Cloud intelligence digital information network transmission system

Abstract

The invention discloses a cloud wisdom digital information network transmission system, and relates to the field of information network transmission, and the key points of the technical scheme are that the method comprises the steps: collecting the original information and transmission path information of an initial data packet, carrying out the preliminary preprocessing of the initial data packet at a transmission starting end point through employing a forward error correction technology, and obtaining an actual data packet; extracting transmission path information of each actual data packet, constructing a transmission directed graph, and obtaining an optimal transmission path of the actual data packet in the transmission directed graph by using an optimized particle swarm algorithm; controlling the number of actual data packets at each transmission starting endpoint and transmission ending endpoint by dynamically adjusting the sizes of the starting window and the ending window, and sequentially performing node transmission on the actual data packets according to the optimal transmission path until the transmission ending endpoint is stopped; a visual tool is provided for checking the transmission process in real time, interface interaction is carried out, and optimized transmission of the cloud intelligent digital information network is achieved.

Description

Technical Field

[0001] The present invention relates to the field of information network transmission, and more specifically, to a cloud intelligence digital information network transmission system. Background Art

[0002] In the process of cloud wisdom digital information network transmission, data loss or damage is a key issue, which may lead to data integrity damage, and then cause system failure or business interruption. Data is modified, truncated or lost during transmission, resulting in data inaccuracy or incompleteness. Network transmission error signal interference (such as electromagnetic interference, signal attenuation in wireless networks, etc.), transmission medium defects (aging or damage of physical lines, such as optical fiber breakage, copper cable damage) and insufficient transmission protocol (the protocol used fails to effectively detect and correct errors) cause data packets to be lost or erroneously transmitted during the transmission from the source node to the target node due to network instability or signal interference. Once data is lost or damaged, recovery work may be complicated and time-consuming. Data problems will increase additional maintenance and repair work, affecting overall operational efficiency.

[0003] Therefore, based on the above problems, it is necessary to design a cloud intelligence digital information network transmission system. Summary of the invention

[0004] In view of the problems existing in the prior art, the purpose of the present invention is to provide a cloud intelligence digital information network transmission system to achieve optimized transmission of the cloud intelligence digital information network.

[0005] To achieve the above object, the present invention provides the following technical solution: the cloud wisdom digital information network transmission system comprises:

[0006] Pre-preparation module: collects the original information and transmission path information of the initial data packet, and uses forward error correction technology to pre-process the initial data packet at the transmission starting endpoint to obtain the actual data packet;

[0007] Transmission design module: extracts the transmission path information of each actual data packet, constructs a transmission directed graph, and uses the optimized particle swarm algorithm to obtain the optimal transmission path of the actual data packet in the transmission directed graph;

[0008] Effective transmission module: By dynamically adjusting the size of the start window and the end window, the number of actual data packets at each transmission start endpoint and transmission end endpoint is controlled. The actual data packets are transmitted from node to node in sequence according to the optimal transmission path until they stop at the transmission end endpoint.

[0009] Effective output module: provides visualization tools to view the transmission process in real time and conduct interface interaction.

[0010] Preferably, the original information of the initial data packet includes the attributes of the data packet, and the attributes of the data packet include the size, type and confidentiality requirement level of the data packet. The confidentiality requirement level is pre-set according to the confidentiality standard, and the transmission path information includes the transmission start endpoint, transmission end endpoint and transmission node.

[0011] Preferably, the method of performing preliminary preprocessing on the initial data packet using forward error correction technology comprises:

[0012] Use network monitoring tools to monitor the packet loss rate of the transmission network in real time , Delay time , jitter frequency and remaining bandwidth , and calculate the network evaluation index ,in, Indicates the size of the initial data packet, represents a constant, , and represents the weight coefficient;

[0013] The initial data packet is divided into blocks according to the confidentiality requirement level and network evaluation index. Small chunks of data;

[0014] For each small data block, a redundancy adding scheme for each small data block is obtained according to a machine learning algorithm;

[0015] Redundant information is added to each small block according to the redundant adding scheme, and all processed small blocks are sorted according to their order in the original data packet, integrated into a complete data packet and counted and verified to form an actual data packet.

