P2p remote communication method

Abstract

The invention discloses a peer-to-peer (p2p) remote communication method, which relates to the technical field of communication, and comprises the following steps of: collecting network environment data and communication data, and preprocessing; inputting the preprocessed network environment data into a machine learning model, and predicting the network quality of each path; performing analysis through an intelligent scheduling algorithm based on a network quality prediction result to obtain a distribution strategy selected by each path; according to the distribution strategy selected by each path, splitting the communication data into a plurality of data blocks, and distributing the data blocks to each path in proportion for transmission; based on the transmission of the communication data, the receiving end node and the sending end node exchange network state information to dynamically adjust the distribution of the communication data; and the receiving end node carries out recombination according to the data blocks transmitted by each path to obtain complete communication data. By collecting and preprocessing the network environment and communication data, high-quality communication data input is ensured, and the network quality of each path is predicted by using the long-short-term memory network model.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, in particular to a p2p remote communication method. Background Art

[0002] With the continuous development of Internet technology and information communication networks, p2p communication has been widely used in video conferencing, file transfer, real-time communication and other fields. P2p communication faces many challenges in practical applications, especially in dynamic network environments. How to efficiently select transmission paths and ensure the reliability and real-time performance of data transmission has become an important factor affecting its performance.

[0003] Existing technologies also have shortcomings in bandwidth allocation. Most methods use a fixed-ratio allocation method, lacking intelligent scheduling algorithms and dynamic adjustment mechanisms, and are unable to cope with fluctuations in bandwidth and latency in complex network environments. Such methods often show low adaptability when facing environments with large changes in network quality, resulting in increased transmission delays and packet loss rates, affecting the stability and efficiency of communications. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for p2p remote communication to solve the problems of static bandwidth allocation and lack of intelligent scheduling and dynamic adjustment in the prior art.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for p2p remote communication, comprising collecting network environment data and communication data, and performing pre-processing;

[0008] Input the preprocessed network environment data into the machine learning model to predict the network quality of each path;

[0009] Based on the network quality prediction results, the intelligent scheduling algorithm is used to analyze and obtain the allocation strategy for each path selection;

[0010] According to the allocation strategy selected for each path, the communication data is split into multiple data blocks and proportionally allocated to each path for transmission;

[0011] Based on the transmission of communication data, the receiving and sending nodes exchange network status information to dynamically adjust the communication data allocation;

[0012] The receiving node reassembles the data blocks transmitted from each path to obtain complete communication data.

[0013] As a preferred solution of the p2p remote communication method of the present invention, the collecting of network environment data and communication data and preprocessing thereof include the following steps:

[0014] The collected network data includes bandwidth, delay and packet loss rate, and the collected communication data includes files, video streams and messages. All collected data are denoised and standardized.

[0015] As a preferred solution of the p2p remote communication method of the present invention, wherein: the pre-processed network environment data is input into the machine learning model, and the network quality of each path is predicted, including the following steps:

[0016] The bandwidth, delay and packet loss rate feature data are extracted from the preprocessed network environment data through the principal component analysis method and input into the long short-term memory network model to predict the network quality of each path. The expression is:

[0017]

[0018] Among them, Q i (t) is the network quality prediction value of the ith path at time t, x k (t) is the value of the kth network quality feature at time t, w k is the weight of the kth network quality feature value, z(t) is the sum of network quality features at time t, n is the number of paths, β is the adjustment parameter of the sum of network quality features, k is the network quality feature index, and f is the nonlinear activation function.

[0019] As a preferred solution of the p2p remote communication method of the present invention, wherein: based on the network quality prediction result, the intelligent scheduling algorithm is used to analyze and obtain the allocation strategy for each path selection, including the following steps:

[0020] Use SNMP protocol to extract the actual flow and maximum available bandwidth of each path from network devices to obtain bandwidth utilization;

[0021] The ICMP ping command is used to measure the round-trip time from the sender to the receiver on each path to obtain the transmission delay.

[0022] Based on the network quality prediction value, bandwidth utilization and transmission delay, the intelligent scheduling algorithm is used to dynamically predict the data of each path, and the allocation strategy for each path selection is obtained. The expression is:

[0023]

[0024] Among them, F(p) is the fitness score of the allocation strategy p, B i is the bandwidth utilization of the ith path, Di The transmission delay of the i-th path, α i The importance coefficient of the i-th path, μ is the average value of all paths on the network quality prediction value, γ is the sensitivity of the control to the change of network quality, δ is the reference point of network quality, j1 is the power of the network quality prediction value, o2 is the power of bandwidth utilization, l3 is the power of transmission delay, λ is the weight coefficient of controlling the transmission delay to the fitness score of the allocation strategy, A is the adjustment coefficient of controlling the sensitivity of network quality change, and p is the allocation strategy;

[0025] According to the fitness score of the allocation strategy, all paths are sorted from high to low using the quick sort method to obtain the allocation strategy selected for each path.

