Specific protocol signal identification method of deep convolutional network based on CBAM

Abstract

The invention discloses a specific protocol signal identification method of a deep convolutional network based on CBAM. The method comprises the following steps: acquiring to-be-identified signal information; pre-processing the to-be-identified signal information to obtain target processing signal information; and performing identification processing on the target processing signal information to obtain target signal identification result information.

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing, and in particular, to a signal recognition method and device. Background Art

[0002] With the rapid development of computer technology and the popularization of its applications, information technology has been widely penetrated and deeply rooted in life. In the ever-changing external environment, all kinds of information need to be transmitted instantaneously, interconnected at any time and place, and closely connected between people. As a result, various wireless communication technologies and systems emerge in an endless stream. In the radio communication frequency band, there are a wide variety of communication rules and protocols, limited bandwidth, highly dense signals, and complex electromagnetic environments. For non-cooperative receivers without any prior knowledge, it is difficult to identify the communication protocol of signals through signal analysis. At present, most traditional communication signal protocol recognition methods still stay at the level of expert feature extraction, extracting the attribute features of target signals in ideal models or environments through a large amount of manual analysis. This method requires a large amount of manual analysis time, and is not suitable for complex environments or complex perception tasks, and is easily affected by the external environment, resulting in inaccurate recognition results and unable to ensure the applicability of the perceived tasks. This method based on manual feature extraction and using feature template matching algorithm performs well when there is sufficient prior information, but performs poorly in identifying unknown signals, has poor scalability, and is sensitive to signal-to-noise ratio, and is prone to false alarms under low signal-to-noise ratio. At present, for the recognition of specific protocol signals based on template matching and deep learning, there are problems such as few training signal samples, and it is not suitable for the detection and recognition of weak signals in complex electromagnetic environments, and the recognition accuracy is relatively low. Therefore, a signal recognition method and device are provided to improve the signal recognition accuracy, and further solve the problem of low recognition accuracy of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and poor transmission environment. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a signal recognition method and device, which are beneficial to improving the signal recognition accuracy, and further solve the problem of low recognition accuracy of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and poor transmission environment.

[0004] To solve the above technical problems, in the first aspect, an embodiment of the present invention discloses a signal recognition method, the method includes:

[0005] Obtain information of a signal to be recognized;

[0006] Perform preprocessing on the information of the signal to be recognized to obtain target processed signal information;

[0007] Perform recognition processing on the target processed signal information to obtain target signal recognition result information.

[0008] In a second aspect of the embodiments of the present invention, a signal recognition device is disclosed, and the device includes:

[0009] An acquisition module, configured to acquire signal information to be recognized;

[0010] A first processing module, configured to perform pre-processing on the signal information to be recognized to obtain target processed signal information;

[0011] A second processing module, configured to perform recognition processing on the target processed signal information to obtain target signal recognition result information.

[0012] In a third aspect of the present invention, another signal recognition device is disclosed, and the device includes:

[0013] A memory storing executable program code;

[0014] A processor coupled to the memory;

[0015] The processor calls the executable program code stored in the memory and executes some or all of the steps in the signal recognition method disclosed in the first aspect of the embodiments of the present invention.

[0016] In a fourth aspect of the present invention, a computer-readable storage medium is disclosed. The computer-readable storage medium stores computer instructions, and when the computer instructions are called, they are used to execute some or all of the steps in the signal recognition method disclosed in the first aspect of the embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0018] Figure 1 is a schematic diagram of the scenario of the signal recognition system provided by the embodiments of the present invention;

[0019] Figure 2 is a schematic flowchart of a signal recognition method disclosed by the embodiments of the present invention;

[0020] Figure 3 is a schematic structural diagram of a signal recognition device disclosed by the embodiments of the present invention;

[0021] Figure 4 is a schematic structural diagram of another signal recognition device disclosed by the embodiments of the present invention;

[0022] Figure 5It is a schematic structural diagram of a target signal recognition model disclosed in an embodiment of the present invention;

[0023] Figure 6 It is a schematic structural diagram of a first feature extraction module disclosed in an embodiment of the present invention;

[0024] Figure 7 It is a schematic structural diagram of a first feature analysis module disclosed in an embodiment of the present invention;

[0025] Figure 8 It is a schematic structural diagram of a first feature processing module disclosed in an embodiment of the present invention;

[0026] Figure 9 It is a schematic diagram showing the effect of a high - order spectrum

[0027] Figure 10 It is a schematic diagram showing the effect of wavelet packet decomposition disclosed in an embodiment of the present invention. Detailed implementation manners

[0028] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present invention.

[0029] The terms "first", "second", etc. in the specification and claims of the present invention and the above - mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non - exclusive inclusion. For example, a process, method, device, product or equipment that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or equipment.

[0030] Referring to "embodiment" in this context means that a specific feature, structure or characteristic described in connection with the embodiment can be included in at least one embodiment of the present invention. The phrase appears in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0031] In this application, the term "exemplary" is used to mean "serving as an example, illustration, or instance". Any embodiment described as "exemplary" in this application is not necessarily to be construed as more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use this application. In the following description, details are set forth for purposes of explanation. It should be understood that those of ordinary skill in the art can recognize that this application can be implemented without these specific details. In other instances, well-known structures and processes are not elaborated in detail to avoid obscuring the description of this application with unnecessary details. Therefore, this application is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed in this application.

[0032] It should be noted that since the method of the embodiments of this application is executed in a computer device, the processing objects of each computer device exist in the form of data or information. For example, time, which is actually time information. It can be understood that in subsequent embodiments, if dimensions, quantities, positions, etc. are mentioned, they are all corresponding data existences for the computer device to process, and specific details are not elaborated here.

[0033] A brief introduction to the artificial intelligence-related technologies that may be involved in this application is provided. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a way similar to human intelligence. Artificial intelligence also studies the design principles and implementation methods of various intelligent machines to enable the machines to have the functions of perception, reasoning, and decision-making.

[0034] Artificial intelligence technology is an interdisciplinary subject involving a wide range of fields, including both hardware-level technologies and software-level technologies. The basic technologies of artificial intelligence generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. The software technologies of artificial intelligence mainly include several major directions such as computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0035] Computer Vision Technology (CV) Computer vision is a science that studies how to enable machines to "see". More specifically, it refers to using cameras and computers to replace human eyes for tasks such as object recognition and measurement in machine vision, and further performing graphic processing to make the computer-processed images more suitable for human eyes to observe or be transmitted to instruments for detection. As a scientific discipline, computer vision research related theories and technologies, and attempts to build artificial intelligence systems that can obtain information from images or multi-dimensional data. Computer vision technology usually includes image processing, image recognition, image semantic understanding, image retrieval, OCR, video processing, video semantic understanding, video content / behavior recognition, three-dimensional object reconstruction, 3D technology, virtual reality, augmented reality, simultaneous localization and mapping, etc. technologies, and also includes common biometric recognition technologies such as face recognition and fingerprint recognition.

[0036] Single-modal information is data of only one type, such as one of the data information like text, image, audio, video, electromagnetic signal, etc. Multi-modal information is data information that includes at least 2 types of single-modal information. Further, multi-modal information is applicable to complex tasks that require integrating multiple information sources, such as sentiment analysis, robot interaction, autonomous driving, etc. By integrating information of multiple modalities, higher performance and accuracy can usually be achieved in tasks.

