Method and system for multi-channel distributed communication based on adaptive prediction

Abstract

The application provides a multi-channel distributed communication method and system based on adaptive prediction, which comprises the following steps: obtaining original first information data through a data collector. Then, in the receiving end, the first information data and corresponding second information data are received and stored in sequence through public and point-to-point private networks respectively. The second information data is processed by a prediction function to predict estimated information data, which is transmitted to an adaptive controller. The controller uses a filter containing an adaptive control rate to compensate the estimated information data and the first information data to obtain compensated information data. Through the compensated information data and the first information data, accurate receiving data is generated and sent. The communication method can effectively identify and predict possible fraud attacks and external disturbances in the communication process, and make adjustments in advance. Moreover, the method can bypass complex modeling design, greatly expanding the application range.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mobile communication, and in particular to a multi-channel distributed communication method and system based on adaptive prediction. BACKGROUND

[0002] With the rapid development of information technology and network communication technology, multi-channel distributed communication systems have been widely applied in various high-performance interactive systems, such as robots, unmanned aerial vehicles, multi-axis robotic arms, etc. These application scenarios not only require real-time transmission of communication data, but also require extremely high transmission quality (QoS). However, existing communication technologies still face many challenges when facing network spoofing attacks and external disturbances, especially in multi-channel communication systems. Spoofing attacks directly affect the reliability and security of communication by injecting false data, while external disturbances can cause data transmission to deviate from the predetermined target, and even cause system failure in severe cases.

[0003] Traditional communication systems mostly use single-input single-output (SISO) or multi-input multi-output (MIMO) systems based on accurate models for control design. Although these systems can achieve effective control in theory, they face the dual challenges of model accuracy and complexity in practical applications. Especially in large-scale communication systems, due to the uncertainty and complexity of the environment, traditional models often fail to accurately describe system behavior, thereby affecting control effectiveness.

[0004] In addition, most existing technologies rely on point-to-point communication mode, ignoring the actual needs of distributed multi-channel communication. Distributed multi-channel systems increase the risk of being attacked and disturbed due to their complexity and wide application background, therefore, there is an urgent need for a communication control method that can adapt to dynamic changes and intelligently predict future states. SUMMARY

[0005] The purpose of the present application is to provide a multi-channel distributed communication method and system based on adaptive prediction, which can accurately predict possible spoofing attacks and external disturbances encountered in communication and adaptively compensate for communication data with disturbances.

[0006] In order to achieve the above-mentioned purpose, the present application provides a multi-channel distributed communication method based on adaptive prediction, which comprises:

[0007] At the sending end, the target object is digitally sampled by a data collector to obtain original first information data;

[0008] At the receiving end, the data corresponding to the first information data is received and stored in sequence through a public communication network to obtain second information data, and the first information data is received and stored in sequence through a point-to-point private network;

[0009] selecting a target data set from the second information data, the target data set including current information data and several historical information data;

[0010] performing prediction processing on the information data in the target data set by a prediction function constructed based on a neural network, to estimate estimation information data corresponding to each of the second information data, and transmitting the estimation information data to an adaptive controller;

[0011] performing adaptive compensation processing on the estimation information data and the first information data by a filter with a preset adaptive control rate, to obtain compensation information data;

[0012] generating reception data according to the compensation information data and the first information data, and sending the reception data to an information receiver of the receiving end.

[0013] Preferably, the target data set includes three second information data, which are current information data and information data at two sampling time points adjacent to the current information data.

[0014] Preferably, the prediction function is a radial basis function network, and a basis function S(x i ) is as follows:

[0015]

[0016] n iq and m iq respectively represent a basis center quantity and a basis width of a q-dimensional neural network node of i, and are preset parameters, and O represents a number of neural networks;

[0017] In the radial basis function network, an expression of the estimation information data is as follows:

[0018]

[0019] wherein, an input vector x i (t) is a matrix: W represents a weight matrix in R OL*O space, and a gradient descent update law corresponding to the weight matrix satisfies the following formula:

[0020]

[0021] h i and λ i respectively represent a momentum factor and a learning rate of i, and are preset parameters;

[0022] In the above formula, t+1 represents the current sampling time, representing the second information data.

