Adaptive flow optimization gateway method and device for satellites and wireless private networks

Abstract

The invention relates to the technical field of communication networks, in particular to a self-adaptive flow optimization gateway method and device for satellites and wireless private networks. The method comprises the following steps: performing flow analysis and semantic analysis on service flow transmitted through a satellite and a wireless private network, and identifying at least one specific service type corresponding to the service flow; detecting current key performance indexes of the satellite and the wireless private network in real time, and aggregating the key performance indexes into link state vectors representing network conditions; performing multi-dimensional analysis on each specific service type, and respectively generating a service portrait of each specific service type; and fusing and analyzing the service portrait and the link state vector, generating an executable target instruction set for each specific service type, and executing the target instruction set. According to the invention, fine traffic scheduling and optimization can be carried out on satellite communication and a wireless private network.

Description

Technical Field

[0001] This invention relates to the field of communication network technology, and in particular to an adaptive traffic optimization gateway method and apparatus for satellite and wireless private networks. Background Technology

[0002] In critical industries such as ocean shipping, energy exploration, and emergency communications, satellite or private wireless networks are often the only available data transmission channels. These links have three inherent drawbacks that are difficult to overcome: they are usually charged by megabytes or gigabytes of data transmitted, resulting in extremely high bandwidth costs; geostationary satellite links have a round-trip delay of more than 600 milliseconds, causing huge propagation delays; and the signal is susceptible to weather, terrain obstruction, and electromagnetic interference during propagation, resulting in unstable channel quality, random packet loss, and severe fluctuations in throughput.

[0003] Existing network optimization technologies still have significant shortcomings in addressing the aforementioned inherent defects. For example, traditional Quality of Service (QoS) devices classify and prioritize traffic based on network layer information such as Internet Protocol (IP) addresses and port numbers. However, with the current trend of dynamic and encrypted application ports, they cannot effectively identify the nature of the business, leading to the mixing of critical services (such as industrial control commands) with ordinary services (such as software updates), resulting in poor scheduling performance. Transmission Control Protocol (TCP) accelerators can mitigate the impact of high latency on TCP throughput, but their one-size-fits-all acceleration strategies (such as ACK spoofing and window expansion) may have negative effects on real-time streaming media (such as exacerbating jitter) and may worsen congestion during periods of severe packet loss. Data compression devices indiscriminately compress all data passing through them; however, performing secondary compression on already highly compressed images, videos, or encrypted data not only fails to save bandwidth but also increases processing latency and power consumption. General-purpose deep packet inspection (DPI) devices are mainly used for traffic statistics, billing, or security filtering. However, their identification results are often disconnected from the link layer optimization decisions. Even if the application type is identified, there is a lack of an intelligent central hub that can dynamically adjust optimization strategies based on real-time network conditions. In summary, there is still a lack of effective methods for fine-grained traffic scheduling and optimization of satellite communications and wireless private networks.

[0004] Therefore, overcoming the shortcomings of the existing technology is an urgent problem to be solved in this technical field. Summary of the Invention

[0005] In view of the above-mentioned defects or improvement needs of the existing technology, the present invention proposes an adaptive traffic optimization gateway method and device for satellite and wireless private networks, which can perform fine-grained traffic scheduling and optimization for satellite communication and wireless private networks.

[0006] The embodiments of the present invention adopt the following technical solutions: In a first aspect, the present invention provides an adaptive traffic optimization gateway method for satellite and wireless private networks, specifically: performing traffic analysis and semantic parsing on service traffic transmitted through satellite and wireless private networks to identify at least one specific service type corresponding to the service traffic; Real-time detection of the current key performance indicators of the satellite and wireless private network, and aggregation of the key performance indicators into a link state vector representing the network status; Perform multi-dimensional analysis on each specific business type and generate a business profile for each specific business type; By integrating and analyzing business profiles and link status vectors, an executable target instruction set is generated for each specific business type, and the target instruction set is executed.