[0016] Preferably, the method for processing the initial data packet in blocks according to the confidentiality requirement level and the network evaluation index comprises:

[0017] Extract the confidentiality requirement level of each part of the data packet and the network evaluation index of the current transmission;

[0018] A CART decision tree model is constructed, and the confidentiality requirement level and network evaluation index are used as segmentation features. The segmentation criterion is based on the Gini coefficient to measure the degree of chaos of the current node. , select the features and partition points that minimize the Gini coefficient for segmentation, recursively divide the samples formed by the data packet into right subsets and left subsets, and then continue to recursively divide each subset until the depth of the tree reaches the preset value;

[0019] in, represents the number of samples in the left subset after partitioning, represents the number of samples in the right subset after division, Indicates the total number of samples of the current node, represents the total number of categories in the left subset after partitioning, represents the total number of categories in the right subset after division, Represents the number of samples of category k in the left subset after the current node is divided, Indicates the number of samples of category k in the right subset after the current node is divided;

[0020] Set the input of the CART decision tree model as the initial data packet and the output as the divided data blocks, collect the initial data packets of the go_p group with known confidentiality requirement levels and network evaluation indexes and the divided data blocks, and use them to train the constructed CART decision tree model to obtain the trained CART decision tree model;

[0021] Use the trained CART decision tree model to process the data packets that need to be divided to obtain the divided data blocks.

[0022] Preferably, the method of obtaining a redundant addition scheme for each data block according to the neural network model specifically includes:

[0023] The neural network model is set to include an input layer, a hidden layer and an output layer. The input of the input layer is a feature vector of a data block, which includes the size and type of the data block, the bandwidth of the transmission network, the packet loss rate and the computing resources. The output of the output layer is a redundancy adding scheme for each data block, which includes the coding type, coding rate, coding lightness and coding parameters.

[0024] The task requirements of the neural network model are set as , the constraints are ;

[0025] in, represents the task requirement assessment index, , and , , and They represent weight coefficients respectively, Indicates the fault tolerance of the redundant addition scheme, Indicates the data recovery speed of the redundancy addition solution, represents a predetermined threshold, Indicates the index of a small block of data, Represents small chunks of data Memory usage resources, Represents small chunks of data The bandwidth overhead, Represents small chunks of data Fault tolerance Represents small chunks of data Real-time performance;

[0026] According to the input and output of the neural network model, historical data is collected to form a training set. The labeled training set is used to iteratively train the optimal redundant addition scheme of the neural network model in different scenarios. The common gradient descent is used to optimize the parameters of the model until the set number of iterations is reached, and the trained neural network model is obtained.

[0027] Input the feature vector of the current data block into the trained neural network model to obtain the required redundancy addition scheme.

[0028] Preferably, the method for constructing the transmission directed graph is:

[0029] The transmission directed graph is set to consist of a starting point, an intermediate point, an end point, and directed edges. The transmission starting endpoint of each data packet is extracted as the starting point, the transit station or processing node that can be passed during the transmission process is set as the intermediate point, and each transmission end point is the end point;

[0030] Set the directed edge as the transmission path, set the direction of the directed edge to be the same as the direction that can be transmitted to the node, and assign attribute restrictions to each directed edge. The attribute restrictions must not exceed the tolerable load, remaining available bandwidth, and maximum delay time. The attribute restrictions of each directed edge are updated in real time.

[0031] Preferably, the method of using the optimized particle swarm algorithm to obtain the best transmission path of the actual data packet in the transmission directed graph includes:

[0032] Get the actual data packets at each starting point in the transmission directed graph, and generate K_lu group particle groups based on the number of actual data packets K_lu using distributed parallelization. The moving route of each particle represents a transmittable directed edge.

[0033] Design a fitness function to evaluate the quality of each moving route. The fitness function ; The restriction condition is ;

[0034] in, represents the load that each directed edge can bear, represents the actual load of each directed edge, represents the remaining available bandwidth of each directed edge, represents the maximum delay time of each directed edge, Represents the actual delay time of each directed edge;

[0035] Initialize the position of each particle group as the starting point, set the algorithm parameters, calculate the fitness value of each particle's corresponding moving route according to the fitness function, record the individual optimal position of each particle in history and the optimal moving route among all particles, update the particle's speed and position, and stop when the maximum number of iterations is reached. For each particle group, record the optimal moving route among all particles as the optimal transmission path to the next node;

[0036] The best transmission paths of all particle swarms are collected, which are the best transmission paths corresponding to the actual data packets transmitted to the next node.

[0037] Preferably, the method for dynamically adjusting the sizes of the start window and the end window includes:

[0038] When the transmission start endpoint starts transmission, it sets an initial window size and uses the initial window size to transmit to the transmission end endpoint. The transmission end endpoint sends an ACK after completely receiving the data packet.

[0039] Set an interval time threshold. When the time it takes for the ACK sent by the transmission termination endpoint to reach the transmission start endpoint exceeds the interval time threshold, adjust the window size in a linear reduction manner to obtain the intermediate window size, and transmit with the intermediate window size.

[0040] When the difference between the time when the transmission termination endpoint sends the confirmation ACK to the transmission start endpoint and the interval time threshold is less than the calibration time, the window size is adjusted in an exponential growth manner to obtain the intermediate window size, and transmission is performed with the intermediate window size.

