CN116390139A - Equipment detection method and device based on neural network, computer equipment and medium - Google Patents

Abstract

Embodiments of the present application provide a neural network-based device detection method, device, computer equipment, and media, which belong to the field of communication technology. The method includes: obtaining a training data set and a test data set; constructing an indicator variable function according to the active state and delay value of the equipment in the above data set; inputting the indicator variable function into the neural network for sum calculation, and outputting a covariance matrix; The matrix, the preset first matrix, the preset second matrix, and the preset learnable parameters perform gradient calculations on the training data set and the test data set to obtain gradient information; iteratively update the gradient information and the indicator variable function to obtain The variable is updated; the neural network is iterated according to the updated variable, and the communication signal of the device to be detected is input into the iterative neural network to determine the active state of the device to be detected. The embodiment of the present application can avoid high-complexity operations in the iterative algorithm, and improve the detection accuracy of active users.

Description

Neural network-based equipment detection method, device, computer equipment and medium

technical field

The present application relates to the field of communication technology, and in particular to a neural network-based device detection method, device, computer equipment and media.

Background technique

With the development of communication technology, large-scale machine communication has become the main way to connect massive IoT terminals. Among them, during the scheduling-free random access process, active users directly perform data transmission on the allocated wireless communication resources. Existing access procedures can be roughly divided into two categories. The first category is to use the burstiness of machine communication services, adopt the method based on compressed sensing, and realize the detection of active equipment through the joint estimation of equipment activity status and instantaneous channel. In contrast, the other type uses the probability distribution of the channel to directly detect active devices based on the sample covariance of the signal received by the access point. Compared with compressive sensing-based methods, covariance-based methods do not need to estimate the instantaneous channel, so more accurate active device detection results can be obtained. Although the existing active device detection algorithms have better detection capabilities to a certain extent, the implementation of these algorithms usually requires solving high-dimensional non-convex problems. However, solving such non-convex problems requires the design of an iterative algorithm with high computational complexity, which does not meet the needs of practical applications, thus affecting the ability to detect active devices.

Contents of the invention

The main purpose of the embodiments of the present application is to propose a neural network-based device detection method, device, computer equipment, and media, which can avoid high-complexity operations in iterative algorithms and improve the detection accuracy of active users.

In order to achieve the above purpose, the first aspect of the embodiment of the present application proposes a neural network-based device detection method, the method comprising:

Acquiring a training data set and a test data set, wherein the training data set and the test data set are calculated by a preset data generation function on communication signals collected by an access point, and the access point includes at least one device;

Constructing an indicator variable function according to the active state and delay value of the equipment in the training data set and the test data set;

Input the indicator variable function into the neural network for sum calculation, and output the covariance matrix;

performing gradient calculation on the training data set and the test data set according to the covariance matrix, the preset first matrix, the preset second matrix and the preset learnable parameters to obtain gradient information;

Iteratively updating the gradient information and the indicator variable function based on the learnable parameters to obtain updated variables;

The neural network is iterated according to the update variable, and the communication signal of the device to be detected is input into the iterated neural network for active detection, and the active state of the device to be detected is determined.

In some embodiments, also include:

Obtaining a characteristic sequence of the device and a delay value corresponding to the characteristic sequence;

An equivalent signature sequence of the device is obtained according to the signature sequence and the delay value.

In some embodiments, the training data set and the test data set are obtained according to the following steps:

Obtaining small-scale fading channel information between the device and the access point and transmit power of the device;

calculating the distance between the device and the access point to obtain a distance value;

inputting the distance value into a preset loss model to obtain large-scale fading information;

generating a first function according to the transmit power, the large-scale fading information, and the equivalent eigensequence;

Obtaining the data generation function according to the small-scale fading channel information, the first function, and a preset Gaussian distribution value;

The communication signal collected by the access point is input into the data generation function for calculation, and the training data set and the test data set are output.

In some embodiments, the input of the indicator variable function into the neural network for sum calculation, and the output covariance matrix includes:

Acquiring the noise power of the access point;

Inputting the noise power, the first function, the indicator variable function, and a preset identity matrix into the neural network for sum calculation, and outputting a covariance matrix.

In some embodiments, the learnable parameters include a penalty factor; the training data set is analyzed according to the covariance matrix, the preset first matrix, the preset second matrix and the preset learnable parameters And the test data set performs gradient calculation to obtain gradient information, including:

performing a matrix approximation operation on the covariance matrix according to the first matrix and the second matrix to obtain an approximate matrix;

Perform gradient calculation on the training data set and the test data set according to the approximation matrix and the penalty factor to obtain the gradient information.

