Portable Intelligent Signal Processing System and Working Method for Rapid Detection and Identification of Communication Signals

Abstract

Disclosed is a portable intelligent signal processing system applicable to rapid detection and recognition of communication signals, which includes two parts: a portable computer hardware device and a host computer software system; the portable computer hardware device adopts a CPU+NPU computing architecture; the portable computer hardware device also includes a memory, an SSD hard disk, a battery, an LED display, and an IO interface; the host computer software system includes a data import / export module, a preprocessing module, a neural network deployment module, an intelligent inference module, a storage and data management module, and a human-computer interaction module. The present invention introduces intelligent signal processing technology, constructs a data set through fine signal analysis, recognition, and annotation of offline signals; on this basis, trains a deep network to endow the existing analysts' capabilities to the deep neural network; can complete rapid analysis and recognition of communication reconnaissance signals, realize the knowledge transfer and iterative improvement of manual offline analysis, and effectively improve the intelligence generation efficiency.

Description

Technical Field

[0001] The present invention relates to intelligent signal processing technology and application systems, belonging to the technical fields of communication countermeasure and electromagnetic spectrum management, and specifically relates to a portable intelligent signal processing system and working method suitable for rapid detection and identification of communication signals. Background Art

[0002] Modern information-based scenarios rely extremely on communication means. As carriers for air traffic control, transmitting voice commands, flight command, situation transmission, and transmission of other important parameter information, communication signals play an extremely important role. In the current communication process, manual processing and verification are required for each link, resulting in very poor timeliness and inability to meet the requirements of information security. On the other hand, when communication signals are subject to significant noise interference or the receiving distance is far, the quality of the collected signals will seriously decline, and it is also not easy to extract and effectively identify signal characteristics through manual means at this time.

[0003] Adopting artificial intelligence technology is an important way to overcome the above-mentioned difficulties and achieve rapid signal recognition in complex spectrum environments. DARPA announced the launch of the "Radio Frequency Machine Learning System RFMLS" project in August 2017, aiming to study the composition of the radio spectrum using the capabilities of machine learning systems, including understanding the types of signals occupying the spectrum, distinguishing important signals from background signals, identifying signals that do not conform to the rules, and using such systems to distinguish radio frequency signals from Internet of Things devices and distinguish these signals from signals attempting to invade these devices.

[0004] There are many literatures on using neural network-based methods to detect and identify communication signals, but specific engineering implementations are not involved. There are also few literatures available for reference on the development of corresponding portable intelligent signal processing platforms. Summary of the Invention

[0005] Aiming at problems such as high manual participation and insufficient signal detection and identification performance in existing communication signal intelligence reconnaissance, the present invention proposes a portable intelligent signal processing system suitable for rapid detection and identification of communication signals, including two parts: a portable computer hardware device and a host computer software system; the portable computer hardware device adopts a CPU+NPU computing architecture; the portable computer hardware device also includes a memory, an SSD hard disk, a battery, an LED display, and an IO interface; specifically as follows:

[0006] The host computer software system includes a data import / export module 1, a preprocessing module 2, a neural network deployment module 3, an intelligent reasoning module 4, a storage and data management module 5, and a human-computer interaction module 6; specifically as follows:

[0007] (1) Data import / export module 1

[0008] The data import / export module 1 receives the data import interaction instruction from the human-machine interaction module 6 as input, and imports the actual acquisition signal data from external devices such as the ground storage array and the mission display and control computer into the system through interfaces such as USB / gigabit Ethernet / 10-gigabit optical port; the data import / export module 1 transfers the file path and associated structure of the imported data to the storage and data management module 5;

[0009] The data import / export module 1 receives the data export interaction instruction from the human-machine interaction module 6 as input, obtains the path and associated structure of the target file from the storage and data management module 5, and exports the file to the external device through interfaces such as USB / gigabit Ethernet / 10-gigabit optical port; the data import / export module 1 feeds back information such as the export progress and error control to the human-machine interaction module 6;

[0010] (2) Preprocessing module 2

[0011] The preprocessing module 2 receives the data preprocessing interaction instruction and preprocessing parameters from the human-machine interaction module 6 as input, preprocesses the data to be processed input from the storage and data management module 5, and feeds back the obtained file list to the human-machine interaction module 6; the preprocessing result output by the preprocessing module 2 is output to the storage and data management module 5 and also output to the intelligent inference module 4;

