Adaptive communication protocol switching method and device

Abstract

The invention discloses a self-adaptive communication protocol switching method and device. The method comprises the following steps: acquiring network environment data corresponding to a target network environment, terminal equipment data corresponding to terminal equipment in the target network environment and service demand data; fusing the three kinds of data to obtain fused feature data; inputting the fused feature data into a network environment prediction model, and processing the fused feature data by using the network environment prediction model to obtain performance prediction data of each candidate network in the target network environment on a future preset time window; according to the adaptation capability value of each communication protocol in the protocol library under the performance prediction data; and after a target communication protocol is selected according to the adaptation capability value and the service demand data, switching to the target communication protocol. According to the invention, the technical problem that a conventional communication system adopts a fixed communication protocol and cannot be dynamically adjusted according to a network environment and service requirements in the prior art, which has great limitation, is solved.

Description

Technical Field

[0001] This invention relates to the field of communication protocol switching technology, and more specifically, to an adaptive communication protocol switching method and apparatus. Background Technology

[0002] With the rapid development of fifth-generation and sixth-generation mobile communication systems (5G / 6G) and the widespread application of Internet of Things (IoT) devices, the convergence of various heterogeneous networks (such as WiFi, cellular networks, and satellite networks) has become a trend. Heterogeneous network convergence aims to provide broader and more reliable connectivity to meet the network performance requirements of different scenarios. However, network convergence brings new challenges, especially for terminal devices, which need to efficiently switch between different networks to maintain optimal connectivity and Quality of Service (QoS). This places higher demands on the network management capabilities and communication protocol flexibility of terminal devices.

[0003] Traditional communication systems often rely on fixed communication protocols and lack the ability to dynamically adjust according to network environment and business needs. This "one-size-fits-all" communication strategy is difficult to cope with complex and ever-changing network conditions, especially in the case of high-density network deployment, multi-frequency band coexistence, and network congestion.

[0004] In addition, common problems during protocol switching include service interruption, data loss, and retransmission delays. These problems often stem from improper timing of the switchover, poor protocol compatibility, and unstable network conditions, leading to a degraded user experience, especially in latency-sensitive applications.

[0005] Existing systems typically only take action after a network performance degradation is detected, lacking proactive prediction of changes in network status. This leads to reactive responses and delayed decision-making, failing to effectively avoid the impact of upcoming network fluctuations on communications.

[0006] Furthermore, the lack of a comprehensive evaluation and selection mechanism when faced with multiple communication protocols makes it difficult for terminal devices to make optimal handover decisions. Existing handover strategies often rely on a single network performance metric, neglecting the combined impact of service requirements, terminal capabilities, and network topology.

[0007] There is currently no effective solution to the above problems. Summary of the Invention

[0008] This invention provides an adaptive communication protocol switching method and apparatus to at least solve the technical problem in related technologies where traditional communication systems use fixed communication protocols and cannot dynamically adjust according to network environment and business needs, resulting in significant limitations.

[0009] According to one aspect of the present invention, an adaptive communication protocol switching method is provided, comprising: acquiring network environment data corresponding to a target network environment, terminal device data corresponding to a terminal device in the target network environment, and service requirement data; fusing the network environment data, the terminal device data, and the service requirement data to obtain fused feature data; inputting the fused feature data into a network environment prediction model to process the fused feature data using the network environment prediction model to obtain performance prediction data of each candidate network in the target network environment in a future preset time window, wherein the network environment prediction model is trained by machine learning using multiple sets of training data, each set of training data including: sample fused feature data and sample performance prediction data corresponding to the sample fused feature data; according to the adaptation capability value of each communication protocol in the protocol library under the performance prediction data; after selecting a target communication protocol according to the adaptation capability value and the service requirement data, switching to the target communication protocol.

[0010] Optionally, the network environment data, the terminal device data, and the service requirement data are fused to obtain fused feature data, including: denoising the network environment data, the terminal device data, and the service requirement data to obtain denoised network environment data, terminal device data, and service requirement data; extracting features from the denoised network environment data, terminal device data, and service requirement data to obtain network environment feature data, terminal device feature data, and service requirement feature data; and standardizing the network environment feature data, terminal device feature data, and service requirement feature data to obtain the fused feature data.

[0011] Optionally, evaluating the adaptability value of each communication protocol in the protocol library under the performance prediction data includes: comparing the protocol feature data of each communication protocol in the protocol library with the performance prediction data to obtain a comparison result; and quantitatively evaluating the adaptability value of each communication protocol under the performance prediction data based on the comparison result.

[0012] Optionally, based on the adaptability values ​​of each communication protocol in the protocol library under the performance prediction data, the process includes: analyzing the performance indicators of each communication protocol in the predicted network environment corresponding to the performance prediction data to obtain protocol evaluation results; and mapping the service requirement data to the protocol evaluation results to calculate the adaptability value of each communication protocol under the performance prediction data.

[0013] Optionally, selecting a target communication protocol based on the adaptation capability value and the business requirement data includes: determining a matching score between each communication protocol and the business requirement corresponding to the business requirement data based on the adaptation capability value; and selecting the target communication protocol from the protocol library based on the matching score.

[0014] Optionally, after selecting a target communication protocol based on the adaptation capability value and the service requirement data, switching to the target communication protocol includes: while maintaining the connection with the original network, initiating a connection request to the target network corresponding to the target communication protocol; after receiving feedback information of the connection request, parsing the feedback information to obtain a parsing result; when the parsing result indicates that a connection with the target network has been successfully established, obtaining a data handover address from the target network, and registering an anchor point through the data handover address, wherein the anchor point, after receiving the registration information, establishes a tunnel between the original network and the target network; and forwarding all data to the target network through the tunnel via the anchor point.

[0015] Optionally, the adaptive communication protocol switching method further includes: after switching to the target network, updating the receive window state and congestion window state to the anchor point through the new link corresponding to the target network.

[0016] According to another aspect of the present invention, an adaptive communication protocol switching device is also provided, comprising: an acquisition unit, configured to acquire network environment data corresponding to a target network environment, terminal device data corresponding to a terminal device in the target network environment, and service requirement data; a fusion unit, configured to fuse the network environment data, the terminal device data, and the service requirement data to obtain fused feature data; an input unit, configured to input the fused feature data into a network environment prediction model, so as to process the fused feature data using the network environment prediction model to obtain performance prediction data of each candidate network in the target network environment in a future preset time window, wherein the network environment prediction model is obtained by machine learning training using multiple sets of training data, each set of multiple sets of training data including: sample fused feature data and sample performance prediction data corresponding to the sample fused feature data; a prediction unit, configured to use the adaptability value of each communication protocol in the protocol library under the performance prediction data; and a switching unit, configured to switch to the target communication protocol after selecting a target communication protocol according to the adaptability value and the service requirement data.

[0017] Optionally, the fusion unit includes: a denoising module for denoising the network environment data, the terminal device data, and the service requirement data to obtain denoised network environment data, terminal device data, and service requirement data; a feature extraction module for extracting features from the denoised network environment data, terminal device data, and service requirement data to obtain network environment feature data, terminal device feature data, and service requirement feature data; and a standardization module for standardizing the network environment feature data, terminal device feature data, and service requirement feature data to obtain the fused feature data.

[0018] Optionally, the prediction unit includes: a comparison module, configured to compare the protocol feature data of each of the communication protocols in the protocol library with the performance prediction data respectively, and obtain a comparison result; and a quantization module, configured to quantify and evaluate the adaptability value of each of the communication protocols under the performance prediction data based on the comparison result.

[0019] Optionally, the prediction unit includes: an analysis module, used to analyze the performance indicators of each of the communication protocols in the predicted network environment corresponding to the performance prediction data, and obtain the protocol evaluation results; and a mapping module, used to map the service requirement data to the protocol evaluation results, so as to calculate the adaptation capability value of each of the communication protocols under the performance prediction data.

[0020] Optionally, the switching unit includes: a determining module, configured to determine a matching score between each of the communication protocols and the business requirements corresponding to the business requirement data based on the adaptability value; and a selecting module, configured to select the target communication protocol from the protocol library based on the matching score.

[0021] Optionally, the switching unit includes: a connection request module, configured to initiate a connection request to a target network corresponding to the target communication protocol while maintaining a connection with the original network; a parsing module, configured to parse the feedback information received from the connection request to obtain a parsing result; an acquisition module, configured to acquire a data handover address from the target network when the parsing result indicates that a connection with the target network has been successfully established, and register an anchor point through the data handover address, wherein the anchor point establishes a tunnel between the original network and the target network after receiving the registration information; and a forwarding module, configured to forward all data to the target network through the tunnel via the anchor point.