[0026] As a preferred solution of the p2p remote communication method of the present invention, wherein: according to the allocation strategy selected by each path, splitting the communication data into multiple data blocks includes the following steps:

[0027] According to the allocation strategy of each path, the communication data to be transmitted is split according to the weight ratio of the path;

[0028] Communication data is split into small data chunks over high-latency and low-bandwidth paths;

[0029] On low-latency and high-bandwidth paths, communication data is broken into large chunks.

[0030] As a preferred solution of the p2p remote communication method of the present invention, wherein: based on the split data blocks, allocating the split data blocks to each path for transmission in proportion includes the following steps:

[0031] According to the bandwidth utilization of each path, the bandwidth allocation ratio of each path is obtained;

[0032] The split data blocks are allocated to each path in proportion according to the bandwidth allocation ratio of each path;

[0033] The data blocks on each path are transmitted using a specified transport protocol.

[0034] As a preferred solution of the p2p remote communication method of the present invention, wherein: based on the transmission of communication data, the receiving end and the sending end node exchange network status information to dynamically adjust the communication data distribution, including the following steps:

[0035] The sender and receiver exchange real-time network status information through the control channel;

[0036] The sender encapsulates all network status information into data packets and sends them to the receiver;

[0037] After receiving the network status information from the sender, the receiver decodes the network performance of each path, extracts bandwidth, delay, and packet loss rate data, and calculates the network health status of each path. The expression is:

[0038]

[0039] Among them, H i (t) is the network health score of the ith path at time t, B max is the maximum bandwidth of all paths, D max is the maximum delay of all paths, κ is the path bandwidth weight coefficient, θ is the delay weight coefficient, φ is the packet loss rate weight coefficient, P i (t) is the packet loss rate of the i-th path at time t;

[0040] According to the network health score, the receiving end will sort the paths to get the path with the highest priority and dynamically adjust the communication data.

[0041] As a preferred solution of the p2p remote communication method of the present invention, wherein: the receiving node reassembles the data blocks transmitted from each path to obtain complete communication data, including the following steps:

[0042] The sender numbers each data block, and the receiver allocates an independent cache area for each path;

[0043] The receiving end numbers the paths to group multiple data blocks from different paths, and sorts the data blocks of each path according to the data block sequence number;

[0044] The receiving end reads the data blocks one by one from the buffer according to the sequence number of the data blocks, and merges the read data blocks into complete communication data.

[0045] In a second aspect, the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the p2p remote communication method as described in the first aspect of the present invention is implemented.

[0046] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the p2p remote communication method as described in the first aspect of the present invention.

[0047] The beneficial effects of the present invention are as follows: by collecting and preprocessing network environment and communication data, high-quality communication data input is ensured, the long short-term memory network model is used to predict the network quality of each path, and a forward-looking evaluation of the future network state is realized, path selection and resource allocation are optimized, and transmission efficiency and user experience are improved. The SNMP protocol and the ICMP ping command are combined to obtain real-time network status information, calculate the fitness score of each path, dynamically adjust the path selection and resource allocation strategy, consider the current network conditions, and foresee future change trends, maximize network resource utilization, and improve the speed and stability of data transmission. The receiving node reorganizes the data blocks transmitted from each path, and adopts an independent cache area and a data block sequence number sorting and reorganization method to ensure the integrity and reliability of the communication data, and ensure an efficient and stable communication experience in a complex and changeable network environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0049] Figure 1 This is a flow chart of the p2p remote communication method in Example 1.

[0050] Figure 2 This is a flow chart of distributing data blocks to various paths in proportion for transmission in Example 1. DETAILED DESCRIPTION

[0051] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0052] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0053] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0054] Example 1, reference Figure 1and Figure 2 , which is the first embodiment of the present invention, and which provides a method for p2p remote communication, comprising the following steps:

[0055] S1. Collect network environment data and communication data, and perform preprocessing.

[0056] The collected network data includes bandwidth, latency and packet loss rate, and the collected communication data includes files, video streams and messages. All collected data are denoised and standardized.