[0012] (3) Neural network deployment module 3

[0013] The neural network deployment module 3 receives the neural network deployment interaction instruction and deployment parameter settings from the human-machine interaction module 6, compiles the network model stored in the SSD hard disk of the portable computer, and outputs the compiled network model to the intelligent inference module 4; the neural network deployment module 3 feeds back the list and path of the selectable network models, the network model type and parameter quantity, and the post-processing required by the network model to the human-machine interaction module 6;

[0014] (4) Intelligent inference module 4

[0015] The intelligent inference module 4 receives the inference interaction instruction from the human-machine interaction module 6, receives the preprocessing result from the preprocessing module 2 and the network model from the neural network deployment module 3, and completes the inference calculation on the dedicated NPU of the hardware device; the inference detection result obtained by the intelligent inference module 4 is output to the human-machine interaction module 6 and also output to the storage and data management module 5;

[0016] (5) Storage and data management module 5 and human-machine interaction module 6

[0017] The interaction control relationship between the storage and data management module 5 and the human-machine interaction module 6 and other modules is as described in the previous four modules.

[0018] In an embodiment of the present invention, the data import / export module 1 is specifically as follows:

[0019] (a) The data import / export interface includes a Gigabit Ethernet interface and a USB 3.0; remote access to the actual acquisition raw data of the ground data storage array and the mission system display and control computer is realized through the Ethernet for import, and the actual acquisition raw data of the ground data storage array and the mission system display and control computer is imported through the USB 3.0 to build a shared network;

[0020] (b) There are two types of imported data. One is the actual acquisition raw data, and the other is the data interacting with the same system. Importing refers to importing from external devices to the local;

[0021] There are three types of exported data: the actual acquisition raw data, the preprocessed time-frequency matrix data, and the inference result data. Exporting refers to exporting local files to external devices.

[0022] A working method of a portable intelligent signal processing system applicable to the rapid detection and recognition of communication signals is proposed, which specifically includes the following steps:

[0023] The first step: Data import

[0024] The data import / export module 1 receives the data import interaction instruction of the human-computer interaction module 6, and imports the actual acquisition signal data from external devices such as the ground storage array and the mission display and control computer through these interfaces of USB / Gigabit Ethernet / 10 Gigabit optical port; the imported data is input to the storage and data management module 5, and the signal data is stored in the hard disk of the hardware device and data management is performed; the data import / export module 1 receives the data export interaction instruction of the human-computer interaction module 6, receives the data file path given by the storage and data management module 5, and exports the file to external devices through these interfaces of USB / Gigabit Ethernet / 10 Gigabit optical port;

[0025] The second step: Data preprocessing

[0026] The preprocessing module 2 receives the data preprocessing interaction instruction and preprocessing parameters of the human-computer interaction module 6, and processes the data to be processed input from the storage and data management module 5; the preprocessing result output by the preprocessing module 2 is managed by the storage and data management module 5 and output to the intelligent inference module 4 for subsequent inference;

[0027] The third step: Neural network deployment

[0028] The neural network deployment module 3 receives the neural network deployment interaction instruction and deployment parameter setting of the human-computer interaction module 6, compiles the network model stored in the portable computer SSD hard disk, and outputs the compiled network model to the intelligent inference module 4;

[0029] Step 4: Intelligent Inference

[0030] Send the time-frequency matrix obtained from preprocessing to the deployed neural network for inference. The NPU completes the inference calculation after the time-frequency matrix is input into the neural network and outputs the result;

[0031] The intelligent inference module 4 receives the inference interaction instruction from the human-computer interaction module 6, receives the preprocessing result from the preprocessing module 2 and the network model from the neural network deployment module 3, and completes the inference calculation on the dedicated NPU of the hardware device. The inference detection result obtained by the NPU is output to the human-computer interaction module 6 for visual display, and is also output to the storage and data management module 5 for result storage and management.