[0007] Preferably, the fusion analysis of the service profile and link state vector generates an executable target instruction set for each specific service type, and executes the target instruction set, including: Based on the business profile of each specific business type, multiple initial strategies are matched from the preset strategy library, where one initial strategy corresponds to one business profile. The link state vector is substituted into the benefit scoring function of each initial strategy for calculation to obtain the final target instruction set to be executed.

[0008] Preferably, the benefit scoring function is as follows: Si=Vi Ci; Vi=α·BW_Savingi+δ·Priority_Bonusi; Ci=β·Latency_Penaltyi+γ·CPU_Costi+λ·Riski; Where Si is the overall score of initial strategy i; Vi is the expected benefit of initial strategy i; Ci is the expected cost of strategy i; BW_Savingi is the estimated bandwidth savings brought by initial strategy i; Priority_Bonusi is the service criticality benchmark value, determined by the service level; α is the bandwidth economy coefficient; δ is the criticality reward coefficient; Latency_Penaltyi is the estimated additional latency introduced by initial strategy i; CPU_Costi is the estimated CPU resource consumption of initial strategy i; Riski is the adaptation risk estimate of initial strategy i under the current link state; β is the latency penalty coefficient; γ is the CPU cost coefficient; and λ is the risk penalty coefficient.

[0009] Preferably, the method further includes: For each initial strategy, the bandwidth savings, service criticality baseline, introduced additional latency, and consumed CPU resources are estimated, wherein the service criticality baseline is determined by the service level. Dynamic parameters are dynamically adjusted based on real-time link status. A comprehensive score for each initial strategy is calculated based on the bandwidth savings, the business criticality benchmark, the introduced additional latency, the consumed CPU resources, and the dynamic parameters. The initial strategy corresponding to the highest comprehensive score is taken as the target instruction set.

[0010] Preferably, the specific service type includes the transmission of data files, and the method further includes: The initial judgment result is obtained based on one or more of the protocol, content feature code or encrypted stream features of the data file. The initial judgment result includes clearly uncompressible, clearly compressible and tentatively compressed. By aggregating the file type of the data file and the initial judgment result, a business profile of the data file transmission is obtained.

[0011] Preferably, when the initial determination result of the data file is a trial compression type, the method further includes: Based on one or more of the connection mode, packet statistical characteristics, TLS handshake information, or protocol context of the data file, an information entropy value is assigned to the data file. Perform compression attempts on data files that are intended for trial compression to obtain an initial compression ratio; Based on the combination of the information entropy and the initial compression rate, a final judgment result is obtained, wherein the final judgment result includes compressible, incompressible, or ambiguous.

[0012] Preferably, when the final judgment result is ambiguous, the method further includes: If the information entropy is less than or equal to the first entropy threshold, the initial compression rate is less than or equal to the first compression threshold, and the current gateway's CPU load rate is less than the preset high load threshold, then conservative compression is enabled for the data file and monitoring is performed. If the information entropy is greater than the second entropy threshold, or the initial compression ratio is greater than the second compression threshold, or the information entropy is greater than the first entropy threshold and less than or equal to the second entropy threshold, and the initial compression ratio is greater than the first compression threshold and less than or equal to the second compression threshold, or the current CPU load rate is greater than or equal to a preset high load threshold, then the data file will be directly passed through.

[0013] Preferably, the method further includes: After all optimization instruction sets have been executed, record multiple metadata related to the business and store these metadata in the performance archive. The performance archive is analyzed periodically to evaluate the overall benefits generated by the execution of each optimized instruction set. Optimized instruction sets that are verified to consistently generate the highest overall benefits are promoted to high-priority strategies and incorporated into the automatic matching and scheduling mechanism of subsequent tasks.

[0014] In a second aspect, the present invention provides an adaptive traffic optimization gateway device for satellite and wireless private networks, the device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the processor to perform the adaptive traffic optimization gateway method for satellite and wireless private networks in the first aspect.