[0057] S1.1 goes a step further and collects the bandwidth utilization of each path through the SNMP protocol, collects the transmission delay of each path through the ICMP Ping command, collects the packet loss rate of each path by sending data packets regularly and monitoring the packet loss, collects the transmitted file data, records the file size, type and transmission requirements, collects the file data to be transmitted, records the file size, type and transmission requirements, and collects the message data transmitted at the application layer, which mainly includes text and image information.

[0058] S1.2 Going further, the collected network environment data is denoised using the median filtering method. Outliers are selectively removed for bandwidth and delay data to ensure that communication data can accurately reflect long-term network trends. When denoising packet loss rate data, the sliding window averaging method is used to reduce the impact of instantaneous network status changes on data prediction. All collected network environment data and communication data are standardized using Z-score standardization. The standardization step ensures that different types of data (such as bandwidth, delay, and packet loss rate data) can be processed at the same scale, thereby improving the accuracy of machine learning model predictions.

[0059] S2. Input the preprocessed network environment data into the machine learning model to predict the network quality of each path.

[0060] S2.1. Extract bandwidth, delay and packet loss rate feature data from the preprocessed network environment data through principal component analysis method, input them into the long short-term memory network model, and predict the network quality of each path. The expression is:

[0061]

[0062] Among them, Q i (t) is the network quality prediction value of the ith path at time t, x k (t) is the value of the kth network quality feature at time t, w kis the weight of the kth network quality feature value, z(t) is the sum of network quality features at time t, n is the number of paths, β is the adjustment parameter of the sum of network quality features, k is the network quality feature index, and f is the nonlinear activation function.

[0063] Furthermore, in the principal component analysis method, the correlation between different bandwidth data is measured by the covariance matrix, and the principal components of the bandwidth data are extracted by eigenvalue decomposition, and finally the most representative principal component is selected to represent the bandwidth characteristics;

[0064] For delay data, the PCA method is applied to identify the main change trend of delay data through the covariance matrix, and the principal components are extracted through eigenvalue decomposition to retain the features that can best explain the delay changes;

[0065] The processing of packet loss rate data is similar to that of bandwidth and latency. First, the packet loss rate is denoised and standardized so that it can be compared on the same scale. The PCA method is used to pass the covariance matrix and extract the most informative principal components, which can effectively represent the network packet loss behavior.

[0066] For each path at time t, it is necessary to calculate the weight w of each feature. k The network quality features are weighted and summed to form the final path

[0067] To enhance the expressive power of the long short-term memory network model, the feature sum z(t) is processed by a nonlinear activation function f. The nonlinear activation function introduces a nonlinear relationship so that the network can better fit the complex network quality prediction task 1+exp(-β·z(t)).

[0068] S3. Based on the network quality prediction results, the intelligent scheduling algorithm is used to analyze and obtain the allocation strategy for each path selection.

[0069] S3.1. Use SNMP protocol to extract the actual flow and maximum available bandwidth of each path from the network device to obtain the bandwidth utilization.

[0070] Furthermore, the SNMP protocol is used to extract the actual flow and maximum available bandwidth of each path from network devices (such as routers and switches) to obtain the bandwidth utilization of the path and provide a basis for subsequent path selection.

[0071] S3.2. Use the ICMP ping command to measure the round-trip time from the sender to the receiver on each path to obtain the transmission delay.

[0072] Furthermore, several data packets (eg, 10) are sent to the receiving node through the ICMP ping command, the round trip time of each data packet is recorded, and the average value is obtained as the transmission delay of the path.

[0073] S3.3, based on the network quality prediction value, bandwidth utilization and transmission delay, the intelligent scheduling algorithm is used to dynamically predict the data of each path, and the allocation strategy for each path selection is obtained, which is expressed as:

[0074]

[0075] Among them, F(p) is the fitness score of the allocation strategy p, B i is the bandwidth utilization of the ith path, D i The transmission delay of the i-th path, α i where μ is the importance coefficient of the ith path, μ is the average value of all paths on the network quality prediction value, γ is the sensitivity of the control to the change of network quality, δ is the reference point of the network quality, j1 is the power of the network quality prediction value, o2 is the power of the bandwidth utilization, l3 is the power of the transmission delay, λ is the weight coefficient of the control transmission delay on the fitness score of the allocation strategy, A is the adjustment coefficient of the control network quality change sensitivity, and p is the allocation strategy.