In some embodiments, the learnable parameter includes an iterative step; the iterative update of the gradient information and the indicator variable function based on the learnable parameter to obtain an updated variable includes:

multiplying the iteration step size by the gradient information to obtain a product result;

The update variable is obtained according to the indicator variable function and the product result.

In some embodiments, iterating the neural network according to the updated variable, and inputting the device to be detected into the iterated neural network for active detection, and determining the active state of the device to be detected includes:

generating an iterative formula according to the update variable and the learnable parameter;

Iterating the neural network according to the iterative formula;

Inputting the device to be detected into the iterated neural network to perform activity detection to determine the active state of the device to be detected.

The second aspect of the embodiment of the present application proposes a device detection device based on a neural network, the device comprising:

A data acquisition module, configured to acquire a training data set and a test data set, wherein the training data set and the test data set are obtained by calculating the communication signals collected by the access point by a preset data generation function, and the access A point includes at least one device;

A function construction module, configured to construct an indicator variable function according to the active state and delay value of the equipment in the training data set and the test data set;

An iterative calculation module, configured to input the indicator variable function into the neural network for sum calculation, and output a covariance matrix;

A gradient calculation module, configured to perform gradient calculation on the training data set and the test data set according to the covariance matrix, the preset first matrix, the preset second matrix and the preset learnable parameters, to obtain gradient information;

an iterative update module, configured to iteratively update the gradient information and the indicator variable function based on the learnable parameters to obtain updated variables;

The active detection module is configured to iterate the neural network according to the update variable, and input the communication signal of the device to be detected into the iterated neural network for active detection, and determine the active state of the device to be detected.

The third aspect of the embodiments of the present application provides a computer device, the computer device includes a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processing The device is configured to execute the method described in any one of the embodiments of the first aspect of the present application.

The fourth aspect of the embodiments of the present application provides a storage medium, the storage medium is a computer-readable storage medium, the computer-readable storage medium stores a computer program, and when the computer program is executed by a computer, the The computer is used to execute the method described in any one of the embodiments of the first aspect of the present application.

The neural network-based equipment detection method, device, computer equipment, and medium proposed in the embodiments of the present application calculate the communication signals collected by the access point according to the data generation function to obtain the training data set and the test data set, and according to the training data The active state and delay value of the equipment in the set and test data set construct an indicator variable function, which is convenient for subsequent in-depth expansion, and the indicator variable function is input into the neural network for summation calculation, and the output covariance matrix is convenient for subsequent training of the neural network. After that, According to the covariance matrix, the preset first matrix, the preset second matrix and the preset learnable parameters, the gradient calculation is performed on the training data set and the test data set, and the gradient information is obtained, and the learnable parameters and the preset matrix are introduced. It can avoid the inversion operation in the calculation process, reduce the computational complexity, and then iteratively update the gradient information and the indicator variable function based on the learnable parameters to obtain the updated variable and improve the detection performance of the neural network for the device. Finally, according to the updated variable The neural network is iterated, and the communication signal of the device to be detected is input into the iterative neural network for active detection to determine the active state of the device to be detected, thereby reducing the number of iterations of the neural network and improving the detection ability of the neural network for active devices.

Description of drawings

Fig. 1 is a flow chart of a neural network-based device detection method provided in an embodiment of the present application;

FIG. 2 is a flowchart of a neural network-based device detection method provided by another embodiment of the present application;

Fig. 3 is the specific flowchart of step S101 in Fig. 1;

Fig. 4 is the specific flowchart of step S103 among Fig. 1;

Fig. 5 is the specific flowchart of step S104 among Fig. 1;

Fig. 6 is the specific flowchart of step S105 among Fig. 1;

Fig. 7 is the specific flowchart of step S106 among Fig. 1;

Fig. 8 is a schematic diagram of a neural network-based device detection method provided in a specific example of the present application;

FIG. 9 is a schematic diagram of a neural network-based device detection method provided in another specific example of the present application;

Figure 10 is a block diagram of the module structure of the active user and data detection device provided by the embodiment of the present application;

FIG. 11 is a schematic diagram of a hardware structure of a computer device provided by an embodiment of the present application.

Detailed ways

In order to make the purpose, technical solution and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application, not to limit the present application.