[0022] Optionally, the adaptive communication protocol switching device further includes an update unit, used to update the receive window state and congestion window state to the anchor point through the new link corresponding to the target network after switching to the target network.

[0023] According to one aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein the program executes the optical adaptive communication protocol switching method described in any one of the above embodiments.

[0024] According to one aspect of the present invention, a processor is provided for running a program, wherein the program executes the adaptive communication protocol switching method described in any one of the above embodiments.

[0025] According to one aspect of the present invention, a computer program product is provided, including computer instructions that, when executed by a processor, perform the adaptive communication protocol switching method described in any one of the above embodiments.

[0026] In this embodiment of the invention, network environment data corresponding to the target network environment, terminal device data corresponding to the terminal devices in the target network environment, and service requirement data are acquired; the network environment data, terminal device data, and service requirement data are fused to obtain fused feature data; the fused feature data is input into a network environment prediction model to process the fused feature data using the network environment prediction model to obtain performance prediction data for each candidate network in the target network environment within a future preset time window. The network environment prediction model is trained using multiple sets of training data through machine learning. Each set of training data includes: sample fused feature data and sample performance prediction data corresponding to the sample fused feature data; the adaptability values ​​of each communication protocol in the protocol library under the performance prediction data are used; after selecting the target communication protocol based on the adaptability values ​​and service requirement data, the target communication protocol is switched to. The above technical solution achieves the following: using a multi-dimensional state awareness module to collect and preprocess data related to network environment, terminal status, and business needs in real time; utilizing a hybrid model of LSTM and GNN to make forward-looking predictions of the network environment; using a protocol adaptation evaluation module to analyze the performance of different protocols under predicted network conditions; and adopting a three-level collaborative "terminal-edge-cloud" architecture to achieve adaptive communication protocol switching. This improves communication quality, ensures business continuity and efficiency, and forms a closed loop from real-time awareness and rapid decision-making to long-term optimization. It supports smooth switching between different network environments, reduces service interruptions and data loss, and solves the technical problem that traditional communication systems use fixed communication protocols and cannot dynamically adjust according to network environment and business needs, resulting in significant limitations. Attached Figure Description

[0027] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0028] Figure 1 This is a hardware structure block diagram of a mobile terminal according to an embodiment of the present invention, which describes an adaptive communication protocol switching method.

[0029] Figure 2 This is a flowchart of an adaptive communication protocol switching method according to an embodiment of the present invention;

[0030] Figure 3 This is a flowchart of an optional adaptive communication protocol switching method according to an embodiment of the present invention;

[0031] Figure 4 This is a schematic diagram of an adaptive communication protocol switching device according to an embodiment of the present invention.

[0032] The above figures include the following reference numerals:

[0033] 102. Processor; 104. Memory; 106. Transmission device; 108. Input / output device. Detailed Implementation

[0034] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0035] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0036] As described in the background section, traditional communication systems in related technologies use fixed communication protocols, which cannot be dynamically adjusted according to network environment and business needs, resulting in significant limitations. Embodiments of this invention provide an adaptive communication protocol switching method and apparatus, a communication protocol, a computer-readable storage medium, a processor, and a computer program product.

[0037] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0038] The methods and embodiments provided in this invention can be executed on a mobile terminal, a computer terminal, or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 This is a hardware structure block diagram of a mobile terminal for an adaptive communication protocol switching method according to an embodiment of the present invention. Figure 1 As shown, a mobile terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the mobile terminal described above. For example, the mobile terminal may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0039] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the adaptive communication protocol switching method in this embodiment of the invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory and non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or send data via a network. Specific examples of the aforementioned networks may include wireless networks provided by the mobile terminal's communication provider. In one example, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0040] Example 1

[0041] According to an embodiment of the present invention, a method embodiment of an adaptive communication protocol switching method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0042] Figure 2 This is a flowchart of an adaptive communication protocol switching method according to an embodiment of the present invention, such as... Figure 2 As shown, the method includes the following steps:

[0043] Step S202: Obtain network environment data corresponding to the target network environment, terminal device data corresponding to the terminal devices in the target network environment, and service requirement data.

[0044] In this embodiment, network environment awareness involves real-time collection of key indicators for each candidate network, including but not limited to: signal strength (RSSI / RSCP), signal-to-noise ratio (SNR), reference signal reception quality (RSRQ), latency, jitter, packet loss rate, available bandwidth, network type (5G NR / 4G LTE / WiFi6 / satellite network, etc.), and network load status; terminal status awareness involves obtaining the current resource usage of the terminal device, including battery level, processor load, memory usage, temperature status, movement speed / direction / trajectory, geographic location information, antenna status, and historical handover records; and service requirement awareness involves identifying the type of currently running service and quality of service (QoS) requirements, including service category (e.g., video call, online game, file download, industrial control, telemedicine, etc.), priority level, latency sensitivity (e.g., <50ms / 100ms / 200ms), bandwidth requirements (e.g., >10Mbps / >50Mbps), reliability requirements (e.g., packet loss rate <1% / 0.1%), security level, and other parameters.

[0045] Optionally, in the real-time communication mechanism, the terminal reports lightweight status summaries (such as location hashes, key network metrics, and service requirement tags) to the associated edge nodes at fixed intervals (e.g., 100ms) or event-driven intervals (e.g., RSSI mutation > 3dB). The edge nodes push switching decision commands or policy parameter updates to the terminal. Lightweight heartbeats (such as the observation mechanism of CoAP) are used to maintain the connection. The terminal discovers the nearest edge node via Bluetooth Low Energy or local DNS-SD. Data synchronization between the edge nodes and the cloud adopts an asynchronous, incremental synchronization strategy. The edge nodes periodically upload aggregated anonymized data (not raw data), local model parameters, and system logs to the cloud. The cloud distributes the globally optimized prediction model, protocol library updates, and security policies to the edge nodes. The core synchronization protocol can use gRPC, which is suitable for batch data transmission and remote calls due to its high performance and support for bidirectional streaming. Real-time decisions (ms-level) are completed at the edge using local prediction models, while non-real-time optimizations (such as model retraining and policy tuning) are completed in the cloud, and the results are then distributed. Security and Privacy in Large-Scale Deployment: All links from the terminal to the edge to the cloud are mandated to use TLS 1.3, providing forward confidentiality and ensuring secure transmission. Sensitive data (such as geographic location and business content) is encrypted on the terminal side using the edge node's public key, ensuring that data remains encrypted during transmission and edge storage. For privacy protection, data minimization and anonymization methods are employed. Personally Identifiable Information (PII) is removed from data uploaded from the terminal, and location information can be generalized to a region ID (such as a cell ID), or differential privacy techniques can be used to add controllable noise to the aggregated data. In federated learning methods, model training can utilize a federated learning framework. Each edge node trains a local copy of the LSTM-GNN model using terminal data, encrypting only the model gradients or parameter updates (not the original data) and uploading them to the cloud for aggregation to generate a global model. This fundamentally protects data privacy. In secure containers and TEE methods, data processing and model inference at the edge nodes run in secure containers (such as Intel SGX, ARM Trust Zone), ensuring the confidentiality and integrity of code and data are protected even if the edge server is compromised.

[0046] This method comprehensively and in real-time collects and analyzes multivariate data related to network handover decisions. Specifically, this step aims to acquire data in three aspects: network environment data of the target network environment, terminal device data of the terminal devices in the target network environment, and currently running service requirement data. Network environment data includes, but is not limited to, key indicators such as signal strength, signal-to-noise ratio, network latency, network jitter, packet loss rate, bandwidth, and network type, which reflect the real-time performance and quality of different networks; terminal device data covers device resource usage, battery level, processor load, memory usage, temperature status, movement speed, direction, trajectory, geographical location, antenna status, and historical handover records, which reveal the current status and environmental adaptability of the terminal devices; service requirement data is used to identify the type of running services and their Quality of Service (QoS) requirements, including parameters such as service priority, latency sensitivity, bandwidth requirements, reliability requirements, and security level.

[0047] By implementing this control method, the system's perception capabilities are significantly enhanced, enabling it to monitor the network environment, terminal device status, and service requirements in real time and with great detail. This multi-dimensional data collection and analysis allows the system to consider more comprehensive factors in network switching decisions, such as potential network congestion, terminal mobility patterns, and real-time changes in service demands. Through precise perception, the system can more accurately predict changes in network status and assess protocol compatibility performance in advance, thereby effectively avoiding service interruptions, data loss, and switching delays, improving communication quality, and ensuring service continuity and user experience stability.

