Short-wave protocol identification method, device and equipment based on multi-head attention network, and storage medium

Abstract

The invention discloses a short-wave protocol identification method based on a multi-head attention network, and the method comprises the steps: S1, obtaining short-wave protocol data for training, and training a classification model based on the multi-head attention network through the short-wave protocol data for training; s2, acquiring short-wave protocol data to be identified, and performing data preprocessing on the data to obtain data in a standard format; s3, the data in the standard format is unitized according to a preset interval, and a unitized sequence is generated; s4, inputting the unitized sequence into the trained classification model based on the multi-head attention network to obtain an output sequence; and S5, according to the value of the category unit in the output sequence, obtaining the probability of dividing the to-be-identified short-wave protocol data into various protocols, wherein the category corresponding to the highest probability is the category of the to-be-identified data. The invention further discloses a shortwave protocol recognition device based on the multi-head attention network, corresponding equipment and a storage medium. According to the invention, the short-wave protocol type identification accuracy can be further improved.

Description

Technical Field

[0001] The present application relates to the field of deep learning technology, and more specifically, to a shortwave protocol recognition method, device, equipment and storage medium based on a multi-head attention network. Background Art

[0002] In the task of shortwave type classification based on intelligent algorithm models, the key to signal classification lies in the characteristic analysis of shortwave protocol signals. When transmitting, shortwave signals will pre-send a preamble sequence (header) containing fixed code element information and other characteristic waveforms for the receiver's capture and synchronization. Generally, it is in the front part of the data frame, which can play the role of frame synchronization, symbol synchronization, and carrier synchronization. Different types of signals have different characteristic waveforms. Specific signals with different modulation methods show different characteristics on the characteristic diagram, and specific signals with the same modulation information but different data frame structures will also produce different characteristics.

[0003] Since the main distinguishing features of shortwave signals are concentrated in the signal waveform and frame structure, and the classification models in the existing technology usually only focus on the shape characteristics of the target and ignore the contextual correlation between the signal frame structures, it is impossible to extract the cross-combination information hidden between different features in the shortwave protocol data with ever-changing situations. Summary of the invention

[0004] In response to at least one defect or improvement need in the prior art, the present invention provides a shortwave protocol recognition method, apparatus, device and storage medium based on a multi-head attention network, which can solve at least one of the problems existing in the above-mentioned background technology.

[0005] To achieve the above object, according to a first aspect of the present invention, a shortwave protocol recognition method based on a multi-head attention network is provided, the method comprising:

[0006] S1 obtains shortwave protocol data for training, and uses the shortwave protocol data for training to train a classification model based on a multi-head attention network, wherein the shortwave protocol data is a baseband burst signal, and the classification model is used to distinguish different shortwave protocol types according to a signal frame format and a waveform context relationship;

[0007] S2 obtains the shortwave protocol data to be identified, and pre-processes the data to obtain data in a standard format;

[0008] S3: Unitize the data in the standard format according to preset intervals, fill in the gaps if there is less than one unit, and generate a unitized sequence;

[0009] S4 inputs the unitized sequence into a trained classification model based on a multi-head attention network to obtain an output sequence;

[0010] S5 obtains the probability of the shortwave protocol data to be identified being classified into various protocols according to the value of the category unit in the output sequence, wherein the corresponding category with the highest probability is the category of the data to be identified.

[0011] Furthermore, in the above-mentioned shortwave protocol recognition method based on a multi-head attention network, the classification model based on a multi-head attention network specifically includes a category encoding layer, a position encoding layer, a normalization layer and a network layer connected in series;

[0012] The category encoding layer is used to add category units at the front of the input sequence to focus the model's sequence output on the protocol category;

[0013] The position encoding layer is used to capture the position information of the protocol frame structure in the input sequence;

[0014] The normalization layer is used to normalize the data transmitted in the network to data with a mean of 0 and a variance of 1;

[0015] The network layer is composed of 12 attention network layers connected in series. Each attention network layer contains a multi-head attention layer and two fully connected layers, which are connected in a residual manner to learn the differences in protocol categories.