[0041] Preferably, the actual data packet is transmitted from node to node in sequence according to the optimal transmission path, including:

[0042] The actual data packet is transmitted from the transmission starting endpoint, and the optimal transmission path to the next node M is obtained according to the optimized particle swarm algorithm. When it reaches the next node M, the next node M is used as the transmission starting endpoint, and the optimal transmission path to the next node M+1 is obtained again according to the optimized particle swarm algorithm. The cycle continues until it reaches the transmission termination endpoint.

[0043] Preferably, the transmission termination endpoint is deployed with a check mechanism, which is used to calculate the redundant check code added in the redundant addition scheme, and when an actual data packet error or loss is detected, an automatic retransmission request mechanism is used to request data retransmission.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] In the present invention, the security of data during transmission is enhanced by adopting forward error correction technology and data block processing based on confidentiality requirement level. The optimized particle swarm algorithm is used to find the best transmission path in the constructed transmission directed graph to ensure that the data packet is transmitted through the optimal path in the network, avoid network congestion, and improve bandwidth utilization. Block processing and redundancy are added according to the confidentiality requirement level of the data packet to ensure that high confidentiality data has a higher level of protection during transmission. By real-time monitoring of network status (such as packet loss rate, delay, jitter, etc.), the transmission strategy is dynamically adjusted to adapt to changes in the network environment. An intuitive and easy-to-use interface interaction tool is provided to facilitate users to manage and monitor the entire transmission process.

[0046] Through the optimal path selection and dynamic adjustment of window size, transmission delay and waste of network resources are reduced. The use of optimization algorithms helps to disperse data traffic, avoid overload of a single path, and achieve network load balancing. The system is designed as multiple modules, each with clear functions, which is easy to maintain and upgrade.

[0047] Through this cloud-intelligent digital information network transmission system, information transmission can be carried out more efficiently, the adaptive optimization mechanism can continuously improve the overall efficiency of intelligent analysis, and multiple modules can collaborate with each other to achieve comprehensive data collection, processing and analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A schematic diagram of the structure of a cloud intelligence digital information network transmission system is proposed for the present invention;

[0049] Figure 2 It is a schematic diagram of the method applied to the cloud wisdom digital information network transmission system in the present invention. DETAILED DESCRIPTION

[0050] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0051] Embodiment 1

[0052] Reference Figure 1 and Figure 2 , Example 1 further illustrates a cloud intelligence digital information network transmission system proposed by the present invention.

[0053] In the process of cloud wisdom digital information network transmission, data loss or damage is a key issue, which may lead to data integrity damage, and then cause system failure or business interruption. Data is modified, truncated or lost during transmission, resulting in data inaccuracy or incompleteness. Network transmission error signal interference (such as electromagnetic interference, signal attenuation in wireless networks, etc.), transmission medium defects (aging or damage of physical lines, such as optical fiber breakage, copper cable damage) and insufficient transmission protocol (the protocol used fails to effectively detect and correct errors) cause data packets to be lost or erroneously transmitted during the transmission from the source node to the target node due to network instability or signal interference. Once data is lost or damaged, recovery work may be complicated and time-consuming. Data problems will increase additional maintenance and repair work, affecting overall operational efficiency;

[0054] Electromagnetic interference (EMI): This is the most common type of interference. Various electronic devices, power lines, radio transmitters, etc. will generate electromagnetic radiation, which may couple with data signals, causing signal distortion or drowning. This interference especially affects copper cable transmission because copper cables are more susceptible to electromagnetic waves. High-frequency signals are more susceptible to EMI than low-frequency signals.

[0055] Signal attenuation in wireless networks: Wireless signals are affected by various factors during propagation, such as distance, obstacles (buildings, trees, etc.), multipath effects (signal reflection and diffraction), etc., which cause signal strength to attenuate. Severely attenuated signals will cause the receiving end to be unable to correctly decode the data, resulting in packet loss or errors. In addition, wireless networks are also susceptible to interference from other wireless devices, such as microwave ovens, Bluetooth devices, etc.

[0056] Crosstalk: When multiple cables are laid in parallel, the signals between different cables may interfere with each other, resulting in signal distortion. This phenomenon is more common in copper cable transmission.

[0057] Atmospheric interference: For wireless transmission, especially in long-distance transmission, water vapor and dust in the atmosphere will absorb and scatter wireless signals, causing signal attenuation. Extreme weather phenomena such as lightning can also generate strong electromagnetic pulses, causing data transmission interruption.

[0058] Physical defects in the transmission medium are another important cause of data transmission errors. These defects can be fiber breaks, copper cable damage, connector failures and aging, etc.

[0059] Therefore, in order to address the problems of network transmission error signal interference and transmission medium defects, it is necessary to effectively select and avoid transmission paths and transmission nodes, and to reduce signal interference to reduce the risks of the transmission process.