[0037] A large model refers to an artificial neural network model with a very large number of parameters. In the field of artificial intelligence, a large model usually refers to a model with hundreds of millions to trillions of parameters. The model usually needs to be trained on a large-scale dataset and requires a large amount of computing resources for optimization and adjustment. Large models are usually used to solve complex natural language processing, computer vision, and speech recognition tasks. Generative AI is a type of AI that can create new content and ideas, including conversations, stories, images, videos, and music. In the embodiments of this application, the large model can be large language models such as ChatGPT, BERT, XLNet, Zhipu Model, Claude, Moonshot AI Model, ChatGLM Model, Tongwen Qianyi Model, MiniMax Model, Spark Model, Llama Model, 360GPT Model, Qwen Model, Baichuan Model, Yunque Model, vivoLM Model, and Wenxin Yiyan, and the embodiments of this application do not make any limitations.

[0038] The embodiments of this application provide a signal recognition method, device, computer device, and computer-readable storage medium, which will be described in detail below.

[0039] Please refer to Figure 1 , Figure 1The figure is a schematic diagram of the scenario of the signal recognition system provided by the embodiments of the present application. The signal recognition system may include a computer device 100, and a signal recognition device is integrated in the computer device 100, such as Figure 1 the computer device in

[0040] In the embodiments of the present application, the computer device 100 is mainly used to obtain signal information to be recognized;

[0041] perform preprocessing on the signal information to be recognized to obtain target processed signal information;

[0042] perform recognition processing on the target processed signal information to obtain target signal recognition result information.

[0043] It can improve the signal recognition accuracy, and further solve the problem of low recognition accuracy of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and poor transmission environment.

[0044] In the embodiments of the present application, the computer device 100 may be an independent server, or a server network or server cluster composed of servers. For example, the computer device 100 described in the embodiments of the present application includes, but is not limited to, a computer, a network host, a single network server, a set of multiple network servers, or a cloud server composed of multiple servers. Among them, the cloud server is composed of a large number of computers or network servers based on cloud computing.

[0045] It can be understood that the computer device 100 used in the embodiments of the present application may be a device that includes both receiving and transmitting hardware, that is, a device having receiving and transmitting hardware capable of performing two-way communication on a two-way communication link. Such devices may include: cellular or other communication devices, which have a single-line display or a multi-line display, or cellular or other communication devices without a multi-line display. Specifically, the computer device 100 may be a desktop terminal or a mobile terminal, and the computer device 100 may specifically also be one of a mobile phone, a tablet computer, a laptop computer, etc.

[0046] Those skilled in the art can understand that Figure 1 the application environment shown in Figure 1 is only one application scenario of the solution of the present application, and does not constitute a limitation on the application scenario of the solution of the present application. Other application environments may also include more or fewer computer devices than those shown in Figure 1 For example, only 1 computer device is shown in Figure 1 . It can be understood that the signal recognition system may further include one or more other services, which are not specifically limited here.

[0047] In addition, as shown in Figure 1As shown, the signal recognition system may further include a memory 200 for storing data such as image data, location information, etc.

[0048] It should be noted that Figure 1 The schematic diagram of the scenario of the signal recognition system shown is only an example. The signal recognition system and scenario described in the embodiments of the present application are for more clearly explaining the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. Those of ordinary skill in the art can know that with the evolution of the signal recognition system and the emergence of new service scenarios, the technical solutions provided by the embodiments of the present application are equally applicable to similar technical problems.

[0049] The present invention discloses a signal recognition method and device, which are beneficial to improving the signal recognition accuracy, and further solve the problem of low recognition accuracy of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and poor transmission environment. The following will be described in detail respectively.

[0050] Embodiment 1

[0051] Please refer to Figure 2 , Figure 2 which is a schematic flowchart of a signal recognition method disclosed in an embodiment of the present invention. Among them, Figure 2 The described signal recognition method is applied in a management system, such as a local server or a cloud server for management, etc., and the embodiments of the present invention do not make limitations. As Figure 2 shown, the signal recognition method may include the following operations:

[0052] 101. Obtain the signal information to be recognized.

[0053] 102. Perform preprocessing on the signal information to be recognized to obtain the target processed signal information.

[0054] 103. Perform recognition processing on the target processed signal information to obtain the target signal recognition result information.

[0055] It should be noted that the above preprocessing of the signal information to be recognized is to perform frequency conversion, filtering, and noise reduction processing on the signal to achieve parameter extraction and recognition analysis of the signal at a lower signal-to-noise ratio, and the embodiments of the present invention do not make limitations.

[0056] It should be noted that the above signal information to be recognized can be received by a radar or obtained by digital signal conversion, and the embodiments of the present invention do not make limitations.

[0057] Further, before performing identification processing on the target processing signal information, blind channel estimation and channel matching filtering processing can also be performed on the target processing signal information, that is, the influence brought by channel interference is suppressed through blind channel estimation and channel matching filtering processing, which is beneficial to improving the robustness of feature extraction to overcome the influence brought by channel distortion. The embodiments of the present invention are not limited thereto.

[0058] In this optional embodiment, as an optional implementation manner, the above preprocessing of the signal information to be recognized to obtain the target processing signal information includes:

[0059] Performing noise reduction processing on the signal information to be recognized to obtain the first processing signal information;

[0060] Performing time-domain waveform feature marking processing on the first processing signal information to obtain the second processing signal information;

[0061] Performing fast Fourier transform on the second processing signal information to obtain frequency-domain signal information;

[0062] Performing frequency-domain feature frame marking processing on the frequency-domain signal information to obtain the target processing signal information.

[0063] It should be noted that the above noise reduction processing can be to perform noise reduction processing on the signal to be recognized by using multi-scale wavelet transform. The embodiments of the present invention are not limited thereto.

[0064] It should be noted that the above time-domain waveform feature marking processing on the first processing signal information is to mark the leading synchronization segment, signal data segment, end segment, etc. of the signal. The embodiments of the present invention are not limited thereto.

[0065] It should be noted that the above frequency-domain feature frame marking processing on the frequency-domain signal information is to mark the frequency-domain frame of the signal and feature segments such as the leading synchronization header, data segment, end segment, etc. of the frame

[0066] It should be noted that according to the unique protocol frame format of the protocol signal, different protocol signals exhibit different signal characteristics in the time domain and frequency domain. At this time, the time-frequency domain characteristics are used as the basis for discriminating the signal protocol type, and at the same time, the signal spectrum characteristics of its different powers can also be used for comprehensive detection. The high-order spectrum characteristics of PSK-like protocol signals are as Figure 9 shown. Different protocol signals will have different preamble sequences, and the signal waveforms formed by different preamble sequences will have very subtle differences. During signal acquisition, when the signal appears, the envelope waveform within a short time after the signal appears is extracted. After noise reduction, it is compared with the preamble signal waveform characteristics in the trained library. The embodiments of the present invention are not limited thereto.