[0016] Furthermore, the above-mentioned shortwave protocol recognition method based on multi-head attention network, wherein the data is subjected to data preprocessing, specifically comprising:

[0017] intercepting a synchronization header sequence and a portion of a data symbol sequence containing key features of the protocol;

[0018] Convert real domain data into I / Q domain data;

[0019] The protocol data with different signal strengths were normalized to data with energy of 1.

[0020] Furthermore, the shortwave protocol recognition method based on the multi-head attention network described above, wherein the data in the standard format is unitized at preset intervals, and the gaps that are less than one unit are filled, specifically includes:

[0021] The data is divided into Floor(S / (L)) units, where the length of each unit is L and the total length of the input data is S. The remaining data that is less than one unit is filled with 0 to a length of L.

[0022] Furthermore, in the above-mentioned shortwave protocol recognition method based on multi-head attention network, the shortwave protocol data includes a communication bandwidth support of the protocol in the range of 3MHz-30MHz.

[0023] Furthermore, in the above-mentioned shortwave protocol recognition method based on a multi-head attention network, the shortwave protocol data includes a signal bandwidth support of the protocol of one or more of 3kHz, 6kHz, 9kHz or 12kHz.

[0024] Furthermore, in the above-mentioned shortwave protocol recognition method based on a multi-head attention network, the shortwave protocol data includes a signal modulation method that supports one or more of a PSK modulation type, a FSK modulation type, a multi-tone type or an AM type.

[0025] According to a second aspect of the present invention, there is also provided a shortwave protocol recognition device based on a multi-head attention network, comprising:

[0026] A model training module, used to obtain shortwave protocol data for training, and use the shortwave protocol data for training to train a classification model based on a multi-head attention network, wherein the shortwave protocol data is a baseband burst signal, and the classification model is used to distinguish different shortwave protocol types according to a signal frame format and a waveform context relationship;

[0027] A preprocessing module is used to obtain the shortwave protocol data to be identified and preprocess the data to obtain data in a standard format;

[0028] A unitization module, used to unitize the data in the standard format according to preset intervals, fill in the gaps where there is less than one unit, and generate a unitized sequence;

[0029] A first acquisition module is used to input the unitized sequence into a trained classification model based on a multi-head attention network to obtain an output sequence;

[0030] The second acquisition module is used to obtain the probability that the shortwave protocol data to be identified is classified into various protocols according to the value of the category unit in the output sequence, wherein the corresponding category with the highest probability is the category of the data to be identified.

[0031] According to the third aspect of the present invention, there is also provided a shortwave protocol recognition device based on a multi-head attention network, which comprises at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program, and when the computer program is executed by the processing unit, the processing unit executes the steps of any one of the above methods.

[0032] According to the fourth aspect of the present invention, a storage medium is also provided, which stores a computer program that can be executed by a shortwave protocol recognition device based on a multi-head attention network. When the computer program runs on the shortwave protocol recognition device based on the multi-head attention network, the shortwave protocol recognition device based on the multi-head attention network executes the steps of any one of the above-mentioned methods.

[0033] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0034] The shortwave protocol recognition method based on a multi-head attention network provided by the present invention is based on a multi-head attention network. A small amount of information reflecting the protocol characteristics is screened out from a large amount of transmission information of the protocol signal through the attention mechanism, thereby improving the recognition ability of the model and selecting sequence features as the input of the model, which is consistent with the selected model that is good at capturing the internal correlation of the sequence, thereby further improving the accuracy of protocol recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 A flowchart of a shortwave protocol recognition method based on a multi-head attention network provided in an embodiment of the present application;

[0037] Figure 2 A schematic diagram of an application scenario of a method for identifying a shortwave protocol based on a multi-head attention network provided in an embodiment of the present application;