[0015] Thirdly, the present invention also provides a non-volatile computer storage medium storing computer-executable instructions, which are executed by one or more processors to perform the adaptive traffic optimization gateway method for satellite and wireless private networks described in the first aspect.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: by analyzing the service traffic transmitted through satellite and wireless private network, the specific service type at present is identified and a service profile is created, the current link status of satellite and wireless private network is obtained, and the service profile and the current link status are integrated and analyzed to generate a target instruction set exclusive to each specific service type. This method does not optimize the link layer indicators in a general way, but performs fine-grained traffic scheduling and optimization of satellite communication and wireless private network to ensure the service quality of each specific service type. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments of the present invention will be briefly described below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0018] Figure 1 This is a flowchart illustrating an adaptive traffic optimization gateway method for satellite and wireless private networks provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the process for obtaining the target instruction set provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the process for obtaining a business profile provided in an embodiment of the present invention; Figure 4 This is a flowchart illustrating the method for obtaining the final judgment result provided in an embodiment of the present invention; Figure 5 This is a flowchart illustrating the method for determining based on sampling compression provided in an embodiment of the present invention; Figure 6 This is a flowchart illustrating a method for further determining data files provided in an embodiment of the present invention; Figure 7 A schematic diagram of the structure of an adaptive traffic optimization gateway device for satellite and wireless private networks provided in an embodiment of the present invention; The reference numerals in the attached figures are as follows: 21: Processor; 22: Memory. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] Unless the context otherwise requires, throughout the specification and claims, the term "comprising" is interpreted as openly inclusive, meaning "including, but not limited to." In the description of the specification, terms such as "one embodiment," "some embodiments," "exemplary embodiment," "example," "specific example," or "some examples" are intended to indicate that a particular feature, structure, material, or characteristic associated with that embodiment or example is included in at least one embodiment or example of this disclosure. The illustrative representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics mentioned may be included in any suitable manner in any one or more embodiments or examples; that is, although they may be incorporated into embodiments or examples using the above terms for reasons such as order and position, it does not limit them to be incorporated in combination by a single embodiment or example.

[0021] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of this disclosure, unless otherwise stated, "a plurality of" means two or more. Furthermore, for example, the description may use the prefix "A" or "B" to describe the same type of nouns as two independent entities. In this case, the corresponding features defined with "A" and "B" are used only to distinguish between similar entities and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features.

[0022] In the description of this invention, the expression “A and / or B” (where A and B are used to formally represent specific features) will be used. The corresponding expression includes the following three combinations: only A, only B, and a combination of A and B.

[0023] As used in this invention, “about,” “approximately,” or “approximately” includes the stated value and the average value within an acceptable range of deviation from a particular value, wherein the acceptable range of deviation is determined by a person skilled in the art taking into account the measurement under discussion and the error associated with the measurement of the particular quantity (i.e., the limitations of the measurement system).

[0024] Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0025] Example 1: This embodiment provides an adaptive traffic optimization gateway system for satellite and wireless private networks to implement the method described in this embodiment. The optimization gateway system is deployed at the edge of high-latency, high-cost, and unstable links such as satellite, maritime, and oil and gas field wireless private networks. The optimization gateway system includes a deep perception and recognition module, a dynamic policy engine, an optimization policy execution unit set, and a learning and policy optimization interface. The modules interact through a high-speed data bus or shared memory, forming an enhanced loop from perception to decision-making to execution to learning. The deep perception and recognition module is responsible for understanding the inherent semantics and external environment of network traffic; the dynamic policy engine, as the intelligent hub of the system, performs real-time calculations and arbitration; the optimization policy execution unit set is responsible for accurately executing various optimization instructions; and the learning and policy optimization interface enables the system's self-evolution and long-term adaptation.

[0026] like Figure 1 As shown, Embodiment 1 of the present invention provides an adaptive traffic optimization gateway method for satellite and wireless private networks, which specifically includes the following steps: Step 101: Perform traffic analysis and semantic parsing on the service traffic transmitted via satellite and wireless private network to identify at least one specific service type corresponding to the service traffic.