[0060] The transmission protocol is responsible for the reliable transmission of data. If the error detection and correction mechanism of the protocol itself is insufficient, it will not be able to effectively deal with errors in the data transmission process. Specifically:

[0061] A. Insufficient error detection mechanism: The error detection mechanism of some transmission protocols is relatively simple, such as simple checksum, which cannot detect all types of errors.

[0062] B. Insufficient error correction mechanism: Some protocols only provide error detection function, but not error correction function. Once an error is detected, the data packet can only be discarded and retransmitted, which will reduce transmission efficiency.

[0063] C. Insufficient congestion control: Network congestion can lead to packet loss. If the congestion control mechanism of the transport protocol is insufficient, it will not be able to effectively cope with network congestion.

[0064] D. Protocol mismatch: Using mismatched protocols can lead to data transmission errors. For example, using the UDP protocol in an application that requires reliable transmission can lead to data loss.

[0065] These three factors are often interrelated. For example, electromagnetic interference may cause copper cable signal attenuation, thereby increasing the probability of transmission errors. Transmission medium defects may cause data packet loss, and insufficient error correction mechanisms in transmission protocols will exacerbate data loss. Therefore, building a reliable network transmission system requires comprehensive consideration of these three factors and improving the reliability and efficiency of the data link layer to reduce the occurrence of errors.

[0066] Pre-preparation module: collects the original information and transmission path information of the initial data packet, and uses forward error correction technology to pre-process the initial data packet at the transmission starting endpoint to obtain the actual data packet;

[0067] The original information of the initial data packet includes the attributes of the data packet. The attributes of the data packet include the size, type and confidentiality requirement level of the data packet. The confidentiality requirement level is pre-set through confidentiality standards. The transmission path information includes the transmission start endpoint, transmission end endpoint and transmission node.

[0068] The data types of different parts in the data packet (for example, image data, video data, and text data) may have different confidentiality requirements.

[0069] The confidentiality standard can be a regulation set by the management personnel based on the analysis of experimental data or experience, such as: highly confidential, confidential, and ordinary. These levels can be mapped to numerical values ​​(for example, highly confidential = 3, confidential = 2, and ordinary = 1). When packaging the data packet, each part is marked with the confidentiality requirement level according to the confidentiality standard, and the confidentiality requirement level is used as the attribute of the data packet.

[0070] Methods for pre-processing initial data packets using forward error correction techniques include:

[0071] Use network monitoring tools (such as SNMP, NetFlow, etc.) to monitor the packet loss rate of the transmission network in real time , Delay time , jitter frequency and remaining bandwidth , and calculate the network evaluation index ,in, Indicates the size of the initial data packet, Represents a constant, because redundant information is added to the data packet, assuming that the redundant data is a fraction of the data packet size. times, the packet size needs to be converted to bits (1 byte = 8 bits), , and Represents the weight coefficient, which is assigned different weights according to the application scenario. For example, for real-time video transmission, delay and jitter may be more important; while for file transmission, packet loss rate may be more critical;

[0072] The initial data packet is divided into blocks according to the confidentiality requirement level and network evaluation index. Small chunks of data;

[0073] For each small data block, a redundancy adding scheme for each small data block is obtained according to a machine learning algorithm;

[0074] Redundant information is added to each small block according to the redundant adding scheme, and all processed small blocks are sorted according to their order in the original data packet, integrated into a complete data packet and counted and verified to ensure data integrity. All added small blocks are integrated to form an actual data packet.

[0075] The redundancy adding scheme for each small data block is obtained according to the machine learning algorithm to realize the dynamic adjustment of data redundancy, and the current network status is evaluated according to the network evaluation index, and then the appropriate forward error correction technology coding scheme is selected. For example, a stronger coding scheme (such as LDPC) is used when the packet loss rate is high, and a weaker coding scheme (such as a simple RS code) is used when the packet loss rate is low. The confidentiality level is determined according to the confidentiality requirement level of the data, and then the data packet is divided into multiple layers, and the small blocks with higher confidentiality levels are placed in the inner layer with stronger forward error correction coding, and the small blocks with lower confidentiality levels are placed in the inner layer with weaker forward error correction coding. A multi-level forward error correction scheme is used to process redundant information of different levels in layers.

[0076] The method of processing the initial data packet into blocks according to the confidentiality requirement level and the network evaluation index includes:

[0077] Extract the confidentiality requirement level of each part of the data packet and the network evaluation index of the current transmission;

[0078] A CART decision tree model was constructed, with confidentiality requirement level and network evaluation index as segmentation features. The segmentation criteria were based on the Gini coefficient to measure the "chaos degree" of the current node. , select the features and partition points that minimize the Gini coefficient for segmentation, recursively divide the samples formed by the data packet into right subsets and left subsets, and then continue to recursively divide each subset until the depth of the tree reaches the preset value;

[0079] in, represents the number of samples in the left subset after partitioning, represents the number of samples in the right subset after division, Indicates the total number of samples of the current node, represents the total number of categories in the left subset after partitioning, represents the total number of categories in the right subset after division, Represents the number of samples of category k in the left subset after the current node is divided, Indicates the number of samples of category k in the right subset after the current node is divided;

[0080] Category refers to the different divisions formed by discrete values ​​of confidentiality requirements and continuous values ​​of network evaluation. Sample number refers to the number of unit data in a data packet. For example, in a video, single data refers to a single frame of image.