[0038] Figure 3 A schematic diagram of the composition of a classification model of a shortwave protocol recognition method based on a multi-head attention network provided in an embodiment of the present application;

[0039] Figure 4 A schematic diagram of the training process of the multi-head attention network model provided in the embodiment of the present application;

[0040] Figure 5 A schematic diagram of the structure of a shortwave protocol recognition device based on a multi-head attention network provided in an embodiment of the present application. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0042] The terms "first", "second", "third", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0043] Figure 1 A flow chart of a shortwave protocol recognition method based on a multi-head attention network provided in an embodiment of the present application is shown as follows: Figure 1 As shown, a shortwave protocol recognition method based on a multi-head attention network provided in an embodiment of the present application includes:

[0044] S1 obtains shortwave protocol data for training, and uses the shortwave protocol data for training to train a classification model based on a multi-head attention network. The shortwave protocol data is a baseband burst signal, and the classification model is used to distinguish different shortwave protocol types according to the relationship between the signal frame format and the waveform context;

[0045] S2 obtains the shortwave protocol data to be identified, and pre-processes the data to obtain data in a standard format;

[0046] S3 unitizes the data in the standard format according to the preset intervals, fills in the gaps that are less than one unit, and generates a unitized sequence;

[0047] S4 inputs the unitized sequence into the trained classification model based on the multi-head attention network to obtain the output sequence;

[0048] S5 obtains the probability of the shortwave protocol data to be identified being classified into various protocols according to the value of the category unit in the output sequence, wherein the corresponding category with the highest probability is the category of the data to be identified.

[0049] Specifically, considering the characteristics of the protocol type recognition task, shortwave protocol recognition can be considered from two directions. In terms of model selection, the protocol recognition model based on intelligent algorithms should have the ability to learn global features, analyze timing, and pay attention to local frame structures. In terms of feature selection, the features of the input model should reflect both the time-frequency information of the signal and the contextual information of the signal between frame structures.

[0050] In terms of the model, a classification model based on a multi-head attention network is selected to avoid the model over-focusing on its own position when encoding the information of the current position. By integrating knowledge from multiple attention convergences, the model pays more attention to the key part of the sequence for the classification of the shortwave protocol, and can have a higher model expression ability and classification accuracy.

[0051] Shortwave protocol data for training is obtained, and a classification model based on a multi-head attention network is trained based on the shortwave protocol data for training, so that the model can distinguish different shortwave protocol types based on the input shortwave protocol data to be identified according to the signal frame format and waveform context relationship.

[0052] In terms of feature selection, the shortwave protocol data to be identified is displayed using its time-frequency diagram as follows Figure 2 As shown, the actual processing in the processing process is to process the sequence in the time domain, not using the time-frequency diagram.

[0053] The shortwave protocol data to be identified needs to be preprocessed before passing through the attention network. Although the communication duration of different shortwave communication protocols is different, the part with obvious distinguishing features is the part containing the synchronization frame header. The frame header area of ​​the shortwave protocol sample to be identified is intercepted. Optionally, in order to improve the calculation efficiency, the first 2s of the sample to be identified can be selected as the feature.

[0054] In order to eliminate the influence of the carrier frequency and center frequency of the shortwave protocol data to be identified on the classification of the protocol category, the real shortwave protocol data is converted into a baseband I / Q signal through Hilbert transform, and its energy is normalized.

[0055] The shortwave protocol data to be identified after the above preprocessing is passed through the trained classification model based on the multi-head attention network. The probability of the shortwave protocol data to be identified being divided into various protocols is obtained according to the value of the category unit in the output sequence. The corresponding category with the highest probability is the category of the data to be identified, and the protocol category of the shortwave protocol data to be identified is output.

[0056] The shortwave protocol recognition method based on a multi-head attention network provided in the embodiment of the present application is based on a multi-head attention network. A small amount of information reflecting the protocol characteristics is screened out from a large amount of transmission information of the protocol signal through the attention mechanism, thereby improving the recognition ability of the model and selecting sequence features as the input of the model, which is consistent with the selected model that is good at capturing the internal correlation of the sequence, thereby further improving the accuracy of protocol recognition.