[0048] Step S204: The network environment data, terminal device data, and business requirement data are fused to obtain fused feature data.

[0049] In this embodiment, the acquired time-series data is denoised using methods such as Kalman filtering and wavelet transform, key feature vectors are extracted through feature engineering, and the fusion representation of multi-source heterogeneous data is achieved through standardization / normalization.

[0050] This method relies on advanced data processing and machine learning technologies to extract the most valuable information for communication protocol adaptation and switching decisions from heterogeneous data. First, the system uses complex data preprocessing steps, such as Kalman filtering and wavelet transform, to denoise and extract features from the raw time-series data, ensuring data accuracy and usability. Then, through standardization or normalization, the system unifies different types of network environment indicators, terminal status parameters, and business requirement indicators to the same order of magnitude, facilitating subsequent fusion analysis. Finally, multi-source data fusion techniques, such as combinatorial coding in feature engineering and feature mapping in deep neural networks, are employed to transform this heterogeneous data into a comprehensive feature vector. This vector not only includes the real-time status of the network environment and the current capabilities of terminal devices but also reflects the urgency and priority of business requirements, providing a comprehensive and accurate data foundation for subsequent intelligent prediction and protocol adaptation evaluation.

[0051] By implementing this control method, the system's intelligence and accuracy in communication protocol switching decisions have been significantly enhanced. Through the construction of integrated feature data, the system can comprehensively analyze network dynamics, terminal real-time status, and specific service requirements, achieving a full consideration of key factors in switching decisions. This integrated analysis enables the system to predict network status changes and identify evolving service needs in advance, thereby making more accurate protocol selections, reducing unnecessary switching, avoiding service interruptions and data loss, and improving communication stability and efficiency.

[0052] Step S206: Input the fused feature data into the network environment prediction model to process the fused feature data using the network environment prediction model, and obtain the performance prediction data of each candidate network in the target network environment in the future preset time window. The network environment prediction model is trained by machine learning using multiple sets of training data. Each set of training data includes: sample fused feature data and sample performance prediction data corresponding to the sample fused feature data.

[0053] In this embodiment, a hybrid model combining Long Short-Term Memory (LSTM) and Graph Neural Network (GNN) is employed. The LSTM layer captures the temporal dependencies of network metrics, while the GNN layer models the dynamic relationship between network topology and user location. Based on current and historical network state data, the performance evolution trend of each candidate network within a future time window (e.g., 300ms to 2s) is predicted, including expected signal quality, available bandwidth, latency changes, network congestion status, and handover availability probability. A performance prediction matrix for each candidate network is generated at multiple future time steps, including confidence assessment and uncertainty quantification indicators, providing a forward-looking basis for subsequent decision-making.

[0054] Optionally, the overall model architecture is as follows: The input layer receives feature vectors from the preprocessing module, with dimensions (Batch_Size, Time_Steps, Feature_Dim). The LSTM branch (capturing temporal dynamics) consists of two stacked LSTM layers; parameters: time step size Time_Steps=10 (corresponding to 1 second of history, assuming one sample every 100ms), hidden units per LSTM layer Hidden_Units=64, dropout rate=0.2 to prevent overfitting; the output of the last time step is taken as the condensed representation H_lstm of the temporal features. The graph construction rules for the GNN branch (capturing network topology and spatial relationships) are as follows: The terminal, surrounding base stations / APs, and edge nodes are modeled as a heterogeneous graph; nodes: terminals (features: terminal status, service requirements), network nodes (features: network performance indicators, load); edges: connecting terminals to visible network nodes. Edge features include: link indicators (predicted RSSI, Latency) and physical distance. Because different neighboring nodes (networks) have different influence weights on the terminal, the GNN model uses a GraphAttention Network (GAT) with two layers. Each GAT layer uses four attention heads, which are then concatenated. The output is a representation of the spatial topological features, H_gnn, obtained by reading out the terminal nodes. In the fusion and output layers, fusion involves concatenating H_lstm and H_gnn [H_lstm; H_gnn]. The fully connected layer consists of two fully connected networks (128 and 64 neurons, ReLU activation function), with the final output layer using linear activation to output a prediction matrix with dimensions (Num_Candidate_Networks, Num_Predicted_Timesteps, Num_Metrics). For example, it can predict RSSI, latency, and bandwidth for the next five time steps (500ms). Uncertainty quantification can be achieved by introducing Monte Carlo Dropout or deep learning ensembles, performing multiple forward propagations during inference, and calculating the mean and variance of the predicted values ​​as confidence assessments. During training, large-scale simulation datasets (such as those generated by NS-3) and real-world data are used for pre-training and fine-tuning. The loss function is Huber Loss (more robust to outliers than MSE) plus a regularization term (encouraging smooth predictions). The optimizer is AdamW with an initial learning rate of 3e-4, and cosine annealing is used for learning rate scheduling.

[0055] The network environment prediction model employed in this method is trained using machine learning algorithms based on a large amount of historical data, particularly deep learning techniques with powerful time-series processing and correlation analysis capabilities, such as Long Short-Term Memory (LSTM) combined with Graph Neural Networks (GNNs). LSTM can capture long-term dependencies in time-series data, while GNNs can handle data in complex network structures and understand the relationships and influences between network nodes. During the training phase, each set of training data includes sample fusion feature data and corresponding sample performance prediction data. The sample fusion feature data reflects the state of the network, terminals, and services under similar past conditions, while the sample performance prediction data predicts network performance changes based on these conditions. By learning the patterns and regularities in these data sets, the model can make predictions about future network performance based on new fusion feature data during the testing phase, including expected signal quality, available bandwidth, latency variations, network congestion status, and the feasibility of network switching.

[0056] By implementing this control method, the system's ability to predict future network environment changes is significantly improved, which is crucial for adaptive communication protocol switching. Through effective processing of fused feature data, the system can predict the performance of each candidate network in the upcoming time window based on current and past network conditions, terminal status, and service requirements. This prediction is not merely an estimate of a single parameter, but a comprehensive, multi-dimensional performance forecast, including signal stability, bandwidth fluctuations, latency increases, and potential network congestion. This proactive prediction allows the system to adjust communication protocols before network environment changes occur, avoiding service interruptions or quality degradation due to network deterioration, and improving communication continuity and reliability.

[0057] Step S208: Based on the adaptation capability values ​​of each communication protocol in the protocol library under the performance prediction data.

[0058] In this embodiment, mainstream communication protocol stacks (such as 802.11ax / be for WiFi 6 / 7, URLLC / eMBB slicing for 5G NR, eMBB / URLLC for 4G LTE, DVB-RCS2 for satellite networks, etc.) are pre-integrated, and performance characteristic parameters of each protocol are extracted under different network environments.

[0059] This method compares and analyzes predicted network performance characteristics with the characteristics of different communication protocols to quantify the performance of each protocol under specific network conditions. First, the system extracts characteristic parameters of each communication protocol from a protocol library, including key indicators such as authentication latency, handover latency, maximum throughput, minimum latency, anti-interference capability, energy efficiency, and security level. Next, these protocol characteristics are combined with the future network performance prediction data obtained in step S206. That is, under predicted signal quality, bandwidth, latency, and network congestion conditions, the system evaluates whether each protocol can operate stably, meet business requirements, and assesses its relative advantages and disadvantages. The system typically employs a multi-dimensional weighted algorithm to score each protocol based on real-time business requirements (such as latency sensitivity, throughput requirements, and reliability), generating an adaptation capability value.

[0060] By implementing this control method, the system's decision-making intelligence is significantly enhanced, enabling it to intelligently select the most suitable communication protocol based on predicted changes in the network environment and specific business needs. Through quantitative evaluation, the system can not only determine which protocols may perform poorly or fail to meet business requirements under future network conditions, but also identify those protocols that provide optimal performance, making switchover arrangements in advance to avoid problems such as service interruptions, data loss, and inefficiency. This decision-making process fully considers the characteristics of the protocols and compatibility with the predicted network environment, effectively reducing resource waste and excessive energy consumption caused by unsuitable protocols.

[0061] Step S210: After selecting the target communication protocol based on the adaptation capability value and business requirement data, switch to the target communication protocol.