It should be noted that although the functional modules are divided in the schematic diagram of the device, and the logical sequence is shown in the flowchart, in some cases, it can be executed in a different order than the module division in the device or the flowchart in the flowchart. steps shown or described. The terms "first", "second" and the like in the specification and claims and the above drawings are used to distinguish similar objects, and not necessarily used to describe a specific sequence or sequence.

Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application, and are not intended to limit the present application.

With the development of communication technology, large-scale machine communication has become the main way to connect massive IoT terminals. Among them, during the scheduling-free random access process, active users directly perform data transmission on the allocated wireless communication resources. Existing access procedures can be roughly divided into two categories. The first category is to use the burstiness of machine communication services, adopt the method based on compressed sensing, and realize the detection of active equipment through the joint estimation of equipment activity status and instantaneous channel. In contrast, the other type uses the probability distribution of the channel to directly detect active devices based on the sample covariance of the communication signal received by the access point. Compared with compressive sensing-based methods, covariance-based methods do not need to estimate the instantaneous channel, so more accurate active device detection results can be obtained.

Although active device detection algorithms in the related art have better detection capabilities to a certain extent, the implementation of these algorithms usually requires solving high-dimensional non-convex problems. However, solving such non-convex problems requires the design of an iterative algorithm with high computational complexity, which does not meet the needs of practical applications. Therefore, it is very necessary to innovate in the access method of large-scale machine communication and explore low-complexity solutions.

Based on this, the main purpose of the embodiments of the present application is to propose a neural network-based device detection method, device, computer equipment and media, which can avoid high-complexity operations in iterative algorithms and improve the detection accuracy of active users.

The neural network-based device detection method provided in the embodiment of the present application may be applied to a terminal, may also be applied to a server, and may also be software running on the terminal or the server. In some embodiments, the terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, or a smart watch; the server end can be configured as an independent physical server, or as a server cluster composed of multiple physical servers or as a distributed The system can also be configured to provide cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery network (Content Delivery Network, CDN) and large Cloud servers for basic cloud computing services such as data and artificial intelligence platforms; software can be applications that implement the above methods, but are not limited to the above forms.

The embodiments of the present application can be used in many general-purpose or special-purpose computer system environments or configurations. Examples: personal computers, server computers, handheld or portable devices, tablet-type devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer computing devices, network PCs, minicomputers, mainframe computers, including A distributed computing environment for any of the above systems or devices, etc. This application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including storage devices.

Please refer to FIG. 1 . FIG. 1 is a flowchart of a specific method of a neural network-based device detection method provided by an embodiment of the present application. In some embodiments, the neural network-based device detection method includes but not limited to steps S101 to S106.

Step S101, obtaining a training data set and a testing data set;

It should be noted that the training data set and the test data set are calculated by a preset data generation function on communication signals collected by the access point, and the access point includes at least one device.

In some embodiments, the training data set and the test data set are acquired to facilitate subsequent iteration and training of the neural network.

Step S102, constructing an indicator variable function according to the active state and delay value of the equipment in the training data set and the test data set;

Step S103, input the indicator variable function into the neural network for sum calculation, and output the covariance matrix;

Step S104, performing gradient calculation on the training data set and the test data set according to the covariance matrix, the preset first matrix, the preset second matrix and the preset learnable parameters to obtain gradient information;

Step S105, iteratively updating the gradient information and indicator variable functions based on the learnable parameters to obtain updated variables;

Step S106, iterating the neural network according to the updated variables, inputting the communication signal of the device to be detected into the iterated neural network for active detection, and determining the active state of the device to be detected.

In some embodiments, from step S101 to step S106, the training data set and the test data set are obtained, and an indicator variable function is constructed according to the active state and delay value of the equipment in the training data set and the test data set, so that the active equipment can be combined After the detection, the indicator variable function is input into the neural network for sum calculation to obtain the covariance matrix, which is convenient for subsequent training of the neural network. According to the covariance matrix, the preset first matrix, the preset second matrix and the preset possible The learning parameters perform gradient calculations on the training data set and the test data set to obtain gradient information. The introduction of learnable parameters and preset matrices can avoid inversion operations during the calculation process and reduce computational complexity. Then, the gradient information is calculated based on the learnable parameters. And the indicator variable function is iteratively updated to obtain the updated variable and improve the detection performance of the neural network for the device. Finally, iterate the neural network according to the updated variable, and input the device to be detected into the iterated neural network for active detection to determine the detection performance of the device to be detected. The active state of the device, thereby reducing the number of iterations of the neural network and improving the detection ability of the neural network for active devices.