[0067] It should be noted that since the communication protocol specifies the baseband modulation mode, communication rate, modulation frequency and interval, signal framing format, signal control code format and other parameters of the digital signal, the characteristics of these communication protocols are crucial to the identification of communication protocol attributes. To characterize the time-frequency characteristics of the signal, a centrally symmetrical sliding window is required to intercept the observed signal, perform Fourier transform on the signal in the window, and finally obtain the time-frequency spectrum composed of each segment of the signal. Since the time resolution and frequency resolution have an opposite effect, a compromise needs to be selected during the calculation process. For the ACARS air-to-ground data communication system structure widely used in China, timely and accurate information transmission provides reliable guarantee for civil aviation and shipping safety. There are many types of messages, including uplink messages and downlink messages. Different messages have different protocol formats. Feature extraction and analysis are performed on fixed positions of different protocol messages. For air-to-ground downlink messages and ground-to-air uplink message formats, their text formats are very different, and the positions and lengths of free text segments in the messages are different. These features can be extracted. After these different data sources are modulated into waveform signals, their amplitude values ​​can be displayed on the waveform and time frequency. Feature extraction and analysis are performed by comparing the positions of different signal amplitude strength values ​​such as free text and high-order spectra.

[0068] It should be noted that the signal recognition method of the present application solves the problem of identifying specific protocols under low noise, multipath delay, Doppler frequency shift, strong interference and strong aliasing conditions, can be well used for communication signal protocol recognition, has good classification performance, and has great promotion value, and the embodiments of the present invention are not limited thereto.

[0069] It can be seen that implementing the signal recognition method described in the embodiment of the present invention is conducive to improving the accuracy of signal recognition, thereby solving the problem of low recognition accuracy of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and harsh transmission environment.

[0070] In an optional embodiment, the target processing signal information is identified and processed to obtain target signal identification result information, including:

[0071] Performing a first signal feature extraction on the target processed signal information to obtain first extracted signal feature information;

[0072] Performing a second signal feature extraction on the target processed signal information to obtain second extracted signal feature information;

[0073] The target signal recognition model is used to perform recognition processing on the first extracted signal feature information and the second extracted signal feature information to obtain target signal recognition result information.

[0074] It should be noted that the above first signal feature extraction is to perform wavelet packet decomposition processing, that is, to use the ability of wavelet packet decomposition to represent the local features of the signal in both the time and frequency domains, perform local transformation of the signal in time and frequency, so as to fit the signal waveform characteristics that the high-frequency signal has a short duration and the low-frequency signal has a long duration. Not only the low-frequency part is decomposed, but also the high-frequency part is decomposed twice to achieve efficient signal denoising and signal analysis and extraction. The embodiments of the present invention are not limited. As Figure 10 shown, it is a schematic diagram of the multi-resolution analysis ability of wavelet packet with multi-layer scale decomposition. S represents the original signal; A represents the low-frequency; D represents the high-frequency; the serial number at the end represents the number of decomposition layers (i.e., the scale number). Assuming that the frequency range of the signal is [Fl, Fh], the original signal S can be considered as the decomposition transformation at scale zero. Each time the signal S passes through wavelet packet decomposition, it is equivalent to passing through a high-pass and a low-pass filter at the same time. After the original signal S passes through one-layer wavelet packet transformation, it is divided into two non-overlapping high-frequency D1 and low-frequency A1 parts with a scale parameter of 1. The frequency change range of the low-frequency A1 is Fl~(Fl + Fh) / 2, and the frequency change range of the high-frequency D1 is (Fl + Fh) / 2~Fh. If one-layer decomposition cannot meet the analysis needs, D1 and A1 can be continuously decomposed by wavelet packet. After two-layer decomposition, four frequency ranges are obtained, namely AA2, DA2, AD2, DD2, and their frequency change ranges are Fl~Fl + ΔF, Fl~Fl + 2ΔF, Fl~Fl + 3ΔF, Fl~Fl + 4ΔF respectively, where ΔF=(Fh - Fl) / 22. By analogy, decomposing continuously, a wavelet packet decomposition tree with a scale parameter of k and a total of 2k nodes can be obtained. Let the serial numbers of the nodes in the kth layer increase sequentially from left to right, that is, 1~2k, then the frequency change range of the jth (1≤j≤2k) node is Fl~Fl + j×(Fn - Fl) / 2k. Each time of decomposition, the frequency band is further divided and the frequency resolution is improved, while the time resolution is reduced.

[0075] In this optional embodiment, as an optional implementation manner, the above first signal feature extraction of the target processed signal information to obtain the first extracted signal feature information includes:

[0076] Performing wavelet packet decomposition processing on the target processed signal information to obtain signal coefficient information;

[0077] Performing coefficient threshold processing on the signal coefficient information to obtain signal threshold information;

[0078] Performing wavelet packet reconstruction processing on the signal coefficient information and the signal threshold information to obtain the first extracted signal feature information.

[0079] It should be noted that the above-mentioned wavelet packet decomposition processing of the target processing signal information selects an appropriate wavelet packet according to the signal characteristics and determines a wavelet decomposition layer number NN. The layer number NN is determined according to the accuracy of the frequency resolution and time resolution required in the algorithm. Then, the noisy signal is subjected to NN-layer wavelet packet decomposition to obtain wavelet decomposition coefficients (the signal coefficient information includes several wavelet decomposition coefficients). This is not limited in the embodiments of the present invention.

[0080] It should be noted that for the above-mentioned coefficient threshold processing of the signal coefficient information, for each wavelet packet decomposition coefficient, an appropriate threshold and threshold function are selected to perform threshold quantization on the coefficient. The selection of the threshold is crucial. For different thresholds, the signal-to-noise ratio has obvious differences. If the threshold is too large, too much signal detail is lost, which may cause signal distortion; if the threshold is too small, the expected noise filtering effect cannot be achieved. The selection of the threshold function can be divided into the hard threshold method and the soft threshold method, but both have their deficiencies. The general method is to take the threshold as where δ is the noise standard deviation. This project will analyze according to the actual situation to determine the selection of the threshold (the signal threshold information includes the signal thresholds corresponding to several wavelet decomposition coefficients). This is not limited in the embodiments of the present invention.

[0081] It should be noted that the above-mentioned wavelet packet reconstruction processing of the signal coefficient information and the signal threshold information is performed according to the wavelet packet decomposition coefficients of the NNth layer and the coefficients after threshold quantization. This is not limited in the embodiments of the present invention.

[0082] It should be noted that the above-mentioned second signal feature extraction is to calculate the high-order spectrum of the signal and mark it according to its characteristics. For the marking of the signal, since the frequency domain can well present the signal characteristics, these unique characteristics can be learned, and at the same time, the protocol can be identified by integrating its high-order spectrum characteristics. This is not limited in the embodiments of the present invention.

[0083] It should be noted that the signal envelope waveform, time-frequency, and high-order spectrum are detected and feature-marked according to the signal protocol characteristics (i.e., the first signal feature extraction and the second signal feature extraction). For signal marking, since the frequency domain can well present signal features, these unique features can be learned, and at the same time, the high-order spectrum features can be combined for protocol recognition. Since protocol signals generally have specific preamble sequences and these preamble sequences are generally sent cyclically, corresponding feature patterns are shown on the envelope. After comparing the denoised signal waveform with the envelope features in the training library, the protocol pattern of the signal can be quickly extracted, thus realizing fast and efficient recognition and classification of the signal. The embodiments of the present invention are not limited thereto. Further, the high-order spectrum signal feature extraction used in the second signal feature extraction is high-order cumulant and high-order spectrum features, which can not only suppress Gaussian noise but also retain the amplitude and phase information of communication signals and are independent of time. Among them, the bispectrum, as the third-order cumulant, has both the advantages of high-order cumulants and relatively simple calculation. The embodiments of the present invention are not limited thereto.