[0081] Set the input of the CART decision tree model as the initial data packet and the output as the divided data blocks, collect the initial data packets of the go_p group with known confidentiality requirement levels and network evaluation indexes and the divided data blocks, and use them to train the constructed CART decision tree model to obtain the trained CART decision tree model;

[0082] Use the trained CART decision tree model to process the data packets that need to be divided to obtain the divided data blocks.

[0083] The redundant addition scheme for each small data block is obtained according to the neural network model. The specific method includes:

[0084] The neural network model is set to include an input layer, a hidden layer and an output layer. The input of the input layer is a feature vector of a data block, which includes the size and type of the data block, the bandwidth of the transmission network, the packet loss rate and the computing resources. The output of the output layer is a redundancy adding scheme for each data block, which includes the coding type, coding rate, coding lightness and coding parameters.

[0085] The size of the data chunks will affect the choice of redundant coding. Larger data chunks may require more complex redundant coding strategies. The type of data (e.g., text, images, video, etc.) will affect the choice of redundant schemes. Certain types of data may be more suitable for a particular redundant coding. The bandwidth, latency, and packet loss rate of the network will affect the choice of redundant schemes. Networks with high packet loss rates may require stronger redundancy, while low-bandwidth networks may require more efficient redundant strategies. The available computing resources are also important. If computing resources are limited, you may need to choose a redundant coding scheme with lower computational overhead.

[0086] The task requirements of the neural network model are set as , the constraints are ;

[0087] in, represents the task requirement assessment index, , and , , and They represent weight coefficients respectively, and different numerical weights are assigned according to the application scenarios. Indicates the fault tolerance of the redundant addition scheme, Indicates the data recovery speed of the redundancy addition solution, represents a predetermined threshold, Indicates the index of a small block of data, Represents small chunks of data Memory usage resources, Represents small chunks of data The bandwidth overhead, Represents small chunks of data The fault tolerance capability is used to indicate the success rate of the redundancy scheme to effectively recover lost data. Represents small chunks of data Real-time performance, which is used to indicate the increased time delay when decoding the redundancy scheme;

[0088] According to the input and output of the neural network model, historical data is collected to form a training set. The redundancy adding scheme in the training set can be generated through historical experience or expert annotation, or through simulation or emulation to generate data. The best redundancy schemes in different environments are analyzed as labels. The labeled training set is used to iteratively train the neural network model for the optimal redundancy adding scheme in different scenarios. Common gradient descent (such as Adam, SGD, etc.) is used to optimize the model parameters until the set number of iterations is reached, and the trained neural network model is obtained.

[0089] Input the feature vector of the current data block into the trained neural network model to obtain the required redundancy addition scheme.

[0090] Transmission design module: extracts the transmission path information of each actual data packet, constructs a transmission directed graph, and uses the optimized particle swarm algorithm to obtain the optimal transmission path of the actual data packet in the transmission directed graph;

[0091] Method for constructing a transmission directed graph:

[0092] The transmission directed graph is set to consist of a starting point, an intermediate point, an end point and directed edges. The transmission starting endpoint of each data packet is extracted as the starting point. The transit stations or processing nodes that can be passed during the transmission process are set as intermediate points. These intermediate points may be network switches, routers or geographical intermediate locations. Each transmission end point is the end point.

[0093] Set the directed edge as a transmission path, such as a cable, optical fiber, or link. Set the direction of the directed edge to be the same as the direction in which the node can be connected, and assign attribute restrictions to each directed edge. The attribute restrictions are that the tolerable load, remaining available bandwidth, and maximum delay time cannot be exceeded.

[0094] The attribute limits of each directed edge are dynamically updated in real time, that is, dynamically updated according to real-time status information (such as node load and bandwidth usage), and the maximum delay time can be customized according to transmission requirements.

[0095] By using the improved optimization algorithm and combining factors such as dynamic load balancing, bandwidth utilization, and delay in network transmission, a dynamic path planning system is designed. The goal is to simultaneously plan transmission paths for multiple data packets and dynamically update the plan at each intermediate node to ultimately achieve efficient transmission.

[0096] Load refers to the amount of data that can be carried on the current link after deducting the actual amount of data. A link with too high a load will affect the transmission performance. Remaining available bandwidth refers to the remaining bandwidth of the link. If the bandwidth is insufficient, the transmission time will increase significantly. Maximum delay time Latency: The maximum delay time that can be accepted by link transmission is pre-defined. Longer links or congested links will have greater delays.