[0062] In this embodiment, a decision-making mechanism is relied upon to comprehensively consider performance prediction data, protocol adaptation capability values, and service QoS requirements to determine which protocol best meets the current service needs under predicted network conditions. A high-level solution to ensure service continuity includes dual-active connectivity and data pre-synchronization: Before the handover decision is triggered, the intelligent prediction engine notifies the protocol adaptation module in advance (e.g., predicting that the current network quality will deteriorate in 200ms) to prepare for the handover. While maintaining a connection with the original network, the terminal establishes a connection with the new target network in parallel, performing security authentication and context establishment. This utilizes 5G's "dual registration" or Wi-Fi's "pre-association" mechanism. Key data (such as TCP sequence numbers and application layer session states) is synchronized between the two links through edge nodes. Smooth migration based on the application layer: For video / games (UDP streams), protocols supporting seamless path switching, such as SRT or WebRTC, are used. The sending end simultaneously sends redundant encoded data streams to two IP addresses (corresponding to two network interfaces), and the receiving end dynamically selects the primary stream based on network quality, achieving sub-second handover seamlessly for the user. For file transfer / TCP services, use MPTCP. During switching, MPTCP substreams can gracefully migrate data traffic from the old path to the new path, without interrupting the TCP session and without the application layer being aware of it.

[0063] In this method, the system performs a weighted calculation of the adaptability value based on preset service priorities and QoS parameters, such as latency, throughput, and reliability requirements, to obtain a final protocol selection score. The protocol with the highest score is selected as the target communication protocol. Subsequently, the system performs a protocol switching operation, which may involve reconfiguring the protocol stack, re-establishing network connections, and seamlessly transferring data streams. The system also needs to ensure data continuity and security during the switching process, meaning that data transmission will not be interrupted or lost under the new protocol, while meeting the security requirements of the service. In addition, the switching mechanism is designed with redundancy and fallback strategies to quickly restore the original protocol in case of switching failure or poor performance of the new protocol, ensuring communication stability and service continuity.

[0064] By implementing this control method, the adaptive capabilities and user experience of the communication system are significantly improved, achieving intelligent matching between communication protocols and business requirements. By selecting the most suitable communication protocol under predicted network conditions, the system can proactively optimize data transmission strategies, ensuring business continuity and efficient data transmission. For example, in low-latency services (such as online games and real-time video conferencing), the system selects protocols with low latency characteristics; in high-bandwidth demand scenarios (such as 4K video streaming), the system tends to select high-throughput protocols. This intelligent switching not only reduces the possibility of service interruptions but also lowers the risk of packet loss, improving communication continuity and stability.

[0065] As described above, in this embodiment, network environment data corresponding to the target network environment, terminal device data corresponding to the terminal devices in the target network environment, and service requirement data are acquired; the network environment data, terminal device data, and service requirement data are fused to obtain fused feature data; the fused feature data is input into the network environment prediction model to process the fused feature data using the network environment prediction model, thereby obtaining performance prediction data for each candidate network in the target network environment within a future preset time window. The network environment prediction model is trained using multiple sets of training data through machine learning. Each set of training data includes: sample fused feature data and sample performance prediction data corresponding to the sample fused feature data; based on the performance of each communication protocol in the protocol library... The system predicts the adaptability value based on the data. After selecting the target communication protocol based on the adaptability value and business requirements, it switches to the target communication protocol. This achieves the goal of using a multi-dimensional state awareness module to collect and preprocess data related to the network environment, terminal status, and business requirements in real time. It also uses a hybrid model of LSTM and GNN to make forward-looking predictions about the network environment. The protocol adaptation evaluation module analyzes the performance of different protocols under the predicted network state. By adopting a three-level collaborative "terminal-edge-cloud" architecture, it achieves the goal of adaptive communication protocol switching. This improves communication quality, ensures business continuity and efficiency, and forms a closed loop from real-time awareness and rapid decision-making to long-term optimization. It supports smooth switching between different network environments and reduces service interruptions and data loss.

[0066] Therefore, the technical solution provided by the above embodiments of the present invention solves the technical problem that traditional communication systems in the related art use fixed communication protocols and cannot be dynamically adjusted according to network environment and business needs, which has great limitations.

[0067] According to the above embodiments of the present invention, fusing network environment data, terminal device data, and service requirement data to obtain fused feature data includes: denoising the network environment data, terminal device data, and service requirement data to obtain denoised network environment data, terminal device data, and service requirement data; extracting features from the denoised network environment data, terminal device data, and service requirement data to obtain network environment feature data, terminal device feature data, and service requirement feature data; and standardizing the network environment feature data, terminal device feature data, and service requirement feature data to obtain fused feature data.

[0068] In this embodiment, the denoising process is specifically implemented as follows: Kalman filtering is applied to the state estimation of linear systems with Gaussian white noise, such as RSSI (signal strength index) and delay. The state vector is... By setting the process noise covariance matrix Q and the observation noise covariance matrix R, along with a relatively large initial state covariance P, Gaussian white noise is effectively filtered out. The process noise covariance matrix Q and the observation noise covariance matrix R need to be estimated online using historical data calibration or an adaptive algorithm, while the initial state covariance P can be set to a relatively large value. Meanwhile, wavelet transform is suitable for denoising non-stationary signals such as jitter and packet loss rates. The 'db4' (Daubechies4) wavelet basis is selected, and a three-level decomposition is performed. For the high-frequency detail coefficients obtained from the decomposition, a soft thresholding method (such as a general threshold) is used. Summarizing and reconstructing the signal effectively filters out burst noise and preserves the true jitter trend. During the feature engineering and extraction phase, a comprehensive and refined set of key feature vectors was constructed, covering instantaneous features, statistical features, cross-features, service context features, and spatiotemporal features. Instantaneous features reflect the current value of various indicators (RSSI, Latency, etc.); statistical features include the mean, variance, maximum, minimum, percentiles (e.g., 95th delay), and trend (first-order difference mean) over a past time window (e.g., 1 second); cross-features include the product of signal strength and movement speed (reflecting fading rate) and the ratio of delay to jitter (reflecting network stability); service context features include the one-hot encoding of the service type, the difference between the demand threshold (e.g., maximum tolerable delay) and the current value; spatiotemporal features include the location grid ID, movement direction (angle), and relative distance to the base station / AP (estimated through signal modeling). During the standardization / normalization phase, missing values ​​were first managed. Transient missing values ​​were handled using linear interpolation or forward imputation strategies, while long-term missing values ​​were considered network unusable. Next, outliers were detected and processed using the IQR method, and values ​​exceeding the limit were removed. Values ​​within the range are considered outliers and replaced with boundary values. Finally, features of different dimensions are unified to the same range through Min-Max normalization (e.g., mapping RSSI from [-120,-60] to [0,1]) and Z-Score standardization (e.g., handling latency: (x-mean) / standard deviation). For periodic features (e.g., orientation angle), sine and cosine transforms are used.

[0069] The feature data generation step in this method is crucial for building an adaptive communication protocol switching service. It achieves data preprocessing and refinement through three consecutive sub-steps: denoising, feature extraction, and standardization. First, denoising utilizes techniques such as Kalman filtering and wavelet transform to filter out Gaussian white noise and non-stationary noise from network environment data, terminal device data, and service requirement data, ensuring data purity and reliability. Next, the feature extraction stage performs in-depth analysis on the denoised data to construct instantaneous features, statistical features, cross features, service context features, and spatiotemporal features. These features comprehensively describe the dynamic state and interrelationships of the network, terminals, and services, providing detailed information support for subsequent intelligent prediction and protocol selection. Finally, standardization transforms the extracted feature data to a unified dimension and scale through Min-Max normalization or Z-Score normalization, as well as sine and cosine transforms of periodic features. This eliminates the impact of dimensional differences on data processing, allowing data from different sources to be compared and fused within the same framework.

[0070] By implementing this control method, the system significantly improves its ability to comprehensively perceive and accurately understand the network environment, terminal status, and service requirements. Noise reduction effectively filters out noise interference from the raw data, improving data accuracy. The multi-dimensional feature vectors constructed during the feature extraction stage not only capture the instantaneous state of network performance but also reveal its trends and changes, while comprehensively considering the urgency and specificity of service requirements, greatly enriching the decision-making basis. Standardization ensures that data from different sources and with different dimensions can be fused and analyzed under a unified standard, avoiding analytical biases caused by differences in dimensions. The final fused feature data not only includes real-time network and terminal status information but also incorporates predictions of future network performance changes and considerations of service requirements. This provides comprehensive, accurate, and standardized data input for the intelligent prediction engine and protocol adaptation evaluation module, enabling the system to make forward-looking network performance predictions and optimal communication protocol selections based on this data. This significantly improves communication quality, ensures service continuity, optimizes network resource utilization, reduces energy consumption, and enhances the system's adaptability and efficiency.