[0023] Preferably, the data model of the filter is as follows:

[0024]

[0025] wherein P and are the filtered ideal output and its derivative, respectively, R, Y are scale parameters, tanh() is the hyperbolic tangent function, and y i () represents the first information data;

[0026] The state quantity P generated based on the model of the filter above is the compensation information data.

[0027] Preferably, the adaptive controller generates the received data u i (t) based on the following formula:

[0028] u i (t) = -k i (p i (t+1) - y i (t+1)) + p i (t) - y i (t),

[0029] wherein k i is a preset scale parameter.

[0030] The present application also provides a multi-channel distributed communication system comprising an information processor arranged at each channel receiving end, the information processor comprising a prediction function and an adaptive controller constructed based on a neural network, and the information processor works based on the adaptive prediction-based multi-channel distributed communication method as described above.

[0031] The present application also provides a multi-channel distributed communication system comprising:

[0032] one or more processors;

[0033] a memory;

[0034] and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the programs comprise instructions for executing the adaptive prediction-based multi-channel distributed communication method as described above.

[0035] The present application also provides a computer readable storage medium comprising a computer program, which can be executed by a processor to complete the adaptive prediction-based multi-channel distributed communication method as described above.

[0036] Compared with the prior art, the communication method provided by the technical scheme of the present application effectively identifies and predicts the fraud attacks and external disturbances that may be encountered in the communication process by using the dynamic prediction capability of the prediction function constructed based on the neural network, and adjusts in advance based on the adaptive controller. This prediction and adjustment mechanism not only reduces the influence of adverse factors on the communication quality, but also ensures the accuracy of data transmission and enhances the anti-interference ability of the system through the real-time compensation function of the adaptive controller. In addition, by bypassing the complex modeling link, the present scheme directly designs the controller for the general MIMO nonlinear system or linear system, greatly expanding the application range. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 The execution flowchart of the communication method in the embodiment of the present application.

[0038] Figure 2 The principle structure diagram of the communication method in the embodiment of the present application. DETAILED DESCRIPTION

[0039] In order to explain the technical content, structural features, achieved purposes and effects of the present application in detail, the following will be described in detail in combination with the embodiments and the accompanying drawings.

[0040] The embodiment discloses a multi-channel distributed communication method for controlling the operation of a multi-channel communication system, thereby eliminating the fraud attacks and external disturbances that are prone to occur in the communication process.

[0041] As Figure 1 and Figure 2 The communication method in the embodiment includes the following steps:

[0042] S1: At the sending end, the target object is digitally sampled by a data collector, and the sampled data is generalized to match the communication system and satisfy the MIMO structure to obtain the original first information data.

[0043] S2: At the receiving end, the data corresponding to the first information data is received and stored in sequence through a public communication network to obtain second information data, and the first information data is received and stored in sequence through a point-to-point private network. Specifically, after obtaining the original first information data, it is transmitted through two lines, the first line being a public communication network, so that the second information data may be the same data as the first information data or data containing interference factors. The second line is a point-to-point private network, so there is no influence of other interference factors, and the data remains unchanged, still being the first information data. The second line in the embodiment is preferably the data link layer of the communication network.

[0044] S3: selecting a target data group from the second information data, the target data group including current information data and a number of historical information data.

[0045] S4: performing prediction processing on the information data in the target data group by a prediction function constructed based on a neural network, to estimate estimation information data corresponding to each of the second information data, and transmitting the estimation information data to an adaptive controller.

[0046] S5: the adaptive controller performing adaptive compensation processing on the estimation information data and the first information data by a filter with a preset adaptive control rate, to obtain compensation information data.

[0047] S6: the adaptive controller generating reception data according to the compensation information data and the first information data, and sending the reception data to an information receiver of the receiving end.

[0048] The execution process and principle of the above communication method will be described in detail below with a specific example.

[0049] Suppose that this adaptive prediction-based multi-channel distributed communication method is applied in an automated manufacturing factory. There are multiple robot arms in the factory performing precise assembly tasks, which require high-precision and high-reliability data communication to ensure the accuracy of operations.

[0050] I: data acquisition

[0051] At the sending end, a data collector is installed on one of the robot arms. The sensors of the robot arm monitor its operating state in real time and perform digital sampling to obtain original first information data such as the position, speed, and acceleration of the robot arm.