[0097] The particle swarm algorithm will be used to dynamically plan the best path for each time period. The optimized particle swarm algorithm is used to obtain the best transmission path for the actual data packet in the transmission directed graph. The specific methods include:

[0098] Get the actual data packets at each starting point in the transmission directed graph, and generate K_lu group particle groups based on the number of actual data packets K_lu using distributed parallelization. The moving route of each particle represents a transmittable directed edge.

[0099] Design a fitness function to evaluate the quality of each moving route. The fitness function ; The restriction condition is ;

[0100] in, represents the load that each directed edge can bear, represents the actual load of each directed edge, represents the remaining available bandwidth of each directed edge, represents the maximum delay time of each directed edge, Represents the actual delay time of each directed edge;

[0101] Initialize the position of each particle swarm as the starting point, set the algorithm parameters (such as the number of particles, acceleration constant, etc.), calculate the fitness value of each particle's corresponding moving route according to the fitness function, record the individual optimal position of each particle in history and the optimal moving route among all particles, update the particle's speed and position, and stop when the maximum number of iterations is reached. For each particle swarm, record the optimal moving route among all particles as the optimal transmission path to the next node;

[0102] The best transmission paths of all particle swarms are collected, which are the best transmission paths corresponding to the actual data packets transmitted to the next node.

[0103] By using the particle swarm optimization algorithm (PSO), combined with real-time factors such as load, bandwidth, and delay, the transmission paths of multiple data packets can be dynamically planned. This solution can effectively avoid transmission congestion, improve bandwidth utilization, and reduce delays. In addition, by optimizing the path in real time at each intermediate node, it can adapt to the dynamic changes of network status.

[0104] Effective transmission module: By dynamically adjusting the size of the start window and the end window, the number of actual data packets at each transmission start endpoint and transmission end endpoint is controlled. The actual data packets are transmitted from node to node in sequence according to the optimal transmission path until they stop at the transmission end endpoint.

[0105] Dynamically adjust the size of the start window and the end window. The specific methods include:

[0106] When the transmission start endpoint starts transmission, it sets an initial window size (for example, 10 data packets) and uses the initial window size to transmit to the transmission end endpoint. The transmission end endpoint sends an ACK after completely receiving the data packet.

[0107] Set an interval time threshold. When the time it takes for the ACK sent by the transmission termination endpoint to reach the transmission start endpoint exceeds the interval time threshold, adjust the window size in a linear reduction manner to obtain the intermediate window size, and transmit with the intermediate window size.

[0108] When the difference between the time when the transmission termination endpoint sends the confirmation ACK to the transmission start endpoint and the interval time threshold is less than the calibration time, the window size is adjusted in an exponential growth manner to obtain the intermediate window size, and transmission is performed with the intermediate window size.

[0109] The actual data packets are transmitted to nodes in sequence according to the best transmission path, including:

[0110] The actual data packet is transmitted from the transmission starting endpoint, and the optimal transmission path to the next node M is obtained according to the optimized particle swarm algorithm. When it reaches the next node M, the next node M is used as the transmission starting endpoint, and the optimal transmission path to the next node M+1 is obtained again according to the optimized particle swarm algorithm. The cycle continues until it reaches the transmission termination endpoint.

[0111] Effective output module: provides visualization tools to view the transmission process in real time and conduct interface interaction.

[0112] The transmission termination endpoint is deployed with a check mechanism, which is used to calculate the redundant check code added in the redundant addition scheme, and when an actual data packet error or loss is detected, an automatic retransmission request mechanism is used to request data retransmission.

[0113] Embodiment 2

[0114] Reference Figure 1 and Figure 2 , Example 2 further illustrates a cloud intelligence digital information network transmission system proposed by the present invention.

[0115] The cloud intelligence digital information network transmission method comprises the following steps:

[0116] Step G1: Collect the original information and transmission path information of the initial data packet;

[0117] Step G2: At the transmission start endpoint, the initial data packet is pre-processed using forward error correction technology to obtain the actual data packet;

[0118] Step G3: extract the transmission path information of each actual data packet and construct a transmission directed graph;

[0119] Step G4: Use the optimized particle swarm algorithm to obtain the best transmission path of the actual data packet in the transmission directed graph;

[0120] Step G5: Control the number of actual data packets at each transmission start endpoint and transmission end endpoint by dynamically adjusting the size of the start window and the end window;

[0121] Step G6: The actual data packet is transmitted from the transmission starting endpoint, and the optimal transmission path to the next node M is obtained according to the optimized particle swarm algorithm. When the next node M is reached, the next node M is used as the transmission starting endpoint, and the optimal transmission path to the next node M+1 is obtained again according to the optimized particle swarm algorithm, and the cycle is repeated until the transmission termination endpoint is reached.