[0076] Furthermore, by using the Sigmoid function It is used to smoothly adjust the score, so that when the network quality of the path is close to the reference point δ, the score changes slowly, and when it is far away, the score changes faster.

[0077] Using the natural logarithm function The scoring mechanism can distinguish the differences between different paths. For paths whose network quality is significantly different from the average μ, the absolute value ensures that the difference in scores between different paths will be distinguished regardless of whether the network quality is above or below the average.

[0078] The bandwidth utilization and transmission delay of each path are combined together through weighted summation, reflecting the importance of the path.

[0079] S3.4. According to the fitness score of the allocation strategy, use the quick sort method to sort all paths from high to low to obtain the allocation strategy selected for each path.

[0080] Furthermore, a quick sort algorithm is used to sort all paths based on the fitness score of the path. The quick sort algorithm uses a benchmark element to partition all paths according to the score and obtain an allocation strategy for each path selection.

[0081] S4. Split the communication data into multiple data blocks according to the allocation strategy selected by each path.

[0082] S4.1. According to the allocation strategy of each path, the communication data to be transmitted is split according to the weight ratio of the path.

[0083] Furthermore, the communication data is in the form of files, video streams or messages. According to the allocation strategy and the weight of each path, the communication data is divided into multiple data blocks, and the size of each data block is adjusted according to the adaptation score of the allocation strategy and the bandwidth requirement.

[0084] S4.2, communication data is split into small data blocks on high-latency and low-bandwidth paths.

[0085] Furthermore, according to the latency and bandwidth utilization of the path, the allocation strategy is used to control the size of the data block. For paths with high latency and low bandwidth, the communication data is split into smaller data blocks using application layer fragmentation technology on high latency and low bandwidth paths to reduce latency and packet loss, while improving the transmission success rate of these paths to avoid packet loss of too large data blocks due to network fluctuations to reduce latency and packet loss, while improving the transmission success rate of these paths.

[0086] S4.3. On low-latency and high-bandwidth paths, communication data is split into large data blocks.

[0087] Furthermore, using application layer slicing technology, communication data is split into larger data blocks on low-latency and high-bandwidth paths, allowing large amounts of data to be transmitted over these efficient paths.

[0088] S5. Based on the split data blocks, the split data blocks are proportionally distributed to various paths for transmission.

[0089] S5.1. According to the bandwidth utilization of each path, the bandwidth allocation ratio of each path is obtained.

[0090] Furthermore, bandwidth utilization refers to the ratio of the bandwidth used by a path in the current transmission process to the maximum bandwidth. It is used to measure the available bandwidth of each path. The bandwidth allocation ratio of each path is calculated through bandwidth utilization. The expression is:

[0091]

[0092] Among them, P i is the bandwidth allocation ratio of the i-th path, E i It is the sum of bandwidth utilization of all paths.

[0093] The bandwidth allocation ratio reflects the bandwidth load ratio that each path should bear during the entire communication data transmission process. A path with a higher ratio will bear more data block transmission.

[0094] S5.2, using the split data blocks, distribute the data blocks to each path in proportion according to the bandwidth allocation ratio of each path.

[0095] Furthermore, according to the bandwidth allocation ratio, the data blocks are allocated to each path in proportion to ensure that the number of data blocks received by each path matches its bandwidth allocation ratio.

[0096] S5.3. The data blocks on each path are transmitted via a specified transmission protocol.

[0097] Furthermore, the data blocks are distributed to each path and transmitted by selecting the TCP transport protocol, and the data blocks on each path are transmitted using the specified protocol.

[0098] S6. Based on the transmission of communication data, the receiving end and the sending end nodes exchange network status information to dynamically adjust the communication data allocation.

[0099] S6.1. The sender and receiver exchange real-time network status information through the control channel.

[0100] Furthermore, the sender and the receiver periodically exchange network status information through a dedicated control channel (such as a UDP control channel). The exchanged information includes bandwidth, delay, packet loss rate, and traffic status data of each path.

[0101] S6.2. The sender encapsulates all network status information into data packets and sends them to the receiver.

[0102] Furthermore, all the collected network status information is encapsulated into a data packet, and the encapsulated data packet is sent to the receiving end through the control channel.

[0103] S6.3, after receiving the network status information from the sender, the receiver decodes the network performance of each path, extracts the bandwidth, delay and packet loss rate data, and calculates the network health status of each path. The expression is,

[0104]

[0105] Among them, H i (t) is the network health score of the ith path at time t, B max is the maximum bandwidth of all paths, D max is the maximum delay of all paths, κ is the path bandwidth weight coefficient, θ is the delay weight coefficient, φ is the packet loss rate weight coefficient, P i(t) is the packet loss rate of the i-th path at time t.