[0071] According to the above embodiments of the present invention, evaluating the adaptability value of each communication protocol in the protocol library under performance prediction data includes: comparing the protocol feature data of each communication protocol in the protocol library with the performance prediction data to obtain comparison results; and quantitatively evaluating the adaptability value of each communication protocol under performance prediction data based on the comparison results.

[0072] In this embodiment, for each candidate network, the adaptability of each protocol in the predicted future state is evaluated, including key technical indicators such as authentication latency, handover latency, maximum throughput, minimum latency, anti-interference capability, energy efficiency, and security level.

[0073] Optionally, the quantification standards for metrics such as "authentication latency," "handover latency," and "minimum latency" are not fixed values, but rather performance boundaries dynamically determined based on communication protocol specifications, measured statistical data, and service requirement thresholds. Authentication latency quantification standard: Definition: The time consumed by a terminal from initiating an access or handover request to completing the entire process of two-way authentication, key negotiation, and security context establishment with the target network. This latency includes all signaling round-trip time and processing latency of protocol interactions. Quantification benchmark and "reasonableness" judgment: The system has a built-in authentication latency benchmark library based on protocol type and scenario. For example: For WPA3-Enterprise enterprise-grade Wi-Fi access based on 802.1X / EAP, the typical authentication latency benchmark range is set to 50 milliseconds to 300 milliseconds. The specific value depends on the EAP method (e.g., EAP-TLS is usually faster than EAP-PEAP) and the RADIUS server response speed. For initial 5G NR registration (using 5G-AKA or EAP-AKA'), the typical authentication latency benchmark range is set to 100 milliseconds to 500 milliseconds, involving the interaction of multiple network elements (AMF, AUSF, UDM) in the terminal, radio access network, and core network. The "reasonable" judgment logic: The system compares the real-time measured or predicted authentication latency with the tolerance threshold of the current service. For services with extremely high real-time requirements (such as remote control, with a handover tolerance latency <10ms), the system will prioritize protocol and network combinations that support pre-authentication or context advance delivery, aiming to compress the authentication overhead during handover to an extremely low level (e.g., <1ms). In this case, the "reasonable" standard is whether this stringent threshold is met. For non-real-time services, authentication latency is considered acceptable as long as it does not exceed the overall expected time for service establishment (e.g., several seconds). Quantitative standard for handover latency: Definition: Specifically refers to user plane interruption latency, i.e., the time interval between the terminal successfully receiving the last user data packet on the source link (original network / protocol) and successfully receiving the first user data packet on the target link (new network / protocol). This metric directly determines whether a perceptible service interruption occurs. Quantification and the "Seamless" Target: The system defines target values ​​for "seamless" handover based on different handover types: Intra-system handover (e.g., handover between 5G base stations): Through a robust context forwarding mechanism, the target interruption latency is 0 to 10 milliseconds. Cross-system handover (e.g., between 5G and Wi-Fi): Utilizing higher-level session continuity protocols (e.g., connection migration based on MPTCP or QUIC), the target is to control user-perceptible interruption to below 50 milliseconds. For services such as voice and video calls, interruptions below 20 milliseconds are typically imperceptible to the human ear / eye; this is defined as the quantification target for "smooth transition." The system estimates the latency of each potential handover by measuring historical handover latency data and combining it with a predictive model, serving as the evaluation basis.Minimum latency quantification standard: Definition: Under ideal channel conditions of no contention, no packet loss, and no retransmission, and with a specific protocol stack and network path, the theoretical lower limit of end-to-end one-way propagation latency achievable when transmitting a very small data packet (such as a 64-byte ICMPPing or protocol-specific signaling). This metric reflects the inherent components of protocol stack processing overhead and network physical propagation latency. Quantification method: This value is obtained through laboratory benchmark testing, protocol stack analysis and modeling, and extensive probe measurements during periods of low network load in actual operations. For example, the minimum latency of the 5G URLLC protocol stack under ideal conditions can be modeled as 1 millisecond. The minimum latency of the Wi-Fi 6 protocol stack in a very short-range, contention-free scenario can be modeled as 2 to 5 milliseconds. Low-Earth orbit satellite networks, limited by satellite-to-ground propagation distance, have a minimum latency physical lower limit of approximately 20 to 40 milliseconds. Purpose: During the evaluation phase, "minimum latency" serves as a hard screening criterion. If the "minimum latency" value of a protocol-network combination exceeds the maximum tolerable latency required by the current service, the combination will be directly excluded without further comprehensive scoring, thereby improving decision-making efficiency. The "multi-dimensional weighted algorithm" is used to calculate the comprehensive suitability score of each candidate protocol-network combination. Its weight allocation is a dynamic process based on rules and policies. The initial basis for weight allocation is strictly based on the service quality parameters output by the service requirement perception module. The system pre-defines a service type-weight template mapping table. For example: Ultra-reliable low-latency communication services (such as industrial control): reliability weight allocation is 45%, latency weight allocation is 45%, throughput weight allocation is 5%, and energy consumption weight allocation is 5%. Enhanced mobile broadband-interactive services (such as high-definition video calls): latency weight allocation is 35%, jitter weight allocation is 25%, reliability weight allocation is 20%, throughput weight allocation is 15%, and energy consumption weight allocation is 5%. Enhanced mobile broadband-background download services: throughput weight allocation is 55%, energy consumption weight allocation is 20%, cost weight allocation (if considered) is 15%, and latency weight allocation is 10%. Dynamic weight adjustment mechanism: Initial weights will be dynamically adjusted based on real-time context information. The adjustment mechanism is based on a policy engine that responds to the following inputs: Terminal status input: When the battery level is below a preset threshold (e.g., 15%), an energy-saving strategy is triggered, significantly increasing the weight of the "energy efficiency" indicator (e.g., multiplying its weight by a coefficient greater than 1, such as 2.0), and correspondingly reducing the weight of peak performance indicators. When the terminal is detected to be in a high-speed moving state (e.g., speed > 80 km / h), a mobility optimization strategy is triggered, increasing the weight of indicators related to "handover robustness" and "coverage continuity". Network prediction input: When the intelligent prediction engine outputs an indication that a network will experience severe congestion in the future, the weight of the "anti-interference capability" and "stability" indicators is increased in all protocol evaluations for that network.User and Operation / Maintenance Policy Input: The system supports configurable policy levels (such as "Ultimate Experience Mode," "Balanced Mode," and "Energy Saving Mode"). Different modes correspond to different base weight bias coefficients. The operations and maintenance team can issue global weight adjustment parameters through the policy server to optimize the overall network load or ensure the experience of specific user groups.

[0074] In this method, the system takes performance prediction data as input and compares it with the characteristic parameters of various communication protocols in the protocol library. The protocol characteristic data here covers key performance indicators of the protocols, such as authentication latency, handover latency, maximum throughput, minimum latency, anti-interference capability, energy efficiency, and security level. The system establishes a multi-dimensional comparison model to match these protocol characteristics with predicted future performance data such as network signal strength, latency, bandwidth availability, jitter, network load, and congestion status, calculating the performance score of each communication protocol under the predicted conditions, i.e., the adaptability value. This process typically involves comprehensive consideration of the protocol's adaptability to changes in the network environment, the satisfaction of service QoS requirements, and the smoothness of protocol handover and energy consumption impact.

[0075] By implementing this control method, the system's decision-making intelligence and dynamic adaptability are effectively enhanced. It can proactively identify and select the optimal communication protocol based on predicted network environment changes and current business needs. The result of this evaluation process, namely the adaptability value, directly guides the system's protocol switching strategy in different scenarios, ensuring business continuity and improving data transmission efficiency and quality. For example, when network latency is expected to increase, the system can prioritize protocols with strong low-latency characteristics; in environments where bandwidth is predicted to decrease, the system tends to use protocols with low bandwidth consumption and high data compression efficiency.

[0076] According to the above embodiments of the present invention, based on the adaptability values ​​of each communication protocol in the protocol library under the performance prediction data, the method includes: analyzing the performance indicators of each communication protocol in the predicted network environment corresponding to the performance prediction data to obtain the protocol evaluation results; mapping the service requirement data to the protocol evaluation results to calculate the adaptability values ​​of each communication protocol under the performance prediction data.