Please refer to FIG. 2 . FIG. 2 is a flowchart of a specific method of a neural network-based device detection method provided by another embodiment of the present application. In some embodiments, the neural network-based device detection method includes but not limited to steps S201 to S202.

Step S201, obtaining the characteristic sequence of the device and the delay value corresponding to the characteristic sequence;

In step S202, an equivalent signature sequence of the device is obtained according to the signature sequence and the delay value.

其中，L为特征序列长度，由于非同步传输的特征导致特征序列的发送伴随一个未知的时延t_k∈{0,…,T}，其中，T代表最大可能的时延，因此设备的带有时延的等效特征序列可以如下公式(1)所示：In some embodiments, from step S201 to step S202, in the process of devices sending communication signals, each device has a unique signature sequence

Among them, L is the length of the characteristic sequence. Due to the characteristics of asynchronous transmission, the transmission of the characteristic sequence is accompanied by an unknown time delay t _k ∈ {0,...,T}, where T represents the maximum possible time delay. Therefore, the bandwidth of the device The equivalent feature sequence with time delay can be shown in the following formula (1):

It can be understood that when the device k is in the active state, it is recorded as a _k =1, otherwise a _k =0. Therefore, the active state of the device and the delay constitute an indicator variable function, as shown in the following formula (2):

It should be noted that b _k,t =1 if and only when device k is active and the time delay is t. Therefore, the detection of active devices by access points in asynchronous large-scale machine-to-machine communication can be mathematically equivalent to the detection of b _k,t ∈ {0,1}. Specifically, each access point transmits the received communication signal to the central processing unit through the fronthaul link, and then the central processing unit performs joint active device detection. This active device detection task is usually modeled as a high-dimensional non-convex problem. To solve such non-convex problems, an iterative algorithm needs to be designed, and a specific description is given below.

Please refer to FIG. 3 . FIG. 3 is a specific flowchart of step S101 provided by the embodiment of the present application. In some embodiments, step S101 specifically includes but not limited to step S301 and step S306.

Step S301, acquiring small-scale fading channel information between the device and the access point and the transmit power of the device;

以及设备的发射功率β_k，其中，β_k＝23dBm。In some embodiments, the small-scale fading channel information between the kth device and the mth access point is obtained

and the transmit power β _k of the device, where β _k =23dBm.

It should be noted that each element in the small-scale fading channel information obeys the Rayleigh distribution.

Step S302, calculating the distance between the device and the access point to obtain a distance value;

In some embodiments, the distance between the device and the access point is calculated to obtain a distance value d.

Step S303, inputting the distance value into a preset loss model to obtain large-scale fading information;

In some embodiments, the distance value d is input into a preset loss model g _k,m =128.1+37.6log ₁₀ (d) to obtain large-scale fading information g _k,m .

Step S304, generating a first function according to the transmit power, large-scale fading information and equivalent feature sequence;

In some embodiments, the first function z _{k, t,m} is formed according to the transmit power β k , the large-scale fading information g _k,m and the equivalent feature sequence s k _,t .

Step S305, obtaining a data generating function according to the small-scale fading channel information, the first function and the preset Gaussian distribution value;

第一函数z_k,t,m以及预设的高斯分布值W_m得到数据生成函数Y_m，其中，具体过程如下公式(3)所示：In some embodiments, according to the small-scale fading channel information

The first function z _{k, t, m} and the preset Gaussian distribution value W _m obtain the data generation function Y _m , where the specific process is shown in the following formula (3):

It should be noted that each element in W _m obeys a Gaussian distribution with a mean of 0 and a variance of 1.

Step S306, input the communication signal collected by the access point into the data generation function for calculation, and output the training data set and the testing data set.

In some embodiments, the communication signal collected by the access point is input into the data generation function for calculation, and the training data set and the test data set are output, wherein, the number of samples in the training data set and the test data set can be based on the user's In fact, it needs to be set by itself. In this embodiment, 1,280,000 training samples are independently generated according to formula (3) each time to form a training data set, and 100,000 verification set samples are generated to form a test data set. Each training cycle includes 10,000 cycles. The data in the loop is transferred to the neural network in batches, and the batch size is 128 at a time.

Please refer to FIG. 4 . FIG. 4 is a specific flowchart of step S103 provided by the embodiment of the present application. In some embodiments, step S103 specifically includes but not limited to step S401 and step S402.