[0052] II: data transmission

[0053] These first information data are sent through the public communication network of the factory. Due to possible network delays and interference, these data are also sent through a more secure and faster point-to-point private network in parallel. The data transmitted by the public network are received and stored in sequence at the receiving end, referred to as second information data; the data transmitted by the private network are also stored in sequence for subsequent comparison and verification.

[0054] III: data group selection and prediction

[0055] At the receiving end, a target data group including current information data and related historical information data is selected from the stored second information data. These data are processed by a prediction function based on a neural network to detect possible errors or abnormalities in the current data.

[0056] Four: Data compensation

[0057] The estimated information data, i.e. the predicted robot arm operating state, is transmitted to an adaptive controller. The adaptive controller uses a filter with adaptive control rate to combine the original first information data and the estimated information data for adaptive compensation. The purpose of this step is to correct the deviation caused by communication errors or interference, ensuring the accuracy of the data.

[0058] Five: Generating received data

[0059] The compensated information data after adaptive controller processing, combined with the original first information data, generates the final received data. These received data represent the corrected accurate state information of the robot arm, which is then sent to the information receiver at the receiving end for subsequent operation or monitoring.

[0060] Through the above steps, even if interference or errors are encountered during data transmission, this communication method can ensure that the operating data of the robot arm remains highly accurate and reliable, effectively supporting complex industrial automation applications.

[0061] On the other hand, the target data set includes three second information data, namely the current information data and the information data of the previous two sampling time points adjacent to the current information data. That is, the three second information data closest to the current time are selected as the basis data to determine whether the latest second information data contains interference data.

[0062] Specifically, the prediction function is a radial basis function network, and the basis function S(x i ) is as follows:

[0063]

[0064] n iq and m iq represent the basis center quantity and basis width of the q-dimensional neural network node of i, respectively, and O represents the number of neural networks.

[0065] In the radial basis function network, the expression of the estimated information data is as follows:

[0066]

[0067] where the input vector x i (t) is a matrix: W represents the weight matrix in R OL*O space, and the gradient descent update law corresponding to the weight matrix satisfies the following formula:

[0068]

[0069] h i and λ i respectively represent momentum factor and learning rate of i, and are preset parameters.

[0070] It should be noted that in the above formula, t+1 represents the current sampling time, represents the second information data. Then, represents the second information data corresponding to the current sampling time, represents the second information data corresponding to the last sampling time, represents the second information data corresponding to the previous sampling time.

[0071] Further, the data model of the filter is as follows:

[0072]

[0073] wherein P and are the ideal output and its derivative after filtering processing, R, Y are scale parameters, tanh() is the hyperbolic tangent function, and y i represents the first information data.

[0074] The state quantity P generated based on the model of the above filter is the compensation information data

[0075] According to the model of the filter, when the estimated information data is obtained from the prediction function, and the corresponding first information data y i (t+1) are substituted into the above filter model, the compensation information data P can be solved. It can be known that when the communication data in the public network is not disturbed by external interference, then the solved compensation quantity is zero. If the communication data in the public network is disturbed by external interference, the corresponding compensation quantity is generated according to the difference between and y i (t+1).

[0076] On the other hand, when the compensation information data is obtained, the adaptive controller generates the received data u i (t) based on the following formula:

[0077] u i (t) = -k i (p i (t+1) - y i (t+1)) + p i (t) - y i (t).

[0078] wherein k i is a preset scale parameter.

[0079] In the embodiment, the estimated information data and the first information data y i (t+1) corresponding to the current sampling moment are substituted into the data model of the filter to solve P i (t+1), and the estimated information data and the first information data y i (t) corresponding to the last sampling moment are substituted into the data model of the filter to solve P i (t), then P i (t+1), P i (t), y i (t+1), y i (t) are substituted into the expression of u i (t) to obtain the final received data.