[0032] In another embodiment of the present invention, the preprocessing process involved in the second step "data preprocessing" is specifically as follows:

[0033] (1) Data Reading and Data Shaping

[0034] The signal data is stored in the portable computer SSD hard disk one by one in the binary file IFD / dat format; each file IFD / dat consists of a file header with a fixed byte length and a subsequent acquisition data segment; the file header stores parameters such as sampling bandwidth, sampling rate, and center frequency; specifically as follows:

[0035] (a) The preprocessing module 2 reads the header file of the signal data from the portable computer SSD hard disk and records the acquisition time, sampling bandwidth, sampling rate, and center frequency for subsequent display;

[0036] (b) The default format of the acquisition data segment in the signal data file in the portable computer SSD hard disk is in complex form, divided into two channels, the I channel and the Q channel, and is stored in the form of alternating IQIQIQ. After the preprocessing module 2 reads the acquisition data segment, it performs data shaping on it, converting the acquisition data segment into the form of I + jQ, where j is the imaginary unit, for subsequent time-frequency transformation processing;

[0037] (c) The data reading form adopts one of the following two methods: one is that the preprocessing module 2 automatically loads all the signal data from the portable computer SSD hard disk into the memory at one time; the other is to manually load a certain signal data from the portable computer SSD hard disk into the memory;

[0038] (2) Signal Preprocessing

[0039] Perform signal preprocessing operations such as filtering, denoising, and resampling on the data in the form of I + jQ obtained in the previous step to obtain preprocessed data. The preprocessing operations can be extended according to requirements;

[0040] (3) Data Segmentation

[0041] Segment the preprocessed data first; when segmenting the data, set the length and step size of each segment.

[0042] (4) Time-frequency transformation

[0043] After data segmentation, perform time-frequency transformation on each segment of data to obtain a time-frequency matrix with the same size as the training environment, specifically as follows:

[0044] (a) The short-time Fourier transform (STFT) is used as the default time-frequency transformation. The configurable parameters include window type, window overlap, and the number of points of the discrete Fourier transform (DFT). The obtained time-frequency matrix after time-frequency transformation is a complex-valued matrix.

[0045] (b) Perform square / modulus, logarithm, and normalization processing on the complex-valued matrix obtained by time-frequency transformation in sequence. The processed matrix becomes a real-valued matrix.

[0046] (5) Saving the time-frequency matrix

[0047] Store the real-valued time-frequency matrix obtained in the previous step in two forms: image and matrix. For image storage, further convert the time-frequency matrix into the uint8 data format, so that the values of the matrix are transformed into the image value range of [0, 255], and then save it in the image format. Matrix storage is to directly save the original time-frequency matrix.

[0048] In another embodiment of the present invention, in step (4) "Time-frequency transformation" (a), the window type is selected as the Hamming window, and the window overlap and discrete Fourier transform are set to 256 points and 512 points respectively.

[0049] In yet another embodiment of the present invention, in the fourth step "Intelligent inference", multiple NPU parallel inference architectures are adopted for inference, that is, the input time-frequency matrix is sent to multiple NPUs simultaneously, and multiple NPUs calculate simultaneously; different network models are supported for deployment on different NPUs during inference.

[0050] This intelligent communication signal processing system first converts the collected signal time series into a time-frequency matrix through preprocessing, and then inputs it into the neural network deployed at the backend for inference, completing the rapid detection, recognition, and time-frequency parameter estimation of wireless communication signals. At the same time, it can summarize and visualize the detection results. The platform is equipped with an interactive and user-friendly upper computer software and UI interface, which can realize functions such as neural network deployment, data import, preprocessing, inference, storage, and human-computer interaction. This system introduces intelligent signal processing technology, constructs a dataset through fine signal analysis, recognition, and annotation of offline signals; trains a deep network on this basis, and transfers the capabilities of existing analysts to the deep neural network; can complete the rapid analysis and recognition of communication reconnaissance signals, realize the knowledge transfer and iterative improvement of manual offline analysis, and effectively improve the intelligence generation efficiency. Description of the Drawings

[0051] Figure 1 A connection diagram showing a portable intelligent signal processing system applicable to the rapid detection and identification of communication signals according to the present invention;

[0052] Figure 2 A workflow diagram showing a portable intelligent signal processing system applicable to the rapid detection and identification of communication signals according to the present invention;

[0053] Figure 3 Showing the preprocessing process. Detailed Description of the Invention

[0054] I. A portable intelligent signal processing system applicable to the rapid detection and identification of communication signals;

[0055] The present invention provides a portable intelligent signal processing system applicable to the rapid detection and identification of communication signals, which consists of two parts: a portable computer hardware device and a host computer software system.