[0084] It should be noted that for the deep learning of feature extraction, although representation learning techniques have been widely applied in fields such as natural language processing and image processing, their applications in the communication field are few, and most of them directly use existing model algorithms in other fields for corresponding processing. The signal recognition method of this application is based on a target signal recognition model to extract and mark signal features.

[0085] It can be seen that implementing the signal recognition method described in the embodiments of the present invention is beneficial to improving the signal recognition accuracy, and further solves the problem of low recognition accuracy of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and harsh transmission environments.

[0086] In another optional embodiment, as Figure 5 shown, the target signal recognition model includes a first feature dimensionality reduction module, a second feature dimensionality reduction module, a first fusion module, a first convolution module, a first feature extraction module, a second feature extraction module, a third feature extraction module, a fourth feature extraction module, a fifth feature extraction module, a first normalization module, a first activation module, a first pooling module, a first feature analysis module, a second feature analysis module, a third feature analysis module, a fourth feature analysis module, a first feature processing module, a second feature processing module, a third feature processing module, a fourth feature processing module, a fifth feature processing module, a sixth feature processing module, a seventh feature processing module, an eighth feature processing module, a ninth feature processing module, a tenth feature processing module, and an eleventh feature processing module; among them,

[0087] The input end of the first feature dimension reduction module is configured to receive the first model input of the target signal recognition model, and the output end of the first feature dimension reduction module is connected to the input end of the first fusion module; the input end of the second feature dimension reduction module is configured to receive the second model input of the target signal recognition model, and the output end of the second feature dimension reduction module is connected to the input end of the first fusion module; the output end of the first fusion module is connected to the input end of the first convolution module; the output end of the first convolution module is connected to the input end of the first feature extraction module; the output end of the first feature extraction module is connected to the input end of the first normalization module; the output end of the first normalization module is connected to the input end of the first activation module; the output end of the first activation module is connected to the input end of the first pooling module; the output end of the first pooling module is connected to the input end of the first feature analysis module; the output end of the first feature analysis module is connected to the input end of the second feature extraction module; the output end of the second feature extraction module is connected to the input end of the first feature processing module; the output end of the first feature extraction module is connected to the input end of the second feature processing module; the output end of the second feature processing module is connected to the input end of the second feature analysis module; the output end of the second feature analysis module is connected to the input end of the third feature extraction module; the output end of the third feature extraction module is connected to the input end of the third feature processing module; the output end of the third feature processing module is connected to the input end of the fourth feature processing module; the output end of the fourth feature processing module is connected to the input end of the fifth feature processing module; the output end of the fifth feature processing module is connected to the input end of the third feature analysis module; the output end of the third feature analysis module is connected to the input end of the fourth feature extraction module; the output end of the fourth feature extraction module is connected to the input end of the sixth feature processing module; the output end of the sixth feature processing module is connected to the input end of the seventh feature processing module; the output end of the seventh feature processing module is connected to the input end of the eighth feature processing module; the output end of the eighth feature processing module is connected to the input end of the ninth feature processing module; the output end of the ninth feature processing module is connected to the input end of the tenth feature processing module; the output end of the tenth feature processing module is connected to the input end of the fourth feature analysis module; the output end of the fourth feature analysis module is connected to the input end of the fifth feature extraction module; the output end of the fifth feature extraction module is connected to the input end of the eleventh feature processing module; the output end of the eleventh feature processing module is configured to output the model output of the target signal recognition model.

[0088] It should be noted that the above target signal recognition model has the characteristics of better extraction of high-level abstract features and network performance, and will not cause the problem of network degradation as the network deepens. It can efficiently extract key features, so as to accurately realize the recognition and classification of signals. The embodiments of the present invention are not limited thereto.

[0089] It should be noted that the above first model input and second model input respectively represent the first extracted signal feature information and the second extracted signal feature information, and the model output represents the target signal recognition result information, which is not limited in the embodiments of the present invention.

[0090] It should be noted that the model architectures of the above first feature dimensionality reduction module and second feature dimensionality reduction module are the same, which is not limited in the embodiments of the present invention.

[0091] It should be noted that the model architectures of the above first feature extraction module, second feature extraction module, third feature extraction module, fourth feature extraction module, and fifth feature extraction module are the same, which is not limited in the embodiments of the present invention.

[0092] It should be noted that the model architectures of the above first feature analysis module, second feature analysis module, third feature analysis module, and fourth feature analysis module are the same, which is not limited in the embodiments of the present invention.

[0093] It should be noted that the model architectures of the above first feature processing module, second feature processing module, third feature processing module, fourth feature processing module, fifth feature processing module, sixth feature processing module, seventh feature processing module, eighth feature processing module, ninth feature processing module, tenth feature processing module, and eleventh feature processing module are the same, which is not limited in the embodiments of the present invention.

[0094] It should be noted that the above first fusion module is constructed based on element-wise addition, which is not limited in the embodiments of the present invention.

[0095] It should be noted that the above first convolutional module can be constructed based on a deep convolutional network with CBAM, which is not limited in the embodiments of the present invention.

[0096] It should be noted that the above first normalization module is constructed based on a batch normalization layer, which is not limited in the embodiments of the present invention.

[0097] It should be noted that the above first activation module is constructed based on the RELU activation function, which is not limited in the embodiments of the present invention.

[0098] It should be noted that the above first pooling module is constructed based on a max pooling layer, which is not limited in the embodiments of the present invention.

[0099] It can be seen that implementing the signal recognition method described in the embodiments of the present invention is beneficial to improving the signal recognition accuracy, and further solves the problem of low recognition accuracy of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and harsh transmission environments.

[0100] In another optional embodiment, as Figure 5As shown in the figure, the first feature dimensionality reduction module includes a first convolutional unit, a second convolutional unit, a first normalization unit, a second normalization unit, a first activation unit, a second activation unit, and a first fusion unit; where,

[0101] The input end of the first convolutional unit and the input end of the first fusion unit are configured to receive the module input of the first feature dimensionality reduction module; the output end of the first convolutional unit is connected to the input end of the first normalization unit; the output end of the first normalization unit is connected to the input end of the first activation unit; the output end of the first activation unit is connected to the input end of the second convolutional unit; the output end of the second convolutional unit is connected to the input end of the second normalization unit; the output end of the second normalization unit is connected to the input end of the first fusion unit; the output end of the first fusion unit is connected to the input end of the second activation unit; the output end of the second activation unit is configured to output the module output of the first feature dimensionality reduction module.

[0102] It should be noted that the module input of the above first feature dimensionality reduction module represents the first model input, and the module output of the first feature dimensionality reduction module represents the data output from the output end of the second activation unit, which is not limited in the embodiments of the present invention.

[0103] It should be noted that the above first feature dimensionality reduction module realizes the dimensionality reduction processing of data through the convolutional layer in the residual structure. After being processed by the first feature dimensionality reduction module, the data dimensionality can be reduced by half to reduce the number of parameters in the subsequent model processing, thereby reducing the calculation amount and improving the processing efficiency of the model for signal recognition and classification, which is not limited in the embodiments of the present invention.

[0104] It should be noted that the convolutional kernels of the above first convolutional unit and second convolutional unit are 1×3 in size and the number of channels is 2, which is not limited in the embodiments of the present invention.