[0077] In this embodiment, based on the current service QoS requirements, the capability-requirement matching score of each protocol is calculated, and the protocol adaptation value is quantified by a multi-dimensional weighted algorithm (considering a typical weight allocation of 40% for latency, 30% for throughput, 20% for reliability, and 10% for energy consumption).

[0078] This method utilizes performance prediction data to deeply analyze the performance of various communication protocols under anticipated network conditions, including key indicators such as authentication latency, handover latency, throughput, delay, anti-interference capability, energy efficiency, and security level. This analysis aims to understand how protocols respond to changes in the network environment, such as fluctuations in signal strength, increased latency, or network congestion, and their stability and efficiency under these conditions. Subsequently, the system maps specific service requirement data (such as latency tolerance, bandwidth requirements, and service priorities) to the protocol evaluation results, calculating the adaptability value of each protocol through a multi-dimensional weighted algorithm. This algorithm fully considers the QoS requirements of services, assigning higher weights to latency sensitivity and lower weights to energy efficiency to balance the various performance aspects of the protocols.

[0079] By implementing this control method, the accuracy and foresight of communication protocol selection are significantly improved, enabling better matching of business needs and dynamic network environments. By closely integrating protocol performance predictions with business requirement data, the system can not only identify protocols that perform well under predicted network conditions but also preemptively exclude protocols that may not meet business QoS requirements, thereby avoiding unnecessary service interruptions and reduced data transmission efficiency. This predictive data-based adaptability assessment allows the system to make reasonable protocol switching decisions before network environment changes, ensuring business continuity and enhancing user experience.

[0080] According to the above embodiments of the present invention, selecting a target communication protocol based on the adaptability value and business requirement data includes: determining the matching degree score between each communication protocol and the business requirement corresponding to the business requirement data based on the adaptability value; and selecting a target communication protocol from the protocol library based on the matching degree score.

[0081] In this embodiment, a lightweight, low-latency protocol is used. MQTT over QUIC is an ideal choice. QUIC, based on UDP, reduces TCP handshake and head-of-line blocking latency, making it well-suited for mobile and dynamic network environments. MQTT's publish / subscribe model facilitates terminal reporting of sensing data and receiving handover decisions from the edge.

[0082] The core of this method lies in the establishment of a matching score and protocol selection. First, the system generates a score matrix by calculating the matching degree between the adaptability value and business requirement data (such as maximum tolerable latency, bandwidth requirements, and security level). Each score represents the ability of each candidate protocol to meet specific business requirements. The calculation of the matching score typically involves multi-dimensional weighted evaluation, that is, assigning different weights to different performance indicators according to the QoS priority of the business to reflect a comprehensive consideration of business requirements. Subsequently, based on this matching score matrix, the system automatically selects the target communication protocol with the highest score from the protocol library using sorting, threshold filtering, or optimization algorithms (such as genetic algorithms, particle swarm optimization, etc.) as the communication protocol to be used under the expected network conditions.

[0083] By implementing this control method, the intelligence and personalization of network connectivity are significantly enhanced. Through precise matching of predicted network conditions and service demands, the system can proactively and rapidly switch to the optimal protocol, avoiding the service quality degradation and user experience impairment that can occur with traditional fixed-protocol solutions. This protocol selection based on matching scores not only ensures the continuity and efficiency of data transmission but also copes with complex and changing network conditions, such as network congestion, signal fluctuations, or device movement, thereby maintaining seamless service operation.

[0084] According to the above embodiments of the present invention, after selecting a target communication protocol based on the adaptability value and service requirement data, switching to the target communication protocol includes: while maintaining the connection with the original network, initiating a connection request to the target network corresponding to the target communication protocol; after receiving feedback information of the connection request, parsing the feedback information to obtain a parsing result; when the parsing result indicates that a connection with the target network has been successfully established, obtaining a data handover address in the target network, and registering an anchor point through the data handover address, wherein, after receiving the registration information, the anchor point establishes a tunnel between the original network and the target network; and forwarding all data to the target network through the tunnel via the anchor point.

[0085] In this embodiment, the high-level solution to ensure service continuity includes: dual-active connection and data pre-synchronization, and smooth migration based on the application layer. Specifically, in the dual-active connection and data pre-synchronization, before the handover decision is triggered, the intelligent prediction engine notifies the protocol adaptation module in advance (e.g., predicting that the current network quality will deteriorate in 200ms) to prepare for the handover. While maintaining a connection with the original network, the terminal establishes a connection with the new target network in parallel, performing security authentication and context establishment. This utilizes the "dual registration" mechanism of 5G or the "pre-association" mechanism of Wi-Fi. Key data (such as TCP sequence numbers and application layer session states) is synchronized between the two links through edge nodes. For smooth migration based on the application layer, for video / games (UDP streams): protocols supporting seamless path switching, such as SRT or WebRTC, are used. The sending end simultaneously sends redundant encoded data streams to two IP addresses (corresponding to two network interfaces). The receiving end dynamically selects the primary stream based on network quality, achieving sub-second switching that is imperceptible to the user. For file transfer / TCP services: MPTCP is used. During the switchover, MPTCP sub-streams can gracefully migrate data traffic from the old path to the new path, with the TCP session remaining uninterrupted and the application layer remaining unaware.

[0086] The protocol switching step in this method employs a smooth transition mechanism, aiming to achieve seamless connectivity on the target communication protocol while ensuring the continuity and integrity of data transmission. The switching process begins by maintaining a stable connection with the original network. The system then initiates a connection request to the target network, based on the characteristics of the previously selected target communication protocol, ensuring efficient and compatible connection establishment. Upon receiving connection feedback from the target network, the system performs detailed analysis of the feedback information to verify the successful connection establishment. If the feedback indicates successful connection establishment, the system obtains the data handover address from the target network and registers with an anchor point on the target network using this address. Upon receiving the registration information, the anchor point establishes a data transmission tunnel between the original and target networks. This tunnel acts as a bridge for data flow, enabling the system to smoothly forward all data from the original network to the target network through the anchor point.

[0087] By implementing this control method, the user experience and business continuity during network handover are significantly improved. During the selection and switching of the target communication protocol, the system achieves a seamless data transition, avoiding connection delays and packet loss common in traditional handover methods, ensuring the stability and reliability of data transmission. This strategy of pre-establishing connections and data tunnels not only reduces handover time but also improves the handover success rate, enabling terminal devices to seamlessly switch to the optimal communication protocol in constantly changing network environments, maintaining high-quality business services.

[0088] According to the above embodiments of the present invention, the adaptive communication protocol switching method further includes: after switching to the target network, updating the receiving window state and congestion window state to the anchor point through the new link corresponding to the target network.

[0089] In this embodiment, address and routing management: The network layer scheme adopts host-based mobility management, such as MP-DCCP or Proxy Mobile IPv6. The edge node acts as a local mobile anchor point, assigning a fixed home address (HoA) to the terminal. During handover, the terminal obtains a care-of address (CoA) in the new network and registers it with the anchor point. The anchor point forwards all data packets to the terminal's current CoA through a tunnel, ensuring that the IP address remains unchanged and the session continues. In a more aggressive scheme, the application layer uses the QUIC protocol. A QUIC connection is identified by a connection ID, decoupled from the IP address and port. When the terminal's IP address changes due to handover, it only needs to use the same connection ID to restore the connection on the new address without re-handshaking, resulting in extremely low handover latency. Packet processing mechanism: Sequencing and retransmission: At the moment of handover, the old link may still be transmitting packets. The edge anchor is responsible for buffering newly arriving packets and waiting for the last few sequence numbers of the old link to ensure that they are delivered to the terminal in order. For lost packets, the anchor or the terminal initiates a fast retransmission on the new link. Flow control: After handover, the terminal immediately updates its receive window and congestion window status to the sender (or anchor) through the new link to quickly restore the optimal transmission rate. Specific role of the protocol adaptation layer: This module has completed the "capability assessment" of each protocol before handover. When it is decided to switch from "5G URLLC slicing" to "Wi-Fi 6", it not only triggers the handover but also dynamically configures the terminal's protocol stack parameters. For example, it adjusts the TCP initial window and retransmission timeout (RTO) values ​​from smaller values ​​optimized for cellular networks to larger values ​​optimized for Wi-Fi environments to immediately adapt to the latency characteristics of the new network and avoid a sudden drop in performance.