[0056] The portable computer hardware device adopts a CPU+NPU computing architecture, which can support the deployment of models of various neural network frameworks such as Pytorch, Caffe, and Darknet, support multi-core multi-algorithm parallel acceleration, multi-core single-algorithm acceleration, large-image partition multi-core acceleration, multi-algorithm simultaneous loading, algorithm switching during runtime (ms level), one-key configuration of dual algorithms for continuous inference, etc. The portable computer hardware device also includes a memory, an SSD hard disk, a battery, an LED display, an IO interface, etc., as known to those skilled in the art.

[0057] The host computer software system includes a data import / export module 1, a preprocessing module 2, a neural network deployment module 3, an intelligent inference module 4, a storage and data management module 5, and a human-computer interaction module 6. The connection relationship between the six modules and the portable computer hardware device is as Figure 1 shown.

[0058] 1. Data import / export module 1

[0059] The data import / export module 1 receives the data import interaction instruction from the human-computer interaction module 6 as input, and imports the actual acquisition signal data from external devices such as a ground storage array and a mission display and control computer into the system through interfaces such as USB / gigabit network port / 10-gigabit optical port. In addition, during the import process, the data import / export module 1 will transfer the file path and associated structure of the imported data to the storage and data management module 5, which will manage them uniformly to achieve subsequent convenient data file retrieval, screening, and other operations.

[0060] For data export, the data import / export module 1 receives the data export interaction instruction from the human-machine interaction module 6 as input, obtains the path and associated structure of the target file from the storage and data management module 5, and exports the file to an external device through interfaces such as USB / gigabit Ethernet / 10-gigabit optical port. Similarly, during the data export process, the data import / export module 1 feeds back information such as the export progress and error control to the human-machine interaction module 6 and then displays it on the interface.

[0061] (1) The data import / export interfaces include gigabit Ethernet interfaces and USB 3.0. Remote access to the actual acquisition raw data of the ground data storage array and the task system display and control computer can be achieved through Ethernet for import, and the actual acquisition raw data of the ground data storage array and the task system display and control computer can be imported by constructing a shared network through USB 3.0;

[0062] (2) There are two types of imported data. One is the actual acquisition raw data, and the other is the data for interaction with the same system. Importing refers to importing from an external device to the local. The import methods for the two types of data are shown in the following table.

[0063]

[0064]

[0065] There are three types of exported data: actual acquisition raw data, preprocessed time-frequency matrix data, and inference result data. Exporting refers to exporting local files to an external device. The export method is the independent or combined export of the three types of data, as shown in the following table.

[0066]

[0067] (3) File formats that can be supported for import: include but are not limited to formats such as IFD, dat, bin, mat, etc., and these formats can be converted to each other.

[0068] 2. Preprocessing module 2

[0069] The preprocessing module 2 receives the data preprocessing interaction instruction and preprocessing parameters from the human-machine interaction module 6 as input, preprocesses the data to be processed input from the storage and data management module 5, and feeds back the obtained file list to the human-machine interaction module 6 for display. In addition, the preprocessing result output by the preprocessing module 2 is not only managed by the storage and data management module 5 but also output to the intelligent inference module 4 for subsequent inference.

[0070] 3. Neural network deployment module 3

[0071] The neural network deployment module 3 receives the neural network deployment interaction instruction and deployment parameter settings from the human-computer interaction module 6 as inputs, compiles the network model stored in the portable computer's SSD hard drive, and outputs the compiled network model to the intelligent inference module 4. In addition, the neural network deployment module 3 also feeds back information such as the selectable network model list and path, network model type and parameter quantity, and post-processing required by the network model to the human-computer interaction module 6 and then displays it on the interface.

[0072] 4. Intelligent Inference Module 4

[0073] The intelligent inference module 4 receives the inference interaction instruction from the human-computer interaction module 6 as an input. On this basis, the intelligent inference module 4 also takes the preprocessing result from the preprocessing module 2 and the network model from the neural network deployment module 3 as inputs, and completes the inference calculation on the dedicated NPU of the hardware device. The inference detection result obtained by the intelligent inference module 4 is not only output to the human-computer interaction module 6 for visual display, but also output to the storage and data management module 5 for result storage and management.

[0074] 5. Storage and Data Management Module 5 and Human-Computer Interaction Module 6

[0075] The interaction control relationships between the storage and data management module 5 and the human-computer interaction module 6 with other modules are as described in the previous four modules.

