and then dividing by the maximum value in each vector to limit the element values in each row vector to [0, 1]; Then multiplying the element values in each row vector limited to [0, 1] by a specific threshold of the element values to limit the element values in each row vector to a specific interval.

5. The method according to claim 3, wherein The step of supplementing the element data in each row vector to a predetermined size includes: According to the size of the target matrix after reshaping the row vector, padding the elements in each row vector with 0 elements so that the number of elements in each row vector is the same as the number of elements in the target matrix.

6. The method according to claim 2, characterized in that, The step of stacking the real part matrix, imaginary part matrix, and modulus data matrix together according to the RGB color model to obtain a signal image corresponding to the original communication signal includes: Stacking the real part matrix, the imaginary part matrix, and the modulus data matrix together through three color channels of red, green, and blue to form a three-channel color image corresponding to the original communication signal.

7. The method according to claim 1, characterized in that, Also including; Using the signal image corresponding to the historical communication signal and the signal classification label as the training data in the training set of the convolutional neural network, training the convolutional neural network to obtain a convolutional neural network for correspondingly outputting the signal type of the communication signal according to the signal image of the communication signal, wherein the signal classification label is used to identify the signal type of its uniquely corresponding historical communication signal.

8. A discriminator for line-of-sight and non-line-of-sight signals, characterized in that, Including: A signal image generation module, configured to stack the real part, the imaginary part, and the Mo data divided from the received original communication signal in the complex form according to the RGB color mode to form a signal image corresponding to the original communication signal; A signal type determination module, configured to input the signal image into a preset convolutional neural network, so that the convolutional neural network correspondingly outputs the signal type corresponding to the original communication signal, and the signal type includes: line-of-sight signal and non-line-of-sight signal.

9. An electronic device, comprising a processor and a memory, characterized in that, Computer instructions are stored in the memory, and the processor is configured to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images