Step S401, acquiring the noise power of the access point;

Step S402, input the noise power, the first function, the indicator variable function and the preset identity matrix into the neural network for sum calculation, and output the covariance matrix.

并对第一函数进行共轭转置计算，得到/>

将噪声功率/>

第一函数z_k,t,m、指示变量函数/>

以及预设的单位矩阵I_L+T均输入神经网络，并将第一函数z_k,t,m、共轭转置后的第一函数/>

以及指示变量函数/>

进行求和累加，并与噪声功率/>

和预设的单位矩阵I_L+T的乘积进行相加，得到协方差矩阵/>

具体如下公式(4)所示：In some embodiments, from step S401 to step S402, the noise power of the access point is obtained

And perform the conjugate transpose calculation on the first function, get />

will noise power />

first function z _k,t,m , indicator variable function/>

and the preset identity matrix I _L+T are input into the neural network, and the first function z _k,t,m and the first function after conjugate transposition/>

and the indicator variable function />

is summed and accumulated, and compared with the noise power />

Add the product of the preset identity matrix I _L+T to get the covariance matrix />

The details are shown in the following formula (4):

Please refer to FIG. 5 , which is a specific flow chart of step S104 provided by the embodiment of the present application. In some embodiments, step S104 specifically includes but not limited to steps S501 to S502.

It should be noted that the learnable parameters include penalty factors.

Step S501, performing a matrix approximation operation on the covariance matrix according to the first matrix and the second matrix to obtain an approximate matrix;

进行矩阵近似操作，得到近似矩阵

从而减少传统迭代算法的计算复杂度，避免了矩阵的求逆操作，提高设备的检测效率。In some embodiments, since the original liveness detection algorithm needs to invert the covariance matrix, resulting in higher computational complexity and a higher number of iterations, which does not meet the needs of practical applications, therefore, this embodiment The covariance matrix is paired by the first matrix A ⁽ⁱ⁾ and the second matrix B ⁽ⁱ⁾

Perform matrix approximation operations to obtain approximate matrices

Therefore, the calculation complexity of the traditional iterative algorithm is reduced, the inverse operation of the matrix is avoided, and the detection efficiency of the device is improved.

It should be noted that the dimensions of the first matrix and the second matrix need to be consistent with the covariance matrix.

Step S502, performing gradient calculation on the training data set and the testing data set according to the approximate matrix and the penalty factor, to obtain gradient information.

以及惩罚因子ρ对训练数据集以及测试数据集进行梯度计算，得到梯度信息/>

从而避免了矩阵的求逆操作，减少计算复杂度，并且引入惩罚因子提高神经网络的学习能力，提高计算的准确性，其中，具体过程如下公式(5)所示：In some embodiments, according to the approximation matrix

And the penalty factor ρ calculates the gradient of the training data set and the test data set to obtain the gradient information />

In this way, the inversion operation of the matrix is avoided, the computational complexity is reduced, and the penalty factor is introduced to improve the learning ability of the neural network and improve the accuracy of calculation. The specific process is shown in the following formula (5):

代表目标函数在/>

处的梯度信息，其中，目标函数的具体表达形式可以根据基于协方差的极大似然准则推到得出，本实施例不做具体限制。It can be understood that N represents the number of antennas of each access point,

represents the objective function at />

Gradient information at , where the specific expression form of the objective function can be deduced according to the maximum likelihood criterion based on covariance, which is not specifically limited in this embodiment.

Please refer to FIG. 6 . FIG. 6 is a specific flowchart of step S105 provided by the embodiment of the present application. In some embodiments, step S105 specifically includes but not limited to step S601 to step S602.

It should be noted that the learnable parameters include the iteration step size.

Step S601, multiplying the iteration step size and the gradient information to obtain a product result;

In step S602, an update variable is obtained according to the indicator variable function and the result of the product.

进行相乘，得到乘积结果，并根据指示变量函数/>

以及乘积结果得到更新变量/>

其中，具体过程如下公式(6)所示：In some embodiments from step S601 to step S602, iterative step size n and gradient information

Multiply, get the product result, and according to the indicator variable function />

and the result of the product gets the update variable />

Among them, the specific process is shown in the following formula (6):

代表算法一次梯度下降后的更新变量。Among them, η ⁽ⁱ⁾ represents the step size of the ith iteration,

Represents the updated variable after one gradient descent of the algorithm.