[0076] II. Working Method of the Portable Intelligent Signal Processing System Applicable to Rapid Detection and Recognition of Communication Signals;

[0077] The working process of the system from importing the original signal acquisition data to the end of signal detection and recognition inference is as Figure 2 shown, and there are a total of four steps: data import, data preprocessing, neural network deployment, and intelligent inference.

[0078] First Step: Data Import

[0079] Data import involves the data import / export module 1, the storage and data management module 5, and the human-computer interaction module 6. The data import / export module 1 receives the data import interaction instruction from the human-computer interaction module 6 as an input, and imports the actual acquired signal data from external devices such as the ground storage array and the mission display and control computer through interfaces such as USB / gigabit network port / 10-gigabit optical port. The imported data is input to the storage and data management module 5, where the signal data is stored in the hard drive of the hardware device and data management is performed simultaneously. In addition, for data export, the data import / export module 1 receives the data export interaction instruction from the human-computer interaction module 6 as an input, and also takes the data file path given by the storage and data management module 5 as an input, and exports the file to the external device through interfaces such as USB / gigabit network port / 10-gigabit optical port.

[0080] Step 2: Data preprocessing

[0081] Data preprocessing involves a data import / export module 2, an intelligent inference module 4, a storage and data management module 5, and a human-computer interaction module 6. The preprocessing module 2 receives the data preprocessing interaction instructions and preprocessing parameters from the human-computer interaction module 6 as inputs, and processes the data to be processed input from the storage and data management module 5. In addition, the preprocessing results output by the preprocessing module 2 are not only managed by the storage and data management module 5, but also output to the intelligent inference module 4 for subsequent inference.

[0082] The preprocessing process involved in the preprocessing module 2 is as Figure 3 shown, mainly including five steps: data reading and shaping, signal preprocessing, data segmentation, time-frequency transformation, and time-frequency matrix saving, which are specifically as follows:

[0083] (1) Data reading and shaping

[0084] The signal data is stored in the portable computer SSD hard disk one by one in the binary file IFD / dat format. Each file IFD / dat consists of a file header with a fixed byte length and a subsequent acquired data segment. The file header stores parameters such as the sampling bandwidth, sampling rate, and center frequency. Specifically as follows:

[0085] (a) The preprocessing module 2 reads the header file of the signal data from the portable computer SSD hard disk, and records parameters such as the acquisition time, sampling bandwidth, sampling rate, and center frequency for subsequent display.

[0086] (b) The default format of the acquired data segment in the signal data file in the portable computer SSD hard disk is in complex form, divided into two channels: I channel and Q channel (the I channel and Q channel are well-known to those skilled in the art), and is stored in the form of alternating IQIQIQ. After the preprocessing module 2 reads the acquired data segment, it performs data shaping on it, converting the acquired data segment into the form of I + jQ (j is the imaginary unit) for subsequent time-frequency transformation processing.

[0087] (c) There are two ways of data reading. One is that the preprocessing module 2 automatically loads all the signal data from the portable computer SSD hard disk into the memory at one time; the other is that a certain signal data can be manually loaded from the portable computer SSD hard disk into the memory.

[0088] (2) Signal preprocessing

[0089] Signal preprocessing operations such as filtering, denoising, and resampling are performed on the data in the form of I + jQ obtained in the previous step to obtain preprocessed data, and the preprocessing operations can be extended according to requirements. "Signal preprocessing operations such as filtering, denoising, and resampling" are well-known to those skilled in the art and will not be elaborated here.

[0090] (3) Data segmentation

[0091] The time-frequency matrix of a fixed size corresponds to a time series of a fixed length. Therefore, the preprocessed data is segmented first. When segmenting the data, the length and step size of each segment can be set. For example, the length of each segment is 60,000 points and the step size is 30,000 points.

[0092] (4) Time-frequency transformation

[0093] After segmenting the data, perform time-frequency transformation on each segment of data, and a time-frequency matrix with the same size as the training environment can be obtained, specifically as follows:

[0094] (a) The short-time Fourier transform (STFT) is used as the default time-frequency transformation. The parameters that can be set include the window type, window overlap, and the number of points of the discrete Fourier transform (DFT). For example, the window type is selected as the Hamming window, and the window overlap and the discrete Fourier transform are set to 256 points and 512 points respectively. The time-frequency matrix obtained after the time-frequency transformation is a complex-valued matrix.