[0105] It should be noted that the above first normalization unit and second normalization unit are constructed based on the batch normalization layer, which is not limited in the embodiments of the present invention.

[0106] It should be noted that the above first activation unit and second activation unit are constructed based on the RELU activation function, which is not limited in the embodiments of the present invention.

[0107] It should be noted that the above first fusion unit is constructed based on the element-wise addition operation, which is not limited in the embodiments of the present invention.

[0108] It can be seen that implementing the signal recognition method described in the embodiments of the present invention is beneficial to improving the signal recognition accuracy, and further solves the problem of low recognition accuracy of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and poor transmission environment.

[0109] In yet another alternative embodiment, as Figure 6As shown in the figure, the first feature extraction module includes a first pooling unit, a second pooling unit, a third pooling unit, a fourth pooling unit, a fifth pooling unit, a sixth pooling unit, a multi-layer perceptron, a second fusion unit, a third fusion unit, and a fourth fusion unit; among them,

[0110] The input ends of the first pooling unit, the second pooling unit, and the third fusion unit are configured to receive the first module input of the first feature extraction module; the output ends of the first pooling unit and the second pooling unit are both connected to the input end of the multi-layer perceptron; the output end of the multi-layer perceptron is respectively connected to the input ends of the third pooling unit and the fourth pooling unit; the output ends of the third pooling unit and the fourth pooling unit are both connected to the input end of the second fusion unit; the output end of the second fusion unit is connected to the input end of the third fusion unit; the output end of the third fusion unit is respectively connected to the input ends of the fifth pooling unit and the fourth fusion unit; the output end of the fifth pooling unit is connected to the input end of the sixth pooling unit; the output end of the sixth pooling unit is connected to the input end of the fourth fusion unit; the output end of the fourth fusion unit is configured to output the first module output of the first feature extraction module.

[0111] It should be noted that the first module input of the above-mentioned first feature extraction module represents the data output from the output end of the first convolutional module, and the first module output of the first feature extraction module represents the data output from the output end of the first feature extraction module. The embodiments of the present invention are not limited thereto.

[0112] It should be noted that the above-mentioned first feature extraction module improves the feature expression by dynamically focusing on the important regions of the input feature map through the feature extraction of multiple pooling units and perceptrons, extracts more targeted features, and thus effectively extracts and fuses features at different scales to further improve the feature expression ability of the target signal recognition model. The embodiments of the present invention are not limited thereto.

[0113] It should be noted that the above-mentioned second pooling unit, fourth pooling unit, and sixth pooling unit are constructed based on the average pooling layer. The embodiments of the present invention are not limited thereto. The above-mentioned first pooling unit, third pooling unit, and fifth pooling unit are constructed based on the max pooling layer. The embodiments of the present invention are not limited thereto.

[0114] It should be noted that the above-mentioned second fusion unit, third fusion unit, and fourth fusion unit are constructed based on element-wise addition. The embodiments of the present invention are not limited thereto.

[0115] It should be noted that the number of hidden layers of the above-mentioned multi-layer perceptron is greater than 1, and the number of single-layer neurons is not less than 20. The embodiments of the present invention are not limited thereto.

[0116] It can be seen that implementing the signal recognition method described in the embodiments of the present invention is beneficial to improving the signal recognition accuracy, and further solves the problem of low recognition accuracy of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and poor transmission environments.

[0117] In an alternative embodiment, as Figure 7 shown, the first feature analysis module includes a third convolution unit, a fourth convolution unit, a fifth convolution unit, a sixth convolution unit, a third normalization unit, a fourth normalization unit, a fifth normalization unit, a sixth normalization unit, a third activation unit, a fourth activation unit, a fifth activation unit, and a fifth fusion unit; wherein,

[0118] The input ends of the third convolution unit and the sixth convolution unit are both configured to receive the second module input of the first feature analysis module; the output end of the third convolution unit is connected to the input end of the third normalization unit; the output end of the third normalization unit is connected to the input end of the third activation unit; the output end of the third activation unit is connected to the input end of the fourth convolution unit; the output end of the fourth convolution unit is connected to the input end of the fourth normalization unit; the output end of the fourth normalization unit is connected to the input end of the fourth activation unit; the output end of the fourth activation unit is connected to the input end of the fifth convolution unit; the output end of the fifth convolution unit is connected to the input end of the fifth normalization unit; the output end of the fifth normalization unit is connected to the input end of the fifth fusion unit; the output end of the sixth convolution unit is connected to the input end of the sixth normalization unit; the output end of the sixth normalization unit is connected to the input end of the fifth fusion unit; the output end of the fifth fusion unit is connected to the input end of the fifth activation unit; the output end of the fifth activation unit is configured to output the second module output of the first feature analysis module.

[0119] It should be noted that the above-mentioned second module input of the first feature analysis module represents the data output from the output end of the first pooling module, and the second module output of the first feature analysis module represents the data output from the output end of the second module of the first feature analysis module. The embodiments of the present invention do not make any limitations.

[0120] It should be noted that the convolution kernel sizes of the above-mentioned third convolution unit, fourth convolution unit, fifth convolution unit, and sixth convolution unit are 1×1, 3×3, 1×1, and 1×1 respectively, and the number of channels are 4, 4, 1, and 1 respectively. The embodiments of the present invention do not make any limitations.

[0121] It should be noted that the above-mentioned third normalization unit, fourth normalization unit, fifth normalization unit, and sixth normalization unit are all constructed based on the batch normalization layer. The embodiments of the present invention do not make any limitations.

[0122] It should be noted that the above-mentioned third activation unit, fourth activation unit, and fifth activation unit are all constructed based on RELU activation. The embodiments of the present invention do not make any limitations.

[0123] It should be noted that the above fifth fusion unit is constructed based on element-wise addition operation, and the embodiments of the present invention are not limited thereto.

[0124] It should be noted that the above first feature analysis module realizes multi-dimensional extraction of signal features by utilizing the cooperation of different channel numbers of the convolutional unit, thereby realizing the fusion analysis of multi-dimensional feature information of the model, improving the accuracy of signal recognition and classification, and the embodiments of the present invention are not limited thereto.

[0125] It can be seen that implementing the signal recognition method described in the embodiments of the present invention is beneficial to improving the signal recognition accuracy rate, and further solves the problem of low recognition accuracy rate of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and harsh transmission environments.

[0126] In another optional embodiment, as Figure 8 shown, the first feature processing module includes a seventh convolutional unit, an eighth convolutional unit, a ninth convolutional unit, a seventh normalization unit, an eighth normalization unit, a ninth normalization unit, a sixth activation unit, a seventh activation unit, an eighth activation unit, and a sixth fusion unit; wherein,

[0127] The input end of the seventh convolutional unit and the input end of the sixth fusion unit are configured to receive the third module input of the first feature processing module; the output end of the seventh convolutional unit is connected to the input end of the seventh normalization unit; the output end of the seventh normalization unit is connected to the input end of the sixth activation unit; the output end of the sixth activation unit is connected to the input end of the eighth convolutional unit; the output end of the eighth convolutional unit is connected to the input end of the eighth normalization unit; the output end of the eighth normalization unit is connected to the input end of the seventh activation unit; the output end of the seventh activation unit is connected to the input end of the ninth convolutional unit; the output end of the ninth convolutional unit is connected to the input end of the ninth normalization unit; the output end of the ninth normalization unit is connected to the input end of the sixth fusion unit; the output end of the sixth fusion unit is connected to the input end of the eighth activation unit; the output end of the eighth activation unit is configured to output the third module output of the first feature processing module.