[0027] In one embodiment, the depth perception and recognition module provides high-fidelity contextual input for subsequent decision-making. The depth perception and recognition module includes a deep packet inspection and semantic parsing unit, which is used to identify the specific service type of the service traffic. The specific service type includes control commands, video conferencing, data file transmission, etc., and the service traffic is the traffic generated by the transmission of the specific service type through satellite or wireless private network.

[0028] Specifically, the specific business type can be identified by parsing the protocols used in the business traffic. The deep packet inspection and semantic parsing unit integrates a deep parser for industrial protocols (such as Modbus / TCP, DNP3). The deep parser can not only identify the protocol type but also perform semantic-level parsing to obtain the specific business type. For example, it can extract the function code (0x10 indicates a write operation) and register address from the Modbus message to determine whether an instruction is a critical valve control or a routine status read. For encrypted traffic (such as HTTPS), by fusing metadata and flow behavior analysis, key information of the Transport Layer Security (TLS) handshake phase (such as SNI (Service Node Interface)) is extracted. At the same time, combined with machine learning models, the statistical characteristics of the data packets of the business traffic, such as timing, size, and flow direction, are modeled. Thus, even with a fully encrypted payload, it can still accurately identify different business types such as video conferencing, file downloading, and web page interaction.

[0029] Step 102: Real-time detection of the current key performance indicators of the satellite and wireless private network, and aggregation of the key performance indicators into a link state vector representing the network status.

[0030] In one embodiment, the deep perception and recognition module further includes a real-time link status probe unit. This real-time probe unit continuously acquires key performance indicators of the link by actively sending probe packets and passively monitoring the network interface card status. The key performance indicators include round-trip delay, packet loss rate, jitter, and available bandwidth estimation. The acquired key performance indicators are aggregated into a link status vector that characterizes the network status. The link status includes delay, packet loss rate, jitter, and bandwidth margin, for example, {delay: 720ms, packet loss rate: 8%, jitter: ±50ms, bandwidth margin: 1.2Mbps}.

[0031] Step 103: Perform multi-dimensional analysis on each specific business type and generate a business profile for each specific business type.

[0032] In one embodiment, the content of the business profile varies depending on the specific business type. When the specific business type is control command or video conferencing, the corresponding business profile includes business type, criticality level, specific operation, and flow mode. For example, {Business type: Industrial control, Criticality level: P1 (highest), Specific operation: Write to pressure valve, Flow mode: Sudden command}. When the specific business type is data file transmission, the corresponding business profile includes file type and initial judgment result. For example, {File: Text log, Initial judgment: Highly compressible}.

[0033] Step 104: Integrate and analyze the business profile and link status vector to generate an executable target instruction set for each specific business type, and execute the target instruction set.

[0034] In one embodiment, in key industries such as ocean shipping, energy exploration, and emergency communications, satellite or private wireless networks are mainly used for transmitting control commands, periodic data, and data files. For this specific application, this embodiment closely integrates the optimization strategies in the target instruction set with its application scenario, aiming to finely optimize key aspects such as scheduling strategies, Transmission Control Protocol (TCP) behavior, transmission reliability, and data compression, thereby improving overall communication efficiency and system performance. The dynamic policy engine is the brain of the optimization gateway system, upgrading decision-making from static rule matching to dynamic scoring and arbitration based on real-time utility. Through the scoring strategy, the target instruction set corresponding to each specific service type is obtained. Executing the target instruction set allows for fine-grained traffic scheduling and optimization of satellite communication and private wireless networks, ensuring the quality of service and transmission priority of each service type.

[0035] In this embodiment, by analyzing the service traffic transmitted via satellite and wireless private network, the specific service type is identified and a service profile is created. The current link status of the satellite and wireless private network is obtained. The service profile and the current link status are integrated and analyzed to generate a target instruction set specific to each specific service type. This method does not optimize the link layer indicators in a general way, but performs fine-grained traffic scheduling and optimization for satellite communication and wireless private network to ensure the service quality of each specific service type.

[0036] like Figure 2 As shown, this embodiment provides a method for obtaining a target instruction set, which specifically includes the following steps: Step 201: Based on the business profile of each specific business type, multiple candidate strategies are matched from the preset strategy library, where one business profile corresponds to one initial strategy.