[0095] (b) Perform square / modulus, logarithm, and normalization processing on the complex-valued matrix obtained by the time-frequency transformation in sequence. The processed matrix becomes a real-valued matrix. Square / modulus, logarithm, and normalization processing are well-known to those skilled in the art and will not be elaborated here.

[0096] The time-frequency transformation is the key to obtaining the inference time-frequency matrix.

[0097] (5) Saving the time-frequency matrix

[0098] Store the real-valued time-frequency matrix obtained in the previous step in two forms: image and matrix. Image storage is to further convert the time-frequency matrix into the uint8 data format (this data format is well-known to those skilled in the art), so that the values of the matrix are transformed into the image value range of [0, 255], and then saved in the image format, such as.png or.jpg format. Matrix storage is to directly save the original time-frequency matrix.

[0099] Step 3: Neural network deployment

[0100] Neural network deployment involves a neural network deployment module 3, an intelligent inference module 4, a storage and data management module 5, and a human-computer interaction module 6. The neural network deployment module 3 receives the neural network deployment interaction instruction and deployment parameter settings from the human-computer interaction module 6 as inputs, compiles the network model stored in the portable computer SSD hard disk (in this invention, a neural network compilation tool supporting the HKN201NPU chip developed by Xi'an Xiangteng Microelectronics Technology Co., Ltd. is used for compilation), and outputs the compiled network model to the intelligent inference module 4.

[0101] The types of neural networks that can support deployment include CNN (Pytorch), etc. The network deployment precision supports float16 and int8.

[0102] Step 4: Intelligent Inference

[0103] Send the time-frequency matrix obtained by preprocessing into the deployed neural network for inference (the process of inputting the time-frequency matrix into the neural network to obtain the output is called inference, which is well-known to those skilled in the art). The NPU completes the inference calculation after the time-frequency matrix is input into the neural network and outputs it. The classification of the signal type and the regression estimation of the signal parameters (the parameters involve the position where the signal spectrum appears, bandwidth, duration, etc.) can be output, which involves the preprocessing module 2, the neural network deployment module 3, the intelligent inference module 4, the storage and data management module 5, and the human-computer interaction module 6.

[0104] The intelligent inference module 4 receives the inference interaction instruction from the human-computer interaction module 6 as input. On this basis, the intelligent inference module 4 also takes the preprocessing result from the preprocessing module 2 and the network model from the neural network deployment module 3 as inputs, and completes the inference calculation on the dedicated NPU of the hardware device. The inference detection result obtained by the intelligent inference module 4 from the NPU is not only output to the human-computer interaction module 6 for visual display, but also output to the storage and data management module 5 for result storage and management.

[0105] The inference adopts a multi-NPU parallel inference architecture, that is, the input time-frequency matrix is sent to multiple NPUs at the same time, and multiple NPUs calculate simultaneously. Different network models can be supported for deployment on different NPUs during inference. int8 and fp16 precision inference are supported.

[0106] The intelligent processing platform applicable to the rapid detection and recognition of communication signals first converts the collected signal time series into a time-frequency matrix through preprocessing, and then inputs it into the neural network deployed at the backend for inference, finally completing the rapid detection, recognition, and time-frequency parameter estimation of wireless signals, and being able to summarize and visualize the detection results. The data import and export module realizes the import of the actually collected original data and the data interacting with the same system through various methods; at the same time, it realizes the export of the actually collected original data, the time-frequency matrix data after preprocessing, and the inference result data. The preprocessing module uses the short-time Fourier transform to perform preprocessing calculations on the input wireless collected data to form a time-frequency matrix. The neural network deployment module completes the deployment of the trained neural network. The intelligent inference module sends the time-frequency matrix into the deployed neural network for inference, and the NPU completes the calculation, which can output the classification of the signal type and the regression estimation of signal parameters (parameters involve the position where the signal spectrum appears, bandwidth, duration, etc.). The storage and data management module stores and manages the input original collected data, the time-frequency matrix after preprocessing, and the inference results, and at the same time ensures fast retrieval; the human-computer interaction module provides human-computer interaction to achieve fast and friendly operations.