[0080] To sum up, the application discloses a multi-channel distributed communication method based on adaptive prediction, which effectively identifies and predicts possible fraud attacks and external disturbances encountered in the communication process by using the dynamic prediction ability of the prediction function constructed based on the neural network, and adjusts in advance based on the adaptive controller. This prediction and adjustment mechanism not only reduces the influence of adverse factors on the communication quality, but also ensures the accuracy of data transmission and enhances the anti-interference ability of the system through the real-time compensation function of the adaptive controller. In addition, the scheme directly controls the general MIMO nonlinear system or linear system by bypassing the complex modeling link, greatly expanding the application range.

[0081] In another preferred embodiment of the application, a multi-channel distributed communication system is also disclosed, which comprises an information processor arranged at each channel receiving end, the information processor comprising a prediction function constructed based on a neural network and an adaptive controller, and the information processor works based on the multi-channel distributed communication method based on adaptive prediction as described above.

[0082] The application also discloses another communication system, comprising one or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the programs comprise instructions for executing the communication method as described above. The processor can be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for executing related programs to achieve the functions required by the modules in the communication system of the embodiments of the application or to execute the communication method of the method embodiments of the application.

[0083] The application also discloses a computer-readable storage medium comprising a computer program, wherein the computer program can be executed by a processor to complete the communication method as described above. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media. The available medium can be a read-only memory (ROM), a random access memory (RAM), a magnetic medium such as a floppy disk, a hard disk, a magnetic tape, a magnetic disk, or an optical medium such as a digital versatile disc (DVD), or a semiconductor medium such as a solid state disk (SSD), etc.

[0084] The embodiments of the application also disclose a computer program product or a computer program, which comprises computer instructions stored in a computer-readable storage medium. The processor of the electronic device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions to enable the electronic device to execute the communication method.

[0085] The above only discloses the preferred embodiments of the application, and of course cannot limit the scope of the application, so the equivalent changes made in the patent scope of the application still fall within the scope of the application.

Claims

1. A method of multi-channel distributed communication based on adaptive prediction, characterized in that, The method comprises: At the sending end, a target object is digitally sampled by a data collector to obtain original first information data; At the receiving end, data corresponding to the first information data is received and stored in sequence through a public communication network to obtain second information data, and the first information data is received and stored in sequence through a point-to-point private network; A target data group is selected from the second information data, the target data group comprising current information data and a plurality of historical information data; The information data in the target data group is predicted by a prediction function based on a neural network to estimate estimated information data corresponding to each second information data, and the estimated information data is transmitted to an adaptive controller; The adaptive controller performs adaptive compensation processing on the estimated information data and the first information data through a filter with a preset adaptive control rate to obtain compensation information data; The adaptive controller generates receiving data according to the compensation information data and the first information data, and sends the receiving data to an information receiver at the receiving end; The prediction function is a radial basis function network, the basis functions are as follows: , and and denote the base center and base width of the q-dimensional neural network node of i, respectively, and O denotes the number of neural networks; In the radial basis function network, the estimated information data is expressed as follows: ， where the input vector is a matrix: ​ W represents a weight matrix under space, and the gradient descent update law corresponding to the weight matrix satisfies the following formula: ， and denote the momentum factor and learning rate for i, respectively, and are preset parameters; In the above equation, t+1 represents the current sampling time, represents the second information data; The data model of the filter is as follows: , wherein, and are respectively a filtered processed ideal output and its derivative, , Y is a scale parameter, tanh() is a hyperbolic tangent function, represent the first information data; state quantities generated on the basis of the model of the filter for the compensation information data.

2. The method of claim 1, wherein, The target data group includes three second information data, which are current information data and information data at the previous two sampling time points adjacent to the current information data.

3. The method of claim 1, wherein, The adaptive controller generates the received data based on the following equation : ， wherein is a predetermined scale parameter.

4. A multi-channel distributed communication system, characterized by The method comprises an information processor arranged at each channel receiving end, the information processor comprising a prediction function based on a neural network and an adaptive controller, and the information processor is based on the adaptive prediction-based multi-channel distributed communication method according to any one of claims 1 to 3.

5. A multi-channel distributed communication system, characterized by The method comprises: One or more processors; Memory; And one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs comprising instructions for executing the adaptive prediction-based multi-channel distributed communication method according to any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, The computer program can be executed by a processor to complete the adaptive prediction-based multi-channel distributed communication method according to any one of claims 1 to 3.

Images