[0122] A cloud intelligence digital information network transmission system using the method includes:

[0123] Pre-preparation module: collects the original information and transmission path information of the initial data packet, and uses forward error correction technology to pre-process the initial data packet at the transmission starting endpoint to obtain the actual data packet;

[0124] Transmission design module: extracts the transmission path information of each actual data packet, constructs a transmission directed graph, and uses the optimized particle swarm algorithm to obtain the optimal transmission path of the actual data packet in the transmission directed graph;

[0125] Effective transmission module: By dynamically adjusting the size of the start window and the end window, the number of actual data packets at each transmission start endpoint and transmission end endpoint is controlled. The actual data packets are transmitted from node to node in sequence according to the optimal transmission path until they stop at the transmission end endpoint.

[0126] Effective output module: provides visualization tools to view the transmission process in real time and conduct interface interaction;

[0127] The modules are connected to each other via wired and / or wireless means.

[0128] In addition, according to the implementation of the present application, the process described in the figure of a cloud wisdom digital information network transmission system can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, which stores machine-readable instructions, and the machine-readable instructions can be executed by a processor to execute instructions corresponding to the method steps provided in the present application. Of course, the architecture shown in the figure of a cloud wisdom digital information network transmission system is only exemplary. When implementing different devices, adaptive selection or adjustment can be made according to actual needs.

[0129] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0130] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A cloud intelligence digital information network transmission system, characterized in that: The cloud wisdom digital information network transmission system includes: Pre-preparation module: collects the original information and transmission path information of the initial data packet, and uses forward error correction technology to pre-process the initial data packet at the transmission starting endpoint to obtain the actual data packet; Transmission design module: extracts the transmission path information of each actual data packet, constructs a transmission directed graph, and uses the optimized particle swarm algorithm to obtain the optimal transmission path of the actual data packet in the transmission directed graph; Effective transmission module: By dynamically adjusting the size of the start window and the end window, the number of actual data packets at each transmission start endpoint and transmission end endpoint is controlled. The actual data packets are transmitted from node to node in sequence according to the optimal transmission path until they stop at the transmission end endpoint. Effective output module: provides visualization tools to view the transmission process in real time and conduct interface interaction.

2. According to claim 1, a cloud intelligence digital information network transmission system is characterized in that: The original information of the initial data packet includes the attributes of the data packet, which include the size, type and confidentiality requirement level of the data packet. The confidentiality requirement level is pre-set according to the confidentiality standard. The transmission path information includes the transmission start endpoint, transmission end endpoint and transmission node.

3. A cloud intelligence digital information network transmission system according to claim 2, characterized in that: The method for preliminarily preprocessing an initial data packet using a forward error correction technique comprises: Use network monitoring tools to monitor the packet loss rate of the transmission network in real time , Delay time , jitter frequency and remaining bandwidth , and calculate the network evaluation index ,in, Indicates the size of the initial data packet, represents a constant, , and represents the weight coefficient; The initial data packet is divided into blocks according to the confidentiality requirement level and network evaluation index. Small chunks of data; For each small data block, a redundancy adding scheme for each small data block is obtained according to a machine learning algorithm; Redundant information is added to each small block according to the redundant adding scheme, and all processed small blocks are sorted according to their order in the original data packet, integrated into a complete data packet and counted and verified to form an actual data packet.

4. A cloud intelligence digital information network transmission system according to claim 3, characterized in that: The method for processing the initial data packet in blocks according to the confidentiality requirement level and the network evaluation index comprises: Extract the confidentiality requirement level of each part of the data packet and the network evaluation index of the current transmission; A CART decision tree model is constructed, and the confidentiality requirement level and network evaluation index are used as segmentation features. The segmentation criterion is based on the Gini coefficient to measure the degree of chaos of the current node. , select the features and partition points that minimize the Gini coefficient for segmentation, recursively divide the samples formed by the data packet into right subsets and left subsets, and then continue to recursively divide each subset until the depth of the tree reaches the preset value; in, represents the number of samples in the left subset after partitioning, represents the number of samples in the right subset after division, Indicates the total number of samples of the current node, represents the total number of categories in the left subset after partitioning, represents the total number of categories in the right subset after division, Represents the number of samples of category k in the left subset after the current node is divided, Indicates the number of samples of category k in the right subset after the current node is divided; Set the input of the CART decision tree model as the initial data packet and the output as the divided data blocks, collect the initial data packets of the go_p group with known confidentiality requirement levels and network evaluation indexes and the divided data blocks, and use them to train the constructed CART decision tree model to obtain the trained CART decision tree model; Use the trained CART decision tree model to process the data packets that need to be divided to obtain the divided data blocks.