Claims

1. A portable intelligent signal processing system applicable to rapid detection and recognition of communication signals, comprising two parts: a portable computer hardware device and a host computer software system; the portable computer hardware device adopts a CPU+NPU computing architecture; the portable computer hardware device also includes a memory, an SSD hard disk, a battery, an LED display, and an IO interface; characterized in that: The host computer software system includes a data import / export module (1), a preprocessing module (2), a neural network deployment module (3), an intelligent inference module (4), a storage and data management module (5), and a human-computer interaction module (6); specifically as follows: (1) Data import / export module (1) The data import / export module (1) receives the data import interaction instruction of the human-computer interaction module (6) as input, and imports the actual acquisition signal data from the ground storage array and the mission display and control computer into this system through a USB / gigabit network port / 10-gigabit optical port; The data import / export module (1) transfers the file path and associated structure of the imported data to the storage and data management module (5); The data import / export module (1) receives the data export interaction instruction of the human-computer interaction module (6) as input, obtains the path and associated structure of the target file from the storage and data management module 5, and exports the file to an external device through a USB / gigabit network port / 10-gigabit optical port; The data import / export module (1) feeds back the export progress and error control information to the human-computer interaction module (6); (2) Preprocessing module (2) The preprocessing module (2) receives the data preprocessing interaction instruction and preprocessing parameters of the human-computer interaction module (6) as input, preprocesses the data to be processed input from the storage and data management module (5), and feeds back the obtained file list to the human-computer interaction module (6); The preprocessing result output by the preprocessing module (2) is output to the storage and data management module (5), and is also output to the intelligent inference module (4); (3) Neural network deployment module (3) The neural network deployment module (3) receives the neural network deployment interaction instruction and deployment parameter settings of the human-computer interaction module (6), compiles the network model stored in the portable computer SSD hard disk, and outputs the compiled network model to the intelligent inference module (4); the neural network deployment module (3) feeds back the list and path of selectable network models, the network model type and parameter quantity, and the post-processing required by the network model to the human-computer interaction module (6); (4) Intelligent inference module (4) The intelligent inference module (4) receives the inference interaction instruction of the human-computer interaction module (6), receives the preprocessing result from the preprocessing module (2) and the network model from the neural network deployment module (3), and completes the inference calculation on the dedicated NPU of the hardware device; the inference detection result obtained by the intelligent inference module (4) is output to the human-computer interaction module (6), and is also output to the storage and data management module (5); (5) Storage and data management module (5) and human-computer interaction module (6) The interaction control relationship between the storage and data management module (5) and the human-computer interaction module (6) and other modules is as described in the previous four modules.

2. The portable intelligent signal processing system applicable to rapid detection and recognition of communication signals according to claim 1, characterized in that, The data import and export module (1) is as follows: (a) The data import and export interface includes a Gigabit Ethernet interface and a USB 3.0 interface; remote access to the actual acquisition raw data of the ground data storage array and the mission system display and control computer is achieved through Ethernet for import, and the actual acquisition raw data of the ground data storage array and the mission system display and control computer is imported through the USB 3.0 interface to build a shared network; (b) There are two types of imported data. One is the actual acquisition raw data, and the other is the data interacting with the same system. Importing refers to importing from external devices to the local; There are three types of exported data: the actual acquisition raw data, the preprocessed time-frequency matrix data, and the inference result data. Exporting refers to exporting local files to external devices.