[0057] Optionally, the shortwave protocol recognition method based on a multi-head attention network provided in an embodiment of the present application is a classification model based on a multi-head attention network, specifically including a category encoding layer, a position encoding layer, a normalization layer and a network layer connected in series;

[0058] The category encoding layer is used to add category units to the front of the input sequence and focus the model's sequence output on the protocol category;

[0059] The position encoding layer is used to capture the position information of the protocol frame structure in the input sequence;

[0060] The normalization layer is used to normalize the data transmitted in the network to data with a mean of 0 and a variance of 1;

[0061] The network layer is composed of 12 attention network layers connected in series. Each attention network layer contains a multi-head attention layer and two fully connected layers, which are connected in a residual way to learn the differences in protocol categories.

[0062] Specifically, Figure 3 A schematic diagram of the classification model composition of the method for shortwave protocol recognition based on a multi-head attention network provided in an embodiment of the present application is shown in FIG. Figure 3 As shown in Figure 2, the input of the model is the unitized shortwave protocol data. Divided into T units (Token), i.e. tensors Where S is the input data length, B is the batch reading size, and L is the length of each unit. The tensor I is directly input into the model structure.

[0063] The model includes category encoding layer, position encoding layer, normalization layer and network layer.

[0064] The category encoding layer is used to add category units at the front of the input sequence to focus the model's sequence output on the protocol category. The category encoding layer encodes the waveform category into a category unit in the input unit sequence, which plays the role of focusing the model's sequence output on the protocol signal category. The value of the category unit is initialized randomly, and its length is the same as the length of each unit in the input unit sequence, that is, the category unit is Splicing the category unit to the starting position of the input tensor is the output of the category encoding layer:

[0065]

[0066] Cat is a sequence splicing method.

[0067] The position encoding layer is used to capture the position information of the protocol frame structure in the input sequence. The position encoding layer maps each unit to a vector containing the unit position information. In this way, the model can learn both the unit content information and the unit position information. The output of the position encoding layer is:

[0068]

[0069] where ft () is the vector mapping function, and t is the unit sequence index.

[0070] The normalization layer is used to normalize the data transmitted in the network to data with a mean of 0 and a variance of 1, which is beneficial for model training.

[0071] The network layer consists of 12 attention network layers connected in series. Each attention network layer contains a multi-head attention layer and two fully connected layers, which are connected by normalization and residual to learn the differences in protocol categories.

[0072] I A =I+LayerNormal(AN(I))

[0073] Out=I A +LayerNormal(F2(ReLu(F1(I A ))))

[0074] Among them, AN(I) is the output of the multi-head attention layer, Out is the output of each attention network layer, F1 and F I They are the first and second layers of the fully connected layer respectively.

[0075] In the output layer, the output of the multi-head attention network model is The category unit stores the category information of the sequence, and the category unit is extracted separately: After the fully connected layer and the softmax layer, the probability of being classified into various protocol labels is obtained.

[0076] Optionally, the shortwave protocol recognition method based on a multi-head attention network provided in an embodiment of the present application preprocesses the data, specifically including:

[0077] intercepting a synchronization header sequence and a portion of a data symbol sequence containing key features of the protocol;

[0078] Convert real domain data into I / Q domain data;

[0079] The protocol data with different signal strengths were normalized to data with energy of 1.

[0080] Optionally, the shortwave protocol recognition method based on the multi-head attention network provided in the embodiment of the present application unitizes the data in the standard format according to a preset interval, and fills in the gaps that are less than one unit, specifically including:

[0081] The data is divided into Floor(S / L) units, where the length of each unit is L, the total length of the input data is S, and the remaining data that is less than one unit is filled with 0 to a length of L.