[0037] In one embodiment, by matching the business type or file type in the business profile with the application scenario in the preset strategy library, an initial strategy corresponding to a specific business type can be obtained. The initial strategy includes extreme reliability mode, high throughput mode and low latency mode. See Table 1 for a strategy template example. Table 1 is just an example and does not limit this embodiment.

[0038] Table 1. Strategy Template Examples

[0039] Optionally, users can customize and solidify their initial strategies for specific business types to ensure that the business always adopts the specified strategies during operation, unaffected by dynamic system adjustments or other strategy changes.

[0040] Step 202: Substitute the link state vector into the benefit scoring function of each initial strategy to calculate the final target instruction set to be executed.

[0041] In one embodiment, the benefit scoring function is used to quantify the expected benefits (such as bandwidth savings and latency reduction) and costs (such as Central Processing Unit (CPU) consumption and additional latency) of implementing candidate strategies in the current network environment. By quantifying the net benefit of the strategy, the benefit scoring function aims to select the optimal solution under specific network conditions, ultimately ensuring that each specific service type has an optimal strategy to guarantee service. The benefit scoring function is as follows: Si=Vi Ci; Vi=α·BW_Savingi+δ·Priority_Bonusi; Ci=β·Latency_Penaltyi+γ·CPU_Costi+λ·Riski; Where Si is the overall score of initial strategy i; Vi is the expected benefit of initial strategy i; Ci is the expected cost of strategy i; BW_Savingi is the estimated bandwidth savings brought by initial strategy i; Priority_Bonusi is the service criticality benchmark value, determined by the service level; α is the bandwidth economy coefficient; δ is the criticality reward coefficient; Latency_Penaltyi is the estimated additional latency introduced by initial strategy i; CPU_Costi is the estimated CPU resource consumption of initial strategy i; Riski is the adaptation risk estimate of initial strategy i under the current link state; β is the latency penalty coefficient; γ is the CPU cost coefficient; and λ is the risk penalty coefficient.

[0042] The expected benefits measure the direct economic benefits of the initial strategy and its value in ensuring critical business operations. α·BW_Savingi reflects the economic benefits of saving expensive bandwidth, while δ·Priority_Bonusi reflects the business value gained in ensuring highly critical business operations. The dynamic changes of coefficients α and δ determine the real-time trade-off between saving money and ensuring business operations. The expected costs quantify the resource costs incurred in implementing the strategy and the risks of incompatibility with the network environment. β·Latency_Penaltyi represents the experience cost caused by processing latency; γ·CPU_Costi represents the cost of computational resource consumption; and λ·Riski represents the risk cost of transmission failure or increased congestion due to the strategy's incompatibility with the current poor link (such as high packet loss).

[0043] Risk values ​​are obtained in real time through a pre-defined policy-link state mapping table. This mapping table defines the inherent risk levels of various basic network optimization actions (such as TCP acceleration, forward error correction, and data compression) under different network state characteristics (such as high packet loss and high jitter). The engine deconstructs candidate policies into atomic actions and matches them with the current link state classification. It then retrieves the risk values ​​of each action from the table and aggregates them to obtain the overall risk value of the policy. This mapping table can be further linked with the system learning module to dynamically update its risk valuation based on historical performance data, thereby achieving continuous optimization of risk assessment capabilities. Table 2 shows the risk valuation design for some key policy actions.

[0044] Table 2 Risk Valuation Table

[0045] like Figure 3 This embodiment provides a method, which specifically includes the following steps: Step 301: Estimate the bandwidth savings, service criticality baseline, introduced additional latency, and consumed CPU resources for each initial strategy, wherein the service criticality baseline is determined by the service level.