3. The working method of a portable intelligent signal processing system applicable to the rapid detection and recognition of communication signals, characterized in that, Specifically, it includes the following steps: The first step: Data import The data import and export module (1) receives the data import interaction instruction from the human-computer interaction module (6), and imports the actual acquisition signal data from the ground storage array and the mission display and control computer through the USB / Gigabit network port / 10 Gigabit optical port; The imported data is input to the storage and data management module (5), and the signal data is stored in the hard disk of the hardware device and data management is performed; the data import and export module (1) receives the data export interaction instruction from the human-computer interaction module (6), receives the data file path given by the storage and data management module (5), and exports the file to an external device through the USB / Gigabit network port / 10 Gigabit optical port; The second step: Data preprocessing The preprocessing module (2) receives the data preprocessing interaction instruction and preprocessing parameters from the human-computer interaction module (6), and processes the data to be processed input from the storage and data management module (5); The preprocessing result output by the preprocessing module (2) is managed by the storage and data management module (5) and output to the intelligent inference module (4) for subsequent inference; The third step: Neural network deployment The neural network deployment module (3) receives the neural network deployment interaction instruction and deployment parameter settings from the human-computer interaction module (6), compiles the network model stored in the SSD hard disk of the portable computer, and outputs the compiled network model to the intelligent inference module (4); the neural network deployment module (3) feeds back the list and path of the selectable network model, the network model type and the number of parameters, and the post-processing required by the network model to the human-computer interaction module (6); The fourth step: Intelligent inference The preprocessed time-frequency matrix is sent to the deployed neural network for inference. The NPU completes the inference calculation after the time-frequency matrix is input to the neural network and outputs; The intelligent inference module (4) receives the inference interaction instruction from the human-computer interaction module (6), receives the preprocessing result from the preprocessing module (2) and the network model from the neural network deployment module (3), and completes the inference calculation on the dedicated NPU of the hardware device; the inference detection result obtained by the NPU is output to the human-computer interaction module (6) for visual display, and is also output to the storage and data management module (5) for result storage and management.

4. The working method of the portable intelligent signal processing system applicable to rapid detection and recognition of communication signals as described in claim 3, the specific preprocessing process involved in the second step of "data preprocessing" is as follows: (1) Data reading and data shaping The signal data is stored in the SSD hard disk of the portable computer one by one in the binary file IFD / dat format; each file IFD / dat consists of a file header with a fixed byte length and a subsequent acquisition data segment; the file header stores parameters such as sampling bandwidth, sampling rate, and center frequency; specifically as follows: (a) The preprocessing module (2) reads the header file of the signal data from the SSD hard disk of the portable computer, and records the acquisition time, sampling bandwidth, sampling rate, and center frequency for subsequent display; (b) The default format of the acquisition data segment in the signal data file in the SSD hard disk of the portable computer is in complex form, divided into two channels, the I channel and the Q channel, and is stored in the form of alternating IQIQIQ. After the preprocessing module (2) reads the acquisition data segment, it performs data shaping on it, converting the acquisition data segment into the form of I + jQ, where j is the imaginary unit, for subsequent time-frequency transformation processing; (c) The data reading form adopts one of the following two methods: one is that the preprocessing module (2) automatically loads all signal data from the SSD hard disk of the portable computer into the memory at one time; the other is to manually load a certain signal data from the SSD hard disk of the portable computer into the memory; (2) Signal preprocessing Perform signal preprocessing operations such as filtering, denoising, and resampling on the data in the form of I + jQ obtained in the previous step to obtain preprocessed data, and the preprocessing operations can be extended according to requirements; (3) Data segmentation First, segment the preprocessed data; when segmenting the data, set the length and step size of each segment; (4) Time-frequency transformation After data segmentation, perform time-frequency transformation on each segment of data to obtain a time-frequency matrix with the same size as the training environment, specifically as follows: (a) The short-time Fourier transform STFT is used as the default time-frequency transformation, and the settable parameters include window type, window overlap, and the number of points of the discrete Fourier transform DFT; the time-frequency matrix obtained after the time-frequency transformation is a complex-valued matrix; (b) Perform square / modulus, logarithm, and normalization processing on the complex-valued matrix obtained by the time-frequency transformation in sequence, and the processed matrix becomes a real-valued matrix; (5) Saving of the time-frequency matrix Store the real-valued time-frequency matrix obtained in the previous step in two forms: image and matrix; for image storage, the time-frequency matrix is further converted into the uint8 data format, so that the values of the matrix are transformed into the image value range of [0, 255], and then saved in the image format; matrix storage is to directly save the original time-frequency matrix.

5. The working method of the portable intelligent signal processing system applicable to rapid detection and recognition of communication signals as described in claim 4, in step (4) "time-frequency transformation" (a), the window type is selected as the Hamming window, and the window overlap and the discrete Fourier transform are set to 256 points and 512 points respectively.

6. For the working method of the portable intelligent signal processing system applicable to the rapid detection and recognition of communication signals as described in claim 3, in the fourth step, "intelligent reasoning", the reasoning adopts a parallel reasoning architecture of multiple NPUs, that is, the input time-frequency matrix is simultaneously sent to multiple NPUs, and multiple NPUs calculate simultaneously; different network models are supported to be deployed on different NPUs during reasoning.

Images