[0082] Optionally, in the shortwave protocol recognition method based on a multi-head attention network provided in an embodiment of the present application, the shortwave protocol data includes a communication bandwidth support of the protocol in the range of 3MHz-30MHz.

[0083] Optionally, in the shortwave protocol recognition method based on a multi-head attention network provided in an embodiment of the present application, the shortwave protocol data includes a signal bandwidth support of the protocol of one or more of 3kHz, 6kHz, 9kHz or 12kHz.

[0084] Optionally, in the shortwave protocol recognition method based on a multi-head attention network provided in an embodiment of the present application, the shortwave protocol data includes a signal modulation method that supports one or more of PSK modulation type, FSK modulation type, multi-tone type or AM type.

[0085] Figure 4 A schematic diagram of the training process of the multi-head attention network model provided in the embodiment of the present application. The training process of the multi-head attention network model provided in the embodiment of the present application includes:

[0086] Data preparation: The training data used is real data collected and processed by shortwave receivers and manually annotated. The protocol types included include 2GALE, 3GALE, ANDVT, CLOVER2000, HF6000, LINK11-CLEW, LINK11-SLEW, LINK22, M110A, M110B, M110C, M110C (wide), S4285, Unknow, and Voice. The data is divided into training set and validation set in the form of 2:8.

[0087] Model initialization: Model sampling is the same as Figure 3 The classification model shown is based on the multi-head attention network. The internal parameters of the model are initialized.

[0088] Training termination judgment: Set the number of training rounds to 300, update the model once in each training round and calculate the average recognition accuracy on the validation set.

[0089] Training results: A model is saved for each round of training, and the model with the highest statistical average recognition accuracy is selected as the result model.

[0090] Figure 5 A schematic diagram of the structure of a shortwave protocol recognition device based on a multi-head attention network provided in an embodiment of the present application is shown in FIG. Figure 5 As shown, the present application also provides a shortwave protocol recognition device based on a multi-head attention network, comprising:

[0091] A model training module is used to obtain shortwave protocol data for training, and use the shortwave protocol data for training to train a classification model based on a multi-head attention network. The shortwave protocol data is a baseband burst signal, and the classification model is used to distinguish different shortwave protocol types according to the relationship between the signal frame format and the waveform context;

[0092] The preprocessing module is used to obtain the shortwave protocol data to be identified and obtain the data in a standard format through data preprocessing;

[0093] The unitization module is used to unitize the data in the standard format according to the preset intervals, fill in the gaps that are less than one unit, and generate a unitized sequence;

[0094] The first acquisition module is used to input the unitized sequence into the trained classification model based on the multi-head attention network to obtain the output sequence;

[0095] The second acquisition module is used to obtain the probability of the shortwave protocol data to be identified being classified into various protocols according to the value of the category unit in the output sequence, wherein the corresponding category with the highest probability is the category of the data to be identified.

[0096] The present application also provides a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the steps of the above method are implemented. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a micro drive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0097] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0098] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0099] In the several embodiments provided in the present application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of the units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0100] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0101] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0102] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a memory, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned memory includes: U disk, read-only memory (ROM), random access memory (RAM), mobile hard disk, disk or optical disk and other media that can store program code.

[0103] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable memory, which can include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0104] The above is only an exemplary embodiment of the present disclosure, and the scope of the present disclosure cannot be limited thereto. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the specification and practicing the disclosure here, those skilled in the art will easily think of the implementation scheme of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field not recorded in the present disclosure. The description and examples are regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

[0105] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0106] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A shortwave protocol recognition method based on a multi-head attention network, characterized in that: include: S1 obtains shortwave protocol data for training, and uses the shortwave protocol data for training to train a classification model based on a multi-head attention network, wherein the shortwave protocol data is a baseband burst signal, and the classification model is used to distinguish different shortwave protocol types according to a signal frame format and a waveform context relationship; S2 obtains the shortwave protocol data to be identified, and pre-processes the data to obtain data in a standard format; S3: Unitize the data in the standard format according to preset intervals, fill in the gaps if there is less than one unit, and generate a unitized sequence; S4 inputs the unitized sequence into a trained classification model based on a multi-head attention network to obtain an output sequence; S5 obtains the probability of the shortwave protocol data to be identified being classified into various protocols according to the value of the category unit in the output sequence, wherein the corresponding category with the highest probability is the category of the data to be identified.