[0128] It should be noted that the above third module input of the first feature processing module represents the data output from the output end of the second feature extraction module, and the third module output of the first feature processing module represents the data output from the output end of the third module of the first feature processing module, and the embodiments of the present invention are not limited thereto.

[0129] It should be noted that the convolutional kernel sizes of the above seventh convolutional unit, eighth convolutional unit, and ninth convolutional unit are 1×1, 3×3, and 1×1 respectively, and the number of channels are 4, 4, and 1 respectively, and the embodiments of the present invention are not limited thereto.

[0130] It should be noted that the above seventh normalization unit, eighth normalization unit, and ninth normalization unit are all constructed based on the batch normalization layer, and the embodiments of the present invention do not make any limitations.

[0131] It should be noted that the above sixth activation unit, seventh activation unit, and eighth activation unit are all constructed based on the RELU activation function, and the embodiments of the present invention do not make any limitations.

[0132] It should be noted that the above sixth fusion unit is constructed based on element-wise addition, and the embodiments of the present invention do not make any limitations.

[0133] It should be noted that the above first feature processing module is a feature recognition and extraction processing module that does not change the data dimension. That is, the data dimension is reduced from 256 to 64 by using the first 1×1 convolutional unit, and then restored to 256 by the 1×1 convolutional unit corresponding to the ninth convolutional unit, thereby realizing flexible reduction and increase of the data dimension, reducing the amount of data processing in the intermediate process, ensuring both the extraction of high-level abstract features and the processing efficiency of the model, and the embodiments of the present invention do not make any limitations.

[0134] It can be seen that implementing the signal recognition method described in the embodiments of the present invention is beneficial to improving the signal recognition accuracy, and further solves the problem of low recognition accuracy of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and harsh transmission environments.

[0135] Embodiment 2

[0136] Please refer to Figure 3 , Figure 3 which is a schematic structural diagram of a signal recognition device disclosed in the embodiments of the present invention. Among them, Figure 3 the described device can be applied to a management system, such as a local server or a cloud server for management, etc., and the embodiments of the present invention do not make any limitations. As Figure 3 shown, the device may include:

[0137] An acquisition module 201, configured to acquire signal information to be recognized;

[0138] A first processing module 202, configured to perform pre-processing on the signal information to be recognized to obtain target processed signal information;

[0139] A second processing module 203, configured to perform recognition processing on the target processed signal information to obtain target signal recognition result information.

[0140] It can be seen that implementing Figure 3 the described signal recognition device is beneficial to improving the signal recognition accuracy, and further solves the problem of low recognition accuracy of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and harsh transmission environments.

[0141] In another alternative embodiment, as Figure 3 shown, the target processing signal information is identified and processed to obtain target signal recognition result information, including:

[0142] Perform the first signal feature extraction on the target processing signal information to obtain the first extracted signal feature information;

[0143] Perform the second signal feature extraction on the target processing signal information to obtain the second extracted signal feature information;

[0144] Use the target signal recognition model to perform identification processing on the first extracted signal feature information and the second extracted signal feature information to obtain the target signal recognition result information.

[0145] It can be seen that implementing Figure 3 the described signal recognition device is beneficial to improving the signal recognition accuracy, and further solves the problem of low recognition accuracy of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and poor transmission environment.

[0146] In yet another alternative embodiment, as Figure 3 shown, the target signal recognition model includes a first feature dimensionality reduction module, a second feature dimensionality reduction module, a first fusion module, a first convolution module, a first feature extraction module, a second feature extraction module, a third feature extraction module, a fourth feature extraction module, a fifth feature extraction module, a first normalization module, a first activation module, a first pooling module, a first feature analysis module, a second feature analysis module, a third feature analysis module, a fourth feature analysis module, a first feature processing module, a second feature processing module, a third feature processing module, a fourth feature processing module, a fifth feature processing module, a sixth feature processing module, a seventh feature processing module, an eighth feature processing module, a ninth feature processing module, a tenth feature processing module, and an eleventh feature processing module; wherein,

[0147] The input end of the first feature dimension reduction module is configured to receive the first model input of the target signal recognition model, and the output end of the first feature dimension reduction module is connected to the input end of the first fusion module; the input end of the second feature dimension reduction module is configured to receive the second model input of the target signal recognition model, and the output end of the second feature dimension reduction module is connected to the input end of the first fusion module; the output end of the first fusion module is connected to the input end of the first convolution module; the output end of the first convolution module is connected to the input end of the first feature extraction module; the output end of the first feature extraction module is connected to the input end of the first normalization module; the output end of the first normalization module is connected to the input end of the first activation module; the output end of the first activation module is connected to the input end of the first pooling module; the output end of the first pooling module is connected to the input end of the first feature analysis module; the output end of the first feature analysis module is connected to the input end of the second feature extraction module; the output end of the second feature extraction module is connected to the input end of the first feature processing module; the output end of the first feature extraction module is connected to the input end of the second feature processing module; the output end of the second feature processing module is connected to the input end of the second feature analysis module; the output end of the second feature analysis module is connected to the input end of the third feature extraction module; the output end of the third feature extraction module is connected to the input end of the third feature processing module; the output end of the third feature processing module is connected to the input end of the fourth feature processing module; the output end of the fourth feature processing module is connected to the input end of the fifth feature processing module; the output end of the fifth feature processing module is connected to the input end of the third feature analysis module; the output end of the third feature analysis module is connected to the input end of the fourth feature extraction module; the output end of the fourth feature extraction module is connected to the input end of the sixth feature processing module; the output end of the sixth feature processing module is connected to the input end of the seventh feature processing module; the output end of the seventh feature processing module is connected to the input end of the eighth feature processing module; the output end of the eighth feature processing module is connected to the input end of the ninth feature processing module; the output end of the ninth feature processing module is connected to the input end of the tenth feature processing module; the output end of the tenth feature processing module is connected to the input end of the fourth feature analysis module; the output end of the fourth feature analysis module is connected to the input end of the fifth feature extraction module; the output end of the fifth feature extraction module is connected to the input end of the eleventh feature processing module; the output end of the eleventh feature processing module is configured to output the model output of the target signal recognition model.

[0148] It can be seen that implementing Figure 3 the described signal recognition device is beneficial to improving the signal recognition accuracy rate, and further solves the problem of low recognition accuracy rate of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and poor transmission environment.

[0149] In yet another alternative embodiment, as Figure 3As shown, the first feature dimension reduction module includes a first convolutional unit, a second convolutional unit, a first normalization unit, a second normalization unit, a first activation unit, a second activation unit, and a first fusion unit; among them,

[0150] The input end of the first convolutional unit and the input end of the first fusion unit are configured to receive the module input of the first feature dimension reduction module; the output end of the first convolutional unit is connected to the input end of the first normalization unit; the output end of the first normalization unit is connected to the input end of the first activation unit; the output end of the first activation unit is connected to the input end of the second convolutional unit; the output end of the second convolutional unit is connected to the input end of the second normalization unit; the output end of the second normalization unit is connected to the input end of the first fusion unit; the output end of the first fusion unit is connected to the input end of the second activation unit; the output end of the second activation unit is configured to output the module output of the first feature dimension reduction module.