[0106] Furthermore, the network status of each path is decoded using an extended header, and the bandwidth, delay, and packet loss rate data are extracted according to the TCP protocol. The bandwidth weight coefficient κ, the delay weight coefficient θ, and the packet loss rate weight coefficient φ are used. These weight coefficients can flexibly adjust the importance of bandwidth, delay, and packet loss rate to ensure the quality and efficiency of communication.

[0107] S6.4. The receiving end will sort the paths according to the network health score to obtain the path with the highest priority and dynamically adjust the communication data.

[0108] Furthermore, according to the health score ranking results of the paths, the receiving end will adjust the communication data allocation strategy. Paths with higher network health scores will receive more communication data allocation to improve the stability and efficiency of transmission. Communication data is achieved by optimizing the bandwidth allocation ratio, giving priority to paths with larger bandwidth, lower latency, and lower packet loss rate for data transmission.

[0109] S7. The receiving node reassembles the data blocks transmitted from each path to obtain complete communication data.

[0110] S7.1. The sending end numbers each data block, and the receiving end allocates an independent cache area for each path.

[0111] Furthermore, the sender numbers all data blocks, and each data block is assigned a unique sequence number so that the receiver can correctly reassemble the data based on the sequence number. These sequence numbers are globally unique, ensuring that the same data blocks transmitted on different paths will not be confused. The receiver allocates an independent cache area for each path to store data blocks received from different paths.

[0112] S7.2. The receiving end numbers the paths to group multiple data blocks from different paths, and sorts the data blocks of each path according to the data block sequence numbers.

[0113] Furthermore, the receiving end numbers each path and groups the received data blocks based on the path number. The data blocks of each path are stored in the cache according to the path number for subsequent sorting and reorganization. For each path, the receiving end stores all the transmitted data blocks in a separate cache area. Each slot in the cache area corresponds to a data block, and the receiving end sorts the data blocks according to the sequence number. The sorting operation ensures that the received data blocks are in the correct order.

[0114] S7.3, the receiving end reads the data blocks one by one from the buffer area according to the sequence numbers of the data blocks, and merges the read data blocks into complete communication data.

[0115] Furthermore, the receiving end reads the stored blocks one by one according to the data block sequence number in the cache area of ​​each path, and extracts them in the correct order according to the sorting order. The reading operation adopts a sequential reading method, and extracts data blocks in order from the smallest to the largest sequence number. After reading all data blocks, the receiving end merges them into complete communication data. The merging operation is achieved by splicing the contents of each data block in sequence, and finally the communication data originally sent by the sender is obtained.

[0116] This embodiment also provides a computer device, which is suitable for the case of the p2p remote communication method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the p2p remote communication method proposed in the above embodiment.

[0117] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0118] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for realizing p2p remote communication proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0119] In summary, the present invention ensures high-quality communication data input by collecting and preprocessing network environment and communication data, predicts the network quality of each path by using the long short-term memory network model, realizes the forward-looking evaluation of future network status, optimizes path selection and resource allocation, improves transmission efficiency and user experience, combines the SNMP protocol and ICMP ping command to obtain real-time network status information, calculates the fitness score of each path, dynamically adjusts the path selection and resource allocation strategy, takes into account the current network conditions, and can also foresee future changes. Trends, maximize network resource utilization, and at the same time improve the speed and stability of data transmission, the receiving node reorganizes the data blocks transmitted according to each path, and adopts the method of independent cache area and data block sequence number sorting and reorganization to ensure the integrity and reliability of communication data, and ensure efficient and stable communication experience in a complex and changeable network environment.

[0120] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A p2p remote communication method, characterized in that: include, Collect network environment data and communication data, and perform pre-processing; Input the preprocessed network environment data into the machine learning model to predict the network quality of each path; Based on the network quality prediction results, the intelligent scheduling algorithm is used to analyze and obtain the allocation strategy for each path selection; According to the allocation strategy selected for each path, the communication data is split into multiple data blocks and proportionally allocated to each path for transmission; Based on the transmission of communication data, the receiving and sending nodes exchange network status information to dynamically adjust the communication data allocation; The receiving node reassembles the data blocks transmitted from each path to obtain complete communication data.