2. The shortwave protocol recognition method based on a multi-head attention network as claimed in claim 1, characterized in that: The classification model based on the multi-head attention network specifically includes a serially connected category encoding layer, a position encoding layer, a normalization layer and a network layer; The category encoding layer is used to add category units at the front of the input sequence to focus the model's sequence output on the protocol category; The position encoding layer is used to capture the position information of the protocol frame structure in the input sequence; The normalization layer is used to normalize the data transmitted in the network to data with a mean of 0 and a variance of 1; The network layer is composed of 12 attention network layers connected in series. Each attention network layer contains a multi-head attention layer and two fully connected layers, which are connected in a residual manner to learn the differences in protocol categories.

3. The shortwave protocol recognition method based on a multi-head attention network as claimed in claim 1, characterized in that: The data is subjected to data preprocessing, specifically comprising: intercepting a synchronization header sequence and a portion of a data symbol sequence containing key features of the protocol; Convert real domain data into I / Q domain data; The protocol data with different signal strengths were normalized to data with energy of 1.

4. The shortwave protocol recognition method based on a multi-head attention network as claimed in claim 1, characterized in that: The data in the standard format is unitized at preset intervals, and the gaps that are less than one unit are filled, specifically including: The data is divided into Floor(S / L) units, where the length of each unit is L, the total length of the input data is S, and the remaining data that is less than one unit is filled with 0 to a length of L.

5. The shortwave protocol recognition method based on a multi-head attention network as claimed in claim 1, characterized in that: The shortwave protocol data includes a communication bandwidth support of the protocol in the range of 3 MHz to 30 MHz.

6. The shortwave protocol recognition method based on a multi-head attention network as claimed in claim 1, characterized in that: The shortwave protocol data includes a signal bandwidth of the protocol that supports one or more of 3kHz, 6kHz, 9kHz or 12kHz.

7. The shortwave protocol recognition method based on a multi-head attention network as claimed in claim 1, characterized in that: The shortwave protocol data includes a signal modulation method that supports one or more of PSK modulation type, FSK modulation type, multi-tone type or AM type.

8. A shortwave protocol recognition device based on a multi-head attention network, characterized in that: include: A model training module, used to obtain shortwave protocol data for training, and use the shortwave protocol data for training to train a classification model based on a multi-head attention network, wherein the shortwave protocol data is a baseband burst signal, and the classification model is used to distinguish different shortwave protocol types according to a signal frame format and a waveform context relationship; A preprocessing module is used to obtain the shortwave protocol data to be identified and preprocess the data to obtain data in a standard format; A unitization module, used to unitize the data in the standard format according to preset intervals, fill in the gaps if there is less than one unit, and generate a unitized sequence; A first acquisition module is used to input the unitized sequence into a trained classification model based on a multi-head attention network to obtain an output sequence; The second acquisition module is used to obtain the probability that the shortwave protocol data to be identified is classified into various protocols according to the value of the category unit in the output sequence, wherein the corresponding category with the highest probability is the category of the data to be identified.

9. A shortwave protocol recognition device based on a multi-head attention network, characterized in that: The method comprises at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program, and when the computer program is executed by the processing unit, the processing unit executes the steps of the method according to any one of claims 1 to 7.

10. A storage medium, characterized in that: It stores a computer program that can be executed by a shortwave protocol recognition device based on a multi-head attention network. When the computer program runs on the shortwave protocol recognition device based on the multi-head attention network, the shortwave protocol recognition device based on the multi-head attention network executes the steps of the method described in any one of claims 1 to 7.

Images