[0090] In this method, once the terminal successfully switches to the target network, it immediately sends update information to the anchor point via the newly established link, including the receive window state and the congestion window state. The receive window state reflects the maximum buffer size that the terminal can currently receive data from, while the congestion window state indicates the maximum number of bytes the data sender is allowed to send without acknowledgment. These two are key parameters in flow control and congestion control, used to dynamically adjust the data transmission rate and quantity to adapt to the real-time network conditions.

[0091] By implementing this control method, the data transmission performance and resource management capabilities of the adaptive communication protocol switching method in dynamic network environments are significantly enhanced. First, the dynamically adjusted receive window and congestion window states enable the system to more precisely control the data flow rate, avoiding buffer overflows or congestion crashes caused by changes in network conditions, thus ensuring smooth data transmission and proper operation of terminal devices. Second, this mechanism, through refined data flow control, effectively utilizes the target network's bandwidth resources, avoiding resource waste and improving the overall network throughput and efficiency.

[0092] Figure 3 This is a schematic diagram of an adaptive communication protocol switching device according to an embodiment of the present invention, such as... Figure 3 As shown, the system is divided into four core modules: First, the multi-dimensional state perception module collects information on the network environment, terminal status, and service requirements, and outputs feature vectors after data preprocessing; then, the intelligent prediction engine combines historical state data, performs prediction fusion through LSTM time series processing and CNN topology modeling, and generates a performance prediction matrix; subsequently, the protocol adaptation evaluation module calculates the matching degree based on the service QoS requirements and the capability evaluation results of the protocol library, and completes the protocol scoring and ranking; finally, the handover decision engine outputs instructions, and the seamless handover execution module realizes the seamless handover of protocols. Figure 3 The demonstrated structure embodies a complete closed-loop process from state awareness, intelligent prediction, protocol evaluation to decision execution. It is an important component in building a flexible, efficient, and user-friendly future network infrastructure, and has profound significance and application prospects for promoting the development of communication technology, especially in the context of heterogeneous network convergence in the 5G / 6G era.

[0093] As described above, the technical solution provided by the embodiments of the present invention constructs a three-level collaborative adaptive communication protocol switching system of "terminal-edge-cloud," achieving efficient collaboration between perception, decision-making, and execution through a layered architecture. The system is deployed in a typical heterogeneous network environment (including 5G NR, WiFi 6 / 7, and low-Earth orbit satellite networks), covering three typical scenarios: mobile terminals (smartphones / tablets), vehicle communication units, and industrial IoT gateways. The core of this system lies in a clearly defined and highly collaborative layered architecture. It first deploys "sensory nerves" on the terminal side. These terminals, including smartphones, vehicle units, and industrial gateways, are not ordinary devices. They are equipped with high-performance computing platforms and multi-mode communication modules capable of simultaneously connecting to 5G (including millimeter wave), Wi-Fi 6E, and even low-Earth orbit satellite networks. Furthermore, high-precision GNSS positioning modules and nine-axis motion sensors enable the terminals not only to perceive network signal strength but also to combine their precise position, speed, and motion status to provide richer contextual information for decision-making. Local storage on the terminal is used to cache key connection history and data, providing a basis for rapid decision-making. Above the terminals are the "decision-making hubs" deployed at the network edge. These edge nodes can be MEC servers in a 5G network or Wi-Fi 6 controllers for enterprises. Like regional command centers, they possess powerful computing capabilities, enabling them to monitor the "health status" of all network resources within their jurisdiction (e.g., a 500-meter radius) in real time, including channel load at each Wi-Fi access point, the number of users in the 5G cell, and satellite link occupancy. By maintaining real-time communication with the terminals, edge nodes can provide precise and timely switching instructions based on a more comprehensive local network view. For example, when Wi-Fi channel congestion is detected, the terminal can be quickly redirected to an idle 5G network slice. The "intelligent brain" supporting the continuous evolution of the entire system resides in the cloud. A cloud platform built on containerization technology runs a network digital twin engine and a long-term policy optimization model. It aggregates anonymized historical switching data from countless terminals and edge nodes across the network and uses this massive amount of data to continuously train reinforcement learning algorithms. This enables the system not only to address immediate problems but also to continuously learn, predict network congestion, and optimize future switching strategies, thereby achieving a spiral increase in the overall network resource utilization efficiency. In summary, this adaptive communication protocol switching system, through close cooperation between terminals, edge computing, and the cloud, achieves a closed loop from real-time perception and rapid decision-making to long-term optimization. It transforms the challenges of heterogeneous networks into opportunities, providing seamless, stable, and efficient high-quality connectivity for various future mobile and IoT applications.

[0094] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0095] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods of the various embodiments of this application.

[0096] Example 2

[0097] According to embodiments of the present invention, an adaptive communication protocol switching apparatus for implementing the above-described adaptive communication protocol switching method is also provided. Figure 4 This is a schematic diagram of an adaptive communication protocol switching device according to an embodiment of the present invention, such as... Figure 4 As shown, the device includes: an acquisition unit 401, a fusion unit 403, an input unit 405, a prediction unit 407, and a switching unit 409. The device will now be described in detail.

[0098] The acquisition unit 401 is used to acquire network environment data corresponding to the target network environment, terminal device data corresponding to the terminal devices in the target network environment, and service requirement data.

[0099] The fusion unit 403 is used to fuse network environment data, terminal device data and service requirement data to obtain fused feature data.

[0100] The input unit 405 is used to input the fused feature data into the network environment prediction model, so as to process the fused feature data using the network environment prediction model to obtain the performance prediction data of each candidate network in the target network environment in the future preset time window. The network environment prediction model is trained by machine learning using multiple sets of training data. Each set of training data includes: sample fused feature data and sample performance prediction data corresponding to the sample fused feature data.

[0101] The prediction unit 407 is used to predict the adaptability values ​​of each communication protocol in the protocol library under the performance prediction data.

[0102] The switching unit 409 is used to switch to the target communication protocol after selecting the target communication protocol based on the adaptability value and business requirement data.

[0103] It should be noted that the above-mentioned acquisition unit 401, fusion unit 403, input unit 405, prediction unit 407 and switching unit 409 correspond to steps S202 to S210 in the above embodiments. The five units and the corresponding steps implement the same instances and application scenarios, but are not limited to the content disclosed in the above embodiments.

[0104] As can be seen from the above, in the scheme described in the above embodiments of the present invention, an acquisition unit is used to acquire network environment data corresponding to the target network environment, terminal device data corresponding to the terminal devices in the target network environment, and service requirement data; a fusion unit is used to fuse the network environment data, terminal device data, and service requirement data to obtain fused feature data; an input unit is used to input the fused feature data into a network environment prediction model, so that the network environment prediction model can process the fused feature data to obtain performance prediction data of each candidate network in the target network environment in a future preset time window. The network environment prediction model is trained by machine learning using multiple sets of training data, and each set of training data includes: sample fused feature data and sample performance prediction data corresponding to the sample fused feature data; a prediction unit is used to predict the adaptability value of each communication protocol in the protocol library under the performance prediction data; and a switching unit is used to switch to the target communication protocol after selecting the target communication protocol based on the adaptability value and service requirement data. The above solution achieves the following: using a multi-dimensional state awareness module to collect and preprocess data related to network environment, terminal status, and service requirements in real time; utilizing a hybrid model of LSTM and GNN to make forward-looking predictions of the network environment; using a protocol adaptation evaluation module to analyze the performance of different protocols under predicted network conditions; and adopting a three-level collaborative "terminal-edge-cloud" architecture to achieve adaptive communication protocol switching. This improves communication quality, ensures service continuity and efficiency, forms a closed loop from real-time awareness and rapid decision-making to long-term optimization, supports smooth switching between different network environments, and reduces service interruptions and data loss.

[0105] Therefore, the technical solution provided by the above embodiments of the present invention solves the technical problem that traditional communication systems in the related art use fixed communication protocols and cannot be dynamically adjusted according to network environment and business needs, which has great limitations.

[0106] Optionally, the fusion unit includes: a denoising module for denoising network environment data, terminal device data, and service requirement data to obtain denoised network environment data, terminal device data, and service requirement data; a feature extraction module for extracting features from the denoised network environment data, terminal device data, and service requirement data to obtain network environment feature data, terminal device feature data, and service requirement feature data; and a standardization module for standardizing the network environment feature data, terminal device feature data, and service requirement feature data to obtain fused feature data.

[0107] Optionally, the prediction unit includes: a comparison module, used to compare the protocol feature data of each communication protocol in the protocol library with the performance prediction data respectively, and obtain the comparison result; and a quantization module, used to quantify and evaluate the adaptability value of each communication protocol under the performance prediction data based on the comparison result.