Please refer to FIG. 7 . FIG. 7 is a specific flowchart of step S106 provided by the embodiment of the present application. In some embodiments, step S106 specifically includes but not limited to steps S701 to S703.

Step S701, generating an iterative formula according to the update variables and learnable parameters;

In some embodiments, an iterative formula is generated according to the update variable and the learnable parameter, wherein, specifically, it is shown in the following formula (7):

Π_[0,1]代表min(max(·,0),1)。in,

Π _[0,1] represents min(max(·,0),1).

Step S702, iterating the neural network according to the iteration formula;

In some embodiments, the neural network is iterated according to the iterative formula, and an iterative process of the iterative algorithm is regarded as a layer of the neural network, the convergence can be completed through fewer layers, and learnable parameters are introduced in the iterative process to pass Supervised learning is used for training to improve the detection accuracy of the device.

Step S703, input the communication signal of the device to be detected into the iterative neural network to perform activity detection, and determine the active state of the device to be detected.

从而确定待检测设备的活跃状态。In some embodiments, the communication signal of the device to be detected is input into the iterative neural network to detect the state of the device, wherein the communication signal of the device to be detected carries the characteristic sequence of the device to be detected, and the characteristic sequence of the device to be detected is obtained and The delay value corresponding to the characteristic sequence constitutes the equivalent characteristic sequence of the device to be detected, and is calculated according to formulas (3) to (7), and the output

Thereby determining the active state of the device to be detected.

In order to more clearly illustrate the process of the neural network-based device detection method, a specific example will be used below.

Example one:

Referring to FIG. 8, FIG. 8 is a schematic diagram of a neural network-based device detection method provided in a specific example of the present application;

First, the communication signal collected by the access point is input into the data generation function for calculation, and the training data set and the test data set are output. The number of samples in the training data set and the test data set can be set according to the actual needs of the user. , in this embodiment, each time according to formula (3), 1,280,000 training samples are independently generated to form a training data set, and 100,000 verification set samples are generated to form a test data set. Each training cycle includes 10,000 cycles, and the data in each cycle is according to The form of batches is imported into the neural network, and the batch size at a time is 128. Then, based on the device detection method in this embodiment, the training data set and the test data set are input into the neural network for training. The active state and the delay value construct the indicator variable function, and initialize the indicator variable function, input the indicator variable function into the neural network for sum calculation, and obtain the covariance matrix, and then introduce the first matrix and the second matrix to carry out the covariance matrix Matrix approximation operation to obtain an approximate matrix, and calculate the gradient of the training data set and the test data set through the learnable penalty factor to obtain the gradient information, and then iteratively update the gradient information and the indicator variable function based on the learnable parameters to obtain the updated variable , so that the computational complexity of the traditional iterative algorithm can be effectively reduced. Finally, the neural network is iterated according to the update variable, and an iterative process of the iterative algorithm is regarded as a layer of the neural network, and the device to be tested is input into the iterative neural network for Active detection, to determine the active state of the device to be detected, and improve the active device detection capability of the network.

Referring to FIG. 8 , it can be seen that the neural network-based device detection method in this embodiment has good convergence, and requires fewer layers to complete the convergence, where I is a different number of layers of the neural network.

Referring to FIG. 9, FIG. 9 is a schematic diagram of a neural network-based device detection method provided in another specific example of the present application;

Referring to FIG. 9, it can be seen that, compared with the existing coordinate descent algorithm and the penalty function method, the present application has a lower false alarm probability and a lower probability of missed detection, so the method of this embodiment can improve the detection accuracy of the active state of the device accuracy.

Please refer to FIG. 10. FIG. 10 is a block diagram of a module structure of a device detection device based on a neural network provided by an embodiment of the present application, which is used to implement the device detection method based on a neural network in any of the above embodiments. The device includes:

The data acquisition module 801 is configured to acquire a training data set and a test data set, wherein the training data set and the test data set are calculated by a preset data generation function on communication signals collected by the access point, and the access point includes at least one device ;

A function construction module 802, configured to construct an indicator variable function according to the active state and the delay value of the equipment in the training data set and the test data set;

The iterative calculation module 803 is used to input the indicator variable function into the neural network for sum calculation, and output the covariance matrix;

The gradient calculation module 804 is used to perform gradient calculation on the training data set and the test data set according to the covariance matrix, the preset first matrix, the preset second matrix and the preset learnable parameters to obtain gradient information;

An iterative update module 805, configured to iteratively update gradient information and indicator variable functions based on learnable parameters to obtain update variables;

The active detection module 806 is configured to iterate the neural network according to the updated variables, and input the device to be detected into the iterated neural network for liveness detection, so as to determine the active state of the device to be detected.