2. The p2p remote communication method according to claim 1, characterized in that: Collecting network environment data and communication data and preprocessing them includes the following steps: The collected network data includes bandwidth, delay and packet loss rate, and the collected communication data includes files, video streams and messages; All collected data were denoised and normalized.

3. The p2p remote communication method as claimed in claim 2, characterized in that: The preprocessed network environment data is input into the machine learning model to predict the network quality of each path, which includes the following steps: The bandwidth, delay and packet loss rate feature data are extracted from the preprocessed network environment data through the principal component analysis method and input into the long short-term memory network model to predict the network quality of each path. The expression is: Among them, Q i (t) is the network quality prediction value of the ith path at time t, x k (t) is the value of the kth network quality feature at time t, w k is the weight of the kth network quality feature value, z(t) is the sum of network quality features at time t, n is the number of paths, β is the adjustment parameter of the sum of network quality features, k is the network quality feature index, and f is the nonlinear activation function.

4. The p2p remote communication method as claimed in claim 3, characterized in that: Based on the network quality prediction results, the intelligent scheduling algorithm is used to analyze and obtain the allocation strategy for each path selection, which includes the following steps: Use SNMP protocol to extract the actual flow and maximum available bandwidth of each path from network devices to obtain bandwidth utilization; The ICMP ping command is used to measure the round-trip time from the sender to the receiver on each path to obtain the transmission delay. Based on the network quality prediction value, bandwidth utilization and transmission delay, the intelligent scheduling algorithm is used to dynamically predict the data of each path, and the allocation strategy for each path selection is obtained. The expression is: Among them, F(p) is the fitness score of the allocation strategy p, B i is the bandwidth utilization of the ith path, D i The transmission delay of the i-th path, α i where μ is the importance coefficient of the ith path, μ is the average value of all paths on the network quality prediction value, γ is the sensitivity of the control to the change of network quality, δ is the reference point of the network quality, j1 is the power of the network quality prediction value, o2 is the power of the bandwidth utilization, l3 is the power of the transmission delay, λ is the weight coefficient of the control transmission delay on the fitness score of the allocation strategy, A is the adjustment coefficient of the control network quality change sensitivity, and p is the allocation strategy. According to the fitness score of the allocation strategy, all paths are sorted from high to low using the quick sort method to obtain the allocation strategy selected for each path.

5. The p2p remote communication method as claimed in claim 4, characterized in that: According to the allocation strategy selected for each path, splitting the communication data into multiple data blocks includes the following steps: According to the allocation strategy of each path, the communication data to be transmitted is split according to the weight ratio of the path; Communication data is split into small data chunks over high-latency and low-bandwidth paths; On low-latency and high-bandwidth paths, communication data is broken into large chunks.

6. The p2p remote communication method as claimed in claim 5, characterized in that: Based on the split data blocks, allocating the split data blocks to each path for transmission in proportion includes the following steps: According to the bandwidth utilization of each path, the bandwidth allocation ratio of each path is obtained; The split data blocks are allocated to each path in proportion according to the bandwidth allocation ratio of each path; The data blocks on each path are transmitted using a specified transport protocol.

7. The p2p remote communication method according to claim 6, characterized in that: Based on the transmission of communication data, the receiving end and the sending end node exchange network status information to dynamically adjust the communication data distribution, including the following steps: The sender and receiver exchange real-time network status information through the control channel; The sender encapsulates all network status information into data packets and sends them to the receiver; After receiving the network status information from the sender, the receiver decodes the network performance of each path, extracts bandwidth, delay, and packet loss rate data, and calculates the network health status of each path. The expression is: Among them, H i (t) is the network health score of the ith path at time t, B max is the maximum bandwidth of all paths, D max is the maximum delay of all paths, κ is the path bandwidth weight coefficient, θ is the delay weight coefficient, φ is the packet loss rate weight coefficient, P i (t) is the packet loss rate of the i-th path at time t; According to the network health score, the receiving end will sort the paths to get the path with the highest priority and dynamically adjust the communication data.

8. The p2p remote communication method as claimed in claim 7, characterized in that: The receiving node reassembles the data blocks transmitted from each path to obtain complete communication data, including the following steps: The sender numbers each data block, and the receiver allocates an independent cache area for each path; The receiving end numbers the paths, groups multiple data blocks from different paths, and sorts the data blocks of each path according to the data block sequence number; The receiving end reads the data blocks one by one from the buffer according to the sequence number of the data blocks, and merges the read data blocks into complete communication data.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the p2p remote communication method according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the p2p remote communication method according to any one of claims 1 to 8 are implemented.

Images