[0108] Optionally, the prediction unit includes: an analysis module for analyzing the performance indicators of each communication protocol in the predicted network environment corresponding to the performance prediction data, and obtaining the protocol evaluation results; and a mapping module for mapping the business requirement data to the protocol evaluation results, so as to calculate the adaptation capability value of each communication protocol under the performance prediction data.

[0109] Optionally, the switching unit includes: a determining module, used to determine the matching score between each communication protocol and the business requirements corresponding to the business requirements data based on the adaptability value; and a selecting module, used to select the target communication protocol from the protocol library based on the matching score.

[0110] Optionally, the switching unit includes: a connection request module, used to initiate a connection request to a target network corresponding to the target communication protocol while maintaining a connection with the original network; a parsing module, used to parse the feedback information after receiving the connection request and obtain the parsing result; an acquisition module, used to acquire a data handover address from the target network when the parsing result indicates that a connection with the target network has been successfully established, and register an anchor point through the data handover address, wherein the anchor point establishes a tunnel between the original network and the target network after receiving the registration information; and a forwarding module, used to forward all data to the target network through the tunnel via the anchor point.

[0111] Optionally, the adaptive communication protocol switching device further includes an update unit, used to update the receive window state and congestion window state to the anchor point through the new link corresponding to the target network after switching to the target network.

[0112] According to one aspect of the present invention, a processor is provided for running a program, wherein the program executes any of the above-described adaptive communication protocol switching methods during runtime.

[0113] According to one aspect of the present invention, a computer program product is provided, including computer instructions, which, when executed by a processor, perform any of the above-described adaptive communication protocol switching methods.

[0114] According to one aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein the program executes the adaptive communication protocol switching method described in any of the above.

[0115] Optionally, in this embodiment, the computer-readable storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any communication device in a group of communication devices.

[0116] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: acquiring network environment data corresponding to the target network environment, terminal device data corresponding to the terminal devices in the target network environment, and service requirement data; fusing the network environment data, terminal device data, and service requirement data to obtain fused feature data; inputting the fused feature data into a network environment prediction model to process the fused feature data using the network environment prediction model to obtain performance prediction data for each candidate network in the target network environment within a future preset time window, wherein the network environment prediction model is trained using multiple sets of training data through machine learning, and each set of training data includes: sample fused feature data and sample performance prediction data corresponding to the sample fused feature data; according to the adaptation capability value of each communication protocol in the protocol library under the performance prediction data; after selecting the target communication protocol according to the adaptation capability value and the service requirement data, switching to the target communication protocol.

[0117] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: denoising network environment data, terminal device data, and service requirement data to obtain denoised network environment data, terminal device data, and service requirement data; extracting features from the denoised network environment data, terminal device data, and service requirement data to obtain network environment feature data, terminal device feature data, and service requirement feature data; and standardizing the network environment feature data, terminal device feature data, and service requirement feature data to obtain fused feature data.

[0118] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: comparing the protocol feature data of each communication protocol in the protocol library with the performance prediction data to obtain comparison results; and quantitatively evaluating the adaptability value of each communication protocol under the performance prediction data based on the comparison results.

[0119] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: analyzing the performance indicators of each communication protocol in the predicted network environment corresponding to the performance prediction data to obtain the protocol evaluation results; mapping the service requirement data to the protocol evaluation results to calculate the adaptation capability value of each communication protocol under the performance prediction data.

[0120] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: determining the matching score between each communication protocol and the business requirements corresponding to the business requirements data based on the adaptability value; and selecting the target communication protocol from the protocol library based on the matching score.

[0121] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: while maintaining a connection with the original network, initiating a connection request to the target network corresponding to the target communication protocol; after receiving feedback information of the connection request, parsing the feedback information to obtain a parsing result; when the parsing result indicates that a connection with the target network has been successfully established, obtaining a data handover address from the target network and registering an anchor point through the data handover address, wherein the anchor point establishes a tunnel between the original network and the target network after receiving the registration information; and forwarding all data to the target network through the tunnel via the anchor point.

[0122] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: after switching to the target network, updating the receive window state and congestion window state to the anchor point through the new link corresponding to the target network.

[0123] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0124] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0125] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0126] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0127] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0128] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0129] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An adaptive communication protocol switching method, characterized in that, include: Obtain network environment data corresponding to the target network environment, terminal device data corresponding to the terminal devices in the target network environment, and service requirement data; The network environment data, the terminal device data, and the service requirement data are fused to obtain fused feature data; The fused feature data is input into the network environment prediction model to process the fused feature data using the network environment prediction model, thereby obtaining the performance prediction data of each candidate network in the target network environment in a future preset time window. The network environment prediction model is trained by machine learning using multiple sets of training data. Each set of training data includes: sample fused feature data and sample performance prediction data corresponding to the sample fused feature data. Based on the compatibility values ​​of each communication protocol in the protocol library under the performance prediction data; After selecting the target communication protocol based on the adaptation capability value and the business requirement data, switch to the target communication protocol.

2. The adaptive communication protocol switching method according to claim 1, characterized in that, The network environment data, the terminal device data, and the service requirement data are fused to obtain fused feature data, including: The network environment data, the terminal device data, and the service requirement data are denoised to obtain the denoised network environment data, terminal device data, and service requirement data. Feature extraction is performed on the noise-reduced network environment data, terminal device data, and service requirement data to obtain network environment feature data, terminal device feature data, and service requirement feature data. The network environment feature data, the terminal device feature data, and the service requirement feature data are standardized to obtain the fused feature data.

3. The adaptive communication protocol switching method according to claim 1, characterized in that, The evaluation criteria include the adaptability values ​​of each communication protocol in the protocol library under the performance prediction data, including: The protocol feature data of each communication protocol in the protocol library are compared with the performance prediction data to obtain the comparison results; The compatibility value of each of the communication protocols under the performance prediction data is quantitatively evaluated based on the comparison results.

4. The adaptive communication protocol switching method according to claim 1, characterized in that, Based on the adaptability values ​​of each communication protocol in the protocol library under the performance prediction data, including: The performance metrics of each of the aforementioned communication protocols in the predicted network environment corresponding to the performance prediction data are analyzed to obtain protocol evaluation results. The business requirement data is mapped to the protocol evaluation results to calculate the adaptability value of each communication protocol under the performance prediction data.

5. The adaptive communication protocol switching method according to claim 1, characterized in that, The target communication protocol is selected based on the adaptability value and the business requirement data, including: Based on the adaptation capability value, a matching score is determined between each of the communication protocols and the corresponding business requirements of the business requirement data. The target communication protocol is selected from the protocol library based on the matching score.

6. The adaptive communication protocol switching method according to claim 1, characterized in that, After selecting a target communication protocol based on the adaptability value and the business requirement data, switching to the target communication protocol includes: While maintaining the connection with the original network, a connection request is initiated to the target network corresponding to the target communication protocol; After receiving the feedback information of the connection request, the feedback information is parsed to obtain the parsing result; When the parsing result indicates that a connection with the target network has been successfully established, a data handover address is obtained from the target network, and an anchor point is registered through the data handover address. After receiving the registration information, the anchor point establishes a tunnel between the original network and the target network. All data is forwarded to the target network via tunnel through the anchor point.

7. The adaptive communication protocol switching method according to claim 6, characterized in that, The adaptive communication protocol switching method further includes: After switching to the target network, the receive window state and congestion window state are updated to the anchor point through the new link corresponding to the target network.

8. An adaptive communication protocol switching device, characterized in that, include: The acquisition unit is used to acquire network environment data corresponding to the target network environment, terminal device data corresponding to the terminal devices in the target network environment, and service requirement data. The fusion unit is used to fuse the network environment data, the terminal device data, and the service requirement data to obtain fused feature data; An input unit is used to input the fused feature data into a network environment prediction model, so as to process the fused feature data using the network environment prediction model to obtain the performance prediction data of each candidate network in the target network environment in a future preset time window. The network environment prediction model is trained by machine learning using multiple sets of training data. Each set of training data includes: sample fused feature data and sample performance prediction data corresponding to the sample fused feature data. The prediction unit is used to determine the adaptability values ​​of each communication protocol in the protocol library under the performance prediction data. The switching unit is used to switch to the target communication protocol after selecting the target communication protocol based on the adaptation capability value and the service requirement data.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein the program executes the adaptive communication protocol switching method according to any one of claims 1 to 7.

10. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by the processor, the adaptive communication protocol switching method according to any one of claims 1 to 7 is performed.

Images