[0151] It can be seen that implementing Figure 3 the described signal recognition device is beneficial to improving the signal recognition accuracy rate, and further solves the problem of low recognition accuracy rate of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and poor transmission environment.

[0152] In yet another alternative embodiment, as Figure 3 shown, the first feature extraction module includes a first pooling unit, a second pooling unit, a third pooling unit, a fourth pooling unit, a fifth pooling unit, a sixth pooling unit, a multi-layer perceptron, a second fusion unit, a third fusion unit, and a fourth fusion unit; among them,

[0153] The input end of the first pooling unit, the input end of the second pooling unit, and the input end of the third fusion unit are configured to receive the first module input of the first feature extraction module; the output end of the first pooling unit and the output end of the second pooling unit are both connected to the input end of the multi-layer perceptron; the output end of the multi-layer perceptron is respectively connected to the input end of the third pooling unit and the input end of the fourth pooling unit; the output end of the third pooling unit and the output end of the fourth pooling unit are both connected to the input end of the second fusion unit; the output end of the second fusion unit is connected to the input end of the third fusion unit; the output end of the third fusion unit is respectively connected to the input end of the fifth pooling unit and the input end of the fourth fusion unit; the output end of the fifth pooling unit is connected to the input end of the sixth pooling unit; the output end of the sixth pooling unit is connected to the input end of the fourth fusion unit; the output end of the fourth fusion unit is configured to output the first module output of the first feature extraction module.

[0154] It can be seen that implementing Figure 3 the described signal recognition device is beneficial to improving the signal recognition accuracy rate, and further solves the problem of low recognition accuracy rate of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and poor transmission environment.

[0155] In yet another alternative embodiment, as Figure 3 shown, the first feature analysis module includes a third convolutional unit, a fourth convolutional unit, a fifth convolutional unit, a sixth convolutional unit, a third normalization unit, a fourth normalization unit, a fifth normalization unit, a sixth normalization unit, a third activation unit, a fourth activation unit, a fifth activation unit, and a fifth fusion unit; wherein,

[0156] The input ends of the third convolutional unit and the sixth convolutional unit are both configured to receive the second module input of the first feature analysis module; the output end of the third convolutional unit is connected to the input end of the third normalization unit; the output end of the third normalization unit is connected to the input end of the third activation unit; the output end of the third activation unit is connected to the input end of the fourth convolutional unit; the output end of the fourth convolutional unit is connected to the input end of the fourth normalization unit; the output end of the fourth normalization unit is connected to the input end of the fourth activation unit; the output end of the fourth activation unit is connected to the input end of the fifth convolutional unit; the output end of the fifth convolutional unit is connected to the input end of the fifth normalization unit; the output end of the fifth normalization unit is connected to the input end of the fifth fusion unit; the output end of the sixth convolutional unit is connected to the input end of the sixth normalization unit; the output end of the sixth normalization unit is connected to the input end of the fifth fusion unit; the output end of the fifth fusion unit is connected to the input end of the fifth activation unit; the output end of the fifth activation unit is configured to output the second module output of the first feature analysis module.

[0157] It can be seen that implementing Figure 3 the described signal recognition device is beneficial to improving the signal recognition accuracy rate, and further solves the problem of low recognition accuracy rate of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and harsh transmission environments.

[0158] In yet another alternative embodiment, as Figure 3 shown, the first feature processing module includes a seventh convolutional unit, an eighth convolutional unit, a ninth convolutional unit, a seventh normalization unit, an eighth normalization unit, a ninth normalization unit, a sixth activation unit, a seventh activation unit, an eighth activation unit, and a sixth fusion unit; wherein,

[0159] The input end of the seventh convolutional unit and the input end of the sixth fusion unit are configured to receive the third module input of the first feature processing module; the output end of the seventh convolutional unit is connected to the input end of the seventh normalization unit; the output end of the seventh normalization unit is connected to the input end of the sixth activation unit; the output end of the sixth activation unit is connected to the input end of the eighth convolutional unit; the output end of the eighth convolutional unit is connected to the input end of the eighth normalization unit; the output end of the eighth normalization unit is connected to the input end of the seventh activation unit; the output end of the seventh activation unit is connected to the input end of the ninth convolutional unit; the output end of the ninth convolutional unit is connected to the input end of the ninth normalization unit; the output end of the ninth normalization unit is connected to the input end of the sixth fusion unit; the output end of the sixth fusion unit is connected to the input end of the eighth activation unit; the output end of the eighth activation unit is configured to output the third module output of the first feature processing module.

[0160] It can be seen that implementing Figure 3 the described signal recognition device is beneficial to improving the signal recognition accuracy rate, and further solves the problem of low recognition accuracy rate of specific protocol signals in complex electromagnetic environments such as low signal-to-noise ratio and poor transmission environment.

[0161] Embodiment 3

[0162] Please refer to Figure 4 , Figure 4 which is a schematic structural diagram of another signal recognition device disclosed in the embodiments of the present invention. Among them, Figure 4 the described device can be applied to a management system, such as a local server or a cloud server for management, etc., which is not limited in the embodiments of the present invention. As Figure 4 shown, the device may include:

[0163] A memory 301 storing executable program code;

[0164] A processor 302 coupled to the memory 301;

[0165] The processor 302 calls the executable program code stored in the memory 301 to execute the steps in the signal recognition method described in Embodiment 1.

[0166] Embodiment 4

[0167] The embodiments of the present invention disclose a computer-readable storage medium, which stores a computer program for electronic data exchange. Among them, the computer program enables a computer to execute the steps in the signal recognition method described in Embodiment 1.

[0168] Embodiment 5

[0169] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute the steps in the signal recognition method described in Embodiment 1.

[0170] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative labor.

[0171] Through the specific descriptions of the above embodiments, those skilled in the art can clearly understand that each implementation can be realized by means of software plus a necessary general hardware platform, and of course, it can also be realized by hardware. Based on such an understanding, the above technical solution, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electrically-erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc memories, a magnetic disk memory, a tape memory, or any other computer-readable medium capable of carrying or storing data.

[0172] Finally, it should be noted that what is disclosed in a signal recognition method and device disclosed in the embodiments of the present invention is only the preferred embodiments of the present invention, and is only used to illustrate the technical solutions of the present invention, rather than limiting it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A signal recognition method, characterized in that: The method comprises: Obtaining information of a signal to be identified; Pre-processing the signal information to be identified to obtain target processed signal information; The target processing signal information is identified and processed to obtain target signal identification result information.

2. The signal recognition method according to claim 1, characterized in that: The identifying and processing the target processing signal information to obtain target signal identification result information includes: Performing a first signal feature extraction on the target processed signal information to obtain first extracted signal feature information; Performing a second signal feature extraction on the target processed signal information to obtain second extracted signal feature information; The first extracted signal feature information and the second extracted signal feature information are identified and processed using a target signal identification model to obtain target signal identification result information.