5. A cloud intelligence digital information network transmission system according to claim 4, characterized in that: The method of obtaining a redundant addition scheme for each small data block according to the neural network model specifically includes: setting the neural network model to include an input layer, a hidden layer and an output layer, the input of the input layer is a feature vector of the small data block, the feature vector includes the size and type of the small data block, the bandwidth of the transmission network, the packet loss rate and computing resources, the output of the output layer is a redundant addition scheme for each small data block, and the redundant addition scheme includes a coding type, a coding rate, a coding lightness and a coding parameter; The task requirements of the neural network model are set as , the constraints are ; in, represents the task requirement assessment index, , and , , and They represent weight coefficients respectively, Indicates the fault tolerance of the redundant addition scheme, Indicates the data recovery speed of the redundancy addition solution, represents a predetermined threshold, Indicates the index of a small block of data, Represents small chunks of data Memory usage resources, Represents small chunks of data The bandwidth overhead, Represents small chunks of data Fault tolerance Represents small chunks of data Real-time performance; According to the input and output of the neural network model, historical data is collected to form a training set. The labeled training set is used to iteratively train the optimal redundant addition scheme of the neural network model in different scenarios. The common gradient descent is used to optimize the parameters of the model until the set number of iterations is reached, and the trained neural network model is obtained. Input the feature vector of the current data block into the trained neural network model to obtain the required redundancy addition scheme.

6. A cloud intelligence digital information network transmission system according to claim 5, characterized in that: The method for constructing the transmission directed graph: The transmission directed graph is set to consist of a starting point, an intermediate point, an end point, and directed edges. The transmission starting endpoint of each data packet is extracted as the starting point, the transit station or processing node that can be passed during the transmission process is set as the intermediate point, and each transmission end point is the end point; Set the directed edge as the transmission path, set the direction of the directed edge to be the same as the direction that can be transmitted to the node, and assign attribute restrictions to each directed edge. The attribute restrictions must not exceed the tolerable load, remaining available bandwidth, and maximum delay time. The attribute restrictions of each directed edge are updated in real time.

7. A cloud intelligence digital information network transmission system according to claim 6, characterized in that: The method of using the optimized particle swarm algorithm to obtain the best transmission path of the actual data packet in the transmission directed graph includes: Get the actual data packets at each starting point in the transmission directed graph, and generate K_lu group particle groups based on the number of actual data packets K_lu using distributed parallelization. The moving route of each particle represents a transmittable directed edge. Design a fitness function to evaluate the quality of each moving route. The fitness function ; The restriction condition is ; in, represents the load that each directed edge can bear, represents the actual load of each directed edge, represents the remaining available bandwidth of each directed edge, represents the maximum delay time of each directed edge, Represents the actual delay time of each directed edge; Initialize the position of each particle group as the starting point, set the algorithm parameters, calculate the fitness value of each particle's corresponding moving route according to the fitness function, record the individual optimal position of each particle in history and the optimal moving route among all particles, update the particle's speed and position, and stop when the maximum number of iterations is reached. For each particle group, record the optimal moving route among all particles as the optimal transmission path to the next node; The best transmission paths of all particle swarms are collected, which are the best transmission paths corresponding to the actual data packets transmitted to the next node.

8. A cloud intelligence digital information network transmission system according to claim 7, characterized in that: The specific method of dynamically adjusting the sizes of the start window and the end window includes: When the transmission start endpoint starts transmission, it sets an initial window size and uses the initial window size to transmit to the transmission end endpoint. The transmission end endpoint sends an ACK after completely receiving the data packet. Set an interval time threshold. When the time it takes for the ACK sent by the transmission termination endpoint to reach the transmission start endpoint exceeds the interval time threshold, adjust the window size in a linear reduction manner to obtain the intermediate window size, and transmit with the intermediate window size. When the difference between the time when the transmission termination endpoint sends the confirmation ACK to the transmission start endpoint and the interval time threshold is less than the calibration time, the window size is adjusted in an exponential growth manner to obtain the intermediate window size, and transmission is performed with the intermediate window size.

9. A cloud intelligence digital information network transmission system according to claim 8, characterized in that: The method of transmitting the actual data packet to nodes in sequence according to the optimal transmission path includes: The actual data packet is transmitted from the transmission starting endpoint, and the optimal transmission path to the next node M is obtained according to the optimized particle swarm algorithm. When it reaches the next node M, the next node M is used as the transmission starting endpoint, and the optimal transmission path to the next node M+1 is obtained again according to the optimized particle swarm algorithm. The cycle continues until it reaches the transmission termination endpoint.

10. A cloud intelligence digital information network transmission system according to claim 9, characterized in that: The transmission termination endpoint is deployed with a check mechanism, which is used to calculate the redundant check code added in the redundant addition scheme, and when an actual data packet error or loss is detected, an automatic retransmission request mechanism is used to request data retransmission.

Images