The device detection device based on neural network in the embodiment of the present application is used to implement the device detection method based on neural network in the above embodiment, and its specific processing process is the same as the device detection method based on neural network in the above embodiment, which is not mentioned here Let me repeat them one by one.

The embodiment of the present application also provides a computer device, including a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor is used to execute the neural network-based neural network in the above embodiments of the present application. Device detection method.

The hardware structure of the computer device will be described in detail below in conjunction with FIG. 11 . The computer device includes: a processor 910 , a memory 920 , an input/output interface 930 , a communication interface 940 and a bus 950 .

The processor 910 may be implemented by a general-purpose CPU (Central Processing Unit, central processing unit), a microprocessor, an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute Relevant programs to realize the technical solutions provided by the embodiments of the present application;

The memory 920 may be implemented in the form of a read only memory (Read Only Memory, ROM), a static storage device, a dynamic storage device, or a random access memory (Random Access Memory, RAM). The memory 920 can store operating systems and other application programs. When implementing the technical solutions provided by the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 920 and called by the processor 910 to execute the implementation of the present application. Example of device detection method based on neural network;

The input/output interface 930 is used to realize information input and output;

The communication interface 940 is used to realize the communication interaction between this device and other devices, which can realize communication through wired methods (such as USB, network cable, etc.), or can realize communication through wireless methods (such as mobile network, WIFI, Bluetooth, etc.); and bus 950, transmitting information between various components of the device (such as the processor 910, the memory 920, the input/output interface 930, and the communication interface 940);

The processor 910 , the memory 920 , the input/output interface 930 and the communication interface 940 are connected to each other within the device through the bus 950 .

The embodiment of the present application also provides a storage medium, the storage medium is a computer-readable storage medium, and the computer-readable storage medium stores a computer program. When the computer program is executed by a computer, the computer is used to execute the A neural network-based device detection method.

As a non-transitory computer-readable storage medium, memory can be used to store non-transitory software programs and non-transitory computer-executable programs. In addition, the memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory optionally includes memory located remotely from the processor, and these remote memories may be connected to the processor via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The embodiments described in the embodiments of the present application are to illustrate the technical solutions of the embodiments of the present application more clearly, and do not constitute a limitation to the technical solutions provided by the embodiments of the present application. Those skilled in the art know that with the evolution of technology and new For the emergence of application scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

Those skilled in the art can understand that the technical solutions shown in Figures 1 to 7 do not constitute limitations on the embodiments of the present application, and may include more or fewer steps than those shown in the illustrations, or combine certain steps, or different steps.

The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, the functional modules/units in the system, and the device can be implemented as software, firmware, hardware, and an appropriate combination thereof.

The terms "first", "second", "third", "fourth", etc. (if any) in the description of the present application and the above drawings are used to distinguish similar objects and not necessarily to describe specific sequence or sequence. It is to be understood that the data so used are interchangeable under appropriate circumstances such that the embodiments of the application described herein can be practiced in sequences other than those illustrated or described herein. Furthermore, the terms "comprising" and "having", as well as any variations thereof, are intended to cover a non-exclusive inclusion, for example, a process, method, system, product or device comprising a sequence of steps or elements is not necessarily limited to the expressly listed instead, may include other steps or elements not explicitly listed or inherent to the process, method, product or apparatus.

It should be understood that in this application, "at least one (item)" means one or more, and "multiple" means two or more. "And/or" is used to describe the association relationship of associated objects, indicating that there can be three types of relationships, for example, "A and/or B" can mean: only A exists, only B exists, and A and B exist at the same time , where A and B can be singular or plural. The character "/" generally indicates that the contextual objects are an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one item (piece) of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c ", where a, b, c can be single or multiple.

In the several embodiments provided in this application, it should be understood that the disclosed devices and methods may be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or May be integrated into another system, or some features may be ignored, or not implemented. In another point, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be in electrical, mechanical or other forms.

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 may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

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

If the integrated unit is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or part of the contribution 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 storage medium , including multiple instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM for short), random access memory (Random Access Memory, RAM for short), magnetic disk or optical disk, etc., which can store programs. medium.