3. The signal recognition method according to claim 2, characterized in that: The target signal recognition model includes a first feature dimension reduction module, a second feature dimension reduction module, a first fusion module, a first convolution module, a first feature extraction module, a second feature extraction module, a third feature extraction module, a fourth feature extraction module, a fifth feature extraction module, a first normalization module, a first activation module, a first pooling module, a first feature analysis module, a second feature analysis module, a third feature analysis module, a fourth feature analysis module, a first feature processing module, a second feature processing module, a third feature processing module, a fourth feature processing module, a fifth feature processing module, a sixth feature processing module, a seventh feature processing module, an eighth feature processing module, a ninth feature processing module, a tenth feature processing module and an eleventh feature processing module; wherein, The input end of the first feature dimension reduction module is configured to receive the first model input of the target signal recognition model, and the output end of the first feature dimension reduction module is connected to the input end of the first fusion module; the input end of the second feature dimension reduction module is configured to receive the second model input of the target signal recognition model, and the output end of the second feature dimension reduction module is connected to the input end of the first fusion module; the output end of the first fusion module is connected to the input end of the first convolution module; the output end of the first convolution module is connected to the input end of the first feature extraction module; the output end of the first feature extraction module is connected to the input end of the first normalization module; the output end of the first normalization module is connected to the input end of the first activation module; the output end of the first activation module is connected to the input end of the first pooling module; the output end of the first pooling module is connected to the input end of the first feature analysis module; the output end of the first feature analysis module is connected to the input end of the second feature extraction module; the output end of the second feature extraction module is connected to the input end of the first feature processing module; the output end of the first feature extraction module is connected to the input end of the second feature processing module; the output end of the second feature processing module is connected to the input end of the second feature analysis module; the output end of the second feature analysis module is connected to the third feature The output end of the third feature extraction module is connected to the input end of the third processing module; the output end of the third feature processing module is connected to the input end of the fourth feature processing module; the output end of the fourth feature processing module is connected to the input end of the fifth feature processing module; the output end of the fifth feature processing module is connected to the input end of the third feature analysis module; the output end of the third feature analysis module is connected to the input end of the fourth feature extraction module; the output end of the fourth feature extraction module is connected to the input end of the sixth feature processing module; the output end of the sixth feature processing module is connected to the input end of the seventh feature processing module; The output end of the seventh feature processing module is connected to the input end of the eighth feature processing module; the output end of the eighth feature processing module is connected to the input end of the ninth feature processing module; the output end of the ninth feature processing module is connected to the input end of the tenth feature processing module; the output end of the tenth feature processing module is connected to the input end of the fourth feature analysis module; the output end of the fourth feature analysis module is connected to the input end of the fifth feature extraction module; the output end of the fifth feature extraction module is connected to the input end of the eleventh feature processing module; the output end of the eleventh feature processing module is configured to output the model output of the target signal recognition model.

4. The signal recognition method according to claim 3, characterized in that: The first feature dimension reduction module includes a first convolution unit, a second convolution unit, a first normalization unit, a second normalization unit, a first activation unit, a second activation unit and a first fusion unit; wherein, The input end of the first convolution unit and the input end of the first fusion unit are configured to receive the module input of the first feature dimension reduction module; the output end of the first convolution unit is connected to the input end of the first normalization unit; the output end of the first normalization unit is connected to the input end of the first activation unit; the output end of the first activation unit is connected to the input end of the second convolution unit; the output end of the second convolution unit is connected to the input end of the second normalization unit; the output end of the second normalization unit is connected to the input end of the first fusion unit; the output end of the first fusion unit is connected to the input end of the second activation unit; the output end of the second activation unit is configured to output the module output of the first feature dimension reduction module.

5. The signal recognition method according to claim 3, characterized in that: The first feature extraction module includes a first pooling unit, a second pooling unit, a third pooling unit, a fourth pooling unit, a fifth pooling unit, a sixth pooling unit, a multi-layer perceptron, a second fusion unit, a third fusion unit and a fourth fusion unit; wherein, The input end of the first pooling unit, the input end of the second pooling unit and the input end of the third fusion unit are configured to receive the first module input of the first feature extraction module; the output end of the first pooling unit and the output end of the second pooling unit are both connected to the input end of the multi-layer perceptron; the output end of the multi-layer perceptron is respectively connected to the input end of the third pooling unit and the input end of the fourth pooling unit; the output end of the third pooling unit and the output end of the fourth pooling unit are both connected to the input end of the second fusion unit; the output end of the second fusion unit is connected to the input end of the third fusion unit; the output end of the third fusion unit is respectively connected to the input end of the fifth pooling unit and the input end of the fourth fusion unit; the output end of the fifth pooling unit is connected to the input end of the sixth pooling unit; the output end of the sixth pooling unit is connected to the input end of the fourth fusion unit; the output end of the fourth fusion unit is configured to output the first module output of the first feature extraction module.

6. The signal recognition method according to claim 3, characterized in that: The first feature analysis module includes a third convolution unit, a fourth convolution unit, a fifth convolution unit, a sixth convolution unit, a third normalization unit, a fourth normalization unit, a fifth normalization unit, a sixth normalization unit, a third activation unit, a fourth activation unit, a fifth activation unit and a fifth fusion unit; wherein, The input end of the third convolution unit and the input end of the sixth convolution unit are both configured to receive the second module input of the first feature analysis module; the output end of the third convolution unit is connected to the input end of the third normalization unit; the output end of the third normalization unit is connected to the input end of the third activation unit; the output end of the third activation unit is connected to the input end of the fourth convolution unit; the output end of the fourth convolution unit is connected to the input end of the fourth normalization unit; the output end of the fourth normalization unit is connected to the input end of the fourth activation unit; the output end of the fourth activation unit is connected to the input end of the fifth convolution unit; the output end of the fifth convolution unit is connected to the input end of the fifth normalization unit; the output end of the fifth normalization unit is connected to the input end of the fifth fusion unit; the output end of the sixth convolution unit is connected to the input end of the sixth normalization unit; the output end of the sixth normalization unit is connected to the input end of the fifth fusion unit; the output end of the fifth fusion unit is connected to the input end of the fifth activation unit; the output end of the fifth activation unit is configured to output the second module output of the first feature analysis module.

7. The signal recognition method according to claim 3, characterized in that: The first feature processing module includes a seventh convolution unit, an eighth convolution unit, a ninth convolution unit, a seventh normalization unit, an eighth normalization unit, a ninth normalization unit, a sixth activation unit, a seventh activation unit, an eighth activation unit and a sixth fusion unit; wherein, The input end of the seventh convolution unit and the input end of the sixth fusion unit are configured to receive the third module input of the first feature processing module; the output end of the seventh convolution unit is connected to the input end of the seventh normalization unit; the output end of the seventh normalization unit is connected to the input end of the sixth activation unit; the output end of the sixth activation unit is connected to the input end of the eighth convolution unit; the output end of the eighth convolution unit is connected to the input end of the eighth normalization unit; the output end of the eighth normalization unit is connected to the input end of the seventh activation unit; the output end of the seventh activation unit is connected to the input end of the ninth convolution unit; the output end of the ninth convolution unit is connected to the input end of the ninth normalization unit; the output end of the ninth normalization unit is connected to the input end of the sixth fusion unit; the output end of the sixth fusion unit is connected to the input end of the eighth activation unit; the output end of the eighth activation unit is configured to output the third module output of the first feature processing module.

8. A signal recognition device, characterized in that: The device comprises: An acquisition module, used for acquiring information of a signal to be identified; A first processing module, used for pre-processing the signal information to be identified to obtain target processed signal information; The second processing module is used to perform identification processing on the target processing signal information to obtain target signal identification result information.

9. A signal recognition device, characterized in that: The device comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the signal recognition method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the signal recognition method according to any one of claims 1 to 7.

Images