The preferred embodiments of the embodiments of the present application have been described above with reference to the accompanying drawings, which does not limit the scope of rights of the embodiments of the present application. Any modifications, equivalent replacements and improvements made by those skilled in the art without departing from the scope and essence of the embodiments of the present application shall fall within the scope of rights of the embodiments of the present application.

Claims (10)

1. A neural network-based device detection method, characterized in that the method comprises: Acquiring a training data set and a test data set, wherein the training data set and the test data set are calculated by a preset data generation function on communication signals collected by an access point, and the access point includes at least one device; Constructing an indicator variable function according to the active state and delay value of the equipment in the training data set and the test data set; Input the indicator variable function into the neural network for sum calculation, and output the covariance matrix; performing gradient calculation on the training data set and the test data set according to the covariance matrix, the preset first matrix, the preset second matrix and the preset learnable parameters to obtain gradient information; Iteratively updating the gradient information and the indicator variable function based on the learnable parameters to obtain updated variables; The neural network is iterated according to the update variable, and the communication signal of the device to be detected is input into the iterated neural network for active detection, and the active state of the device to be detected is determined. 2. The device detection method according to claim 1, further comprising: Obtaining a characteristic sequence of the device and a delay value corresponding to the characteristic sequence; An equivalent signature sequence of the device is obtained according to the signature sequence and the delay value. 3. The device detection method according to claim 2, wherein the training data set and the test data set are obtained according to the following steps: Obtaining small-scale fading channel information between the device and the access point and transmit power of the device; calculating the distance between the device and the access point to obtain a distance value; inputting the distance value into a preset loss model to obtain large-scale fading information; generating a first function according to the transmit power, the large-scale fading information, and the equivalent eigensequence; Obtaining the data generation function according to the small-scale fading channel information, the first function, and a preset Gaussian distribution value; The communication signal collected by the access point is input into the data generation function for calculation, and the training data set and the test data set are output. 4. The device detection method according to claim 3, wherein the input of the indicator variable function into the neural network is summed and calculated, and the covariance matrix is output, comprising: obtaining the noise power of the access point; Inputting the noise power, the first function, the indicator variable function, and a preset identity matrix into the neural network for sum calculation, and outputting a covariance matrix. 5. The device detection method according to claim 1, wherein the learnable parameters include a penalty factor; the covariance matrix, the preset first matrix, the preset second matrix and the preset The set learnable parameters carry out gradient calculation on the training data set and the test data set to obtain gradient information, including: performing a matrix approximation operation on the covariance matrix according to the first matrix and the second matrix to obtain an approximate matrix; Perform gradient calculation on the training data set and the test data set according to the approximation matrix and the penalty factor to obtain the gradient information. 6. The device detection method according to claim 1, wherein the learnable parameters include an iterative step size; the gradient information and the indicator variable function are iteratively updated based on the learnable parameters, Get updated variables, including: multiplying the iteration step size by the gradient information to obtain a product result; The update variable is obtained according to the indicator variable function and the product result. 7. The device detection method according to claim 1, wherein the neural network is iterated according to the update variable, and the device to be detected is input into the iterated neural network for active detection, and the Active status of the device to be detected, including: generating an iterative formula according to the update variable and the learnable parameter; Iterating the neural network according to the iterative formula; Inputting the device to be detected into the iterated neural network for active detection, and determining the active state of the device to be detected. 8. A device detection device based on a neural network, characterized in that the device comprises: A data acquisition module, configured to acquire a training data set and a test data set, wherein the training data set and the test data set are obtained by calculating the communication signals collected by the access point by a preset data generation function, and the access A point includes at least one device; A function construction module, configured to construct an indicator variable function according to the active state and delay value of the equipment in the training data set and the test data set; An iterative calculation module, configured to input the indicator variable function into the neural network for sum calculation, and output a covariance matrix; A gradient calculation module, configured to perform gradient calculation on the training data set and the test data set according to the covariance matrix, the preset first matrix, the preset second matrix and the preset learnable parameters, to obtain gradient information; an iterative update module, configured to iteratively update the gradient information and the indicator variable function based on the learnable parameters to obtain updated variables; The active detection module is configured to iterate the neural network according to the update variable, and input the communication signal of the device to be detected into the iterated neural network for active detection, and determine the active state of the device to be detected. 9. A computer device, characterized in that the computer device includes a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor is used to execute The method according to any one of claims 1 to 7. 10. A storage medium, wherein the storage medium is a computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a computer, the A computer is used to execute the method according to any one of claims 1-7.

Images