[0046] In one embodiment, the initial strategy not only brings varying degrees of bandwidth savings but also has a profound impact on system performance, user experience, and business continuity. For example, enabling intelligent compression algorithms can significantly reduce the amount of data transmitted, but may increase the CPU load on terminal devices; while deploying edge caching can significantly reduce back-to-origin bandwidth and lower latency, its effectiveness is highly dependent on content popularity and user access patterns. Furthermore, different business scenarios have vastly different tolerances for network behavior. For instance, financial transaction systems require millisecond-level response times and zero packet loss, with their business-critical baseline values ​​far exceeding those of non-critical tasks such as background log synchronization. Therefore, before implementing the final strategy, it is necessary to comprehensively estimate its performance across four dimensions: bandwidth savings, business-critical baseline values, introduced additional latency, and CPU resource consumption. This will achieve an optimal balance between resource constraints and service quality, supporting highly reliable and efficient digital operations.

[0047] Step 302: Dynamically adjust dynamic parameters based on real-time link status.

[0048] In one embodiment, the coefficients α, β, γ, λ, and δ are not fixed values ​​but dynamically adjusted based on the real-time link status. For example, when a link packet loss rate > 10% is detected (poor link performance), the risk penalty coefficient λ is significantly increased, putting the system into a "risk-averse" mode. The score of any high-risk strategy (such as aggressive TCP acceleration) will drop sharply, increasing the critical reward coefficient δ and further widening the revenue gap between different critical service levels. This drives all resources to prioritize P1-level services, reducing the bandwidth economy coefficient α. In this case, ensuring connectivity and reliability is far more important than saving bandwidth. When link jitter > 50ms is detected (unstable link), the latency penalty coefficient β is significantly increased. The system becomes extremely sensitive to any operations that may increase latency (such as complex compression). In this state, strategies including "anti-jitter buffers" are preferred for interactive video, voice, and other services. When link latency is high but packet loss rate is low (stable but slow link), the bandwidth economy coefficient α is increased. This encourages maximizing bandwidth utilization through intelligent compression and other technologies in a stable environment, saving costs and reducing the risk penalty coefficient λ. This allows for more relaxed experimentation with high-return but slightly risky optimization strategies.

[0049] Step 303: Calculate the comprehensive score for each initial strategy based on the bandwidth savings, the business criticality benchmark, the introduced additional latency, the consumed CPU resources, and the dynamic parameters.

[0050] In one embodiment, the bandwidth savings from steps 301 and 302, the business-critical baseline, the introduced additional latency, the consumed CPU resources, and the dynamic parameters are fed into the benefit scoring function to obtain a comprehensive score for each initial strategy.

[0051] Step 304: Use the initial strategy corresponding to the highest comprehensive score as the target instruction set.

[0052] In one embodiment, the comprehensive score obtained in step 303 can ensure that even if the net returns of all strategies are negative in extreme environments, it can accurately identify strategies with relatively lower risk and higher reliability, thereby ensuring the absolute security of the highest-level business and reflecting the core value of intelligent arbitration.

[0053] Furthermore, when multiple service flows simultaneously compete for system resources (such as egress bandwidth), the policy engine employs a two-level arbitration mechanism for global optimization. First, assuming sufficient resources, it independently calculates the scores of all candidate policies for each service flow and records the highest score S_opt and its corresponding policy for each flow. Based on priority-based resource allocation and dynamic allocation, the system pre-sets virtual resource quotas for services of different priority levels according to configuration (e.g., P1 service enjoys 50% guaranteed bandwidth). Within their respective quotas, service flows of each level are scheduled according to their optimal policy score S_opt. When the optimal policy of a high-priority service (such as P1) requires additional resources (e.g., sending a high proportion of FEC (Forward Error Correction) redundant packets), it can dynamically "borrow" quotas from low-priority services. The system then recalculates the policy scores of the affected low-priority services, as the reduction in their available bandwidth leads to an increase in the risk value (Riski) of their high-bandwidth-demand policies, thus driving these services to automatically switch to more economical and conservative alternative policies. This mechanism ensures that low-priority services proactively yield to high-priority services during resource shortages, guaranteeing their absolute needs.

[0054] To clearly describe the solution that yields the optimal strategy, this example provides a detailed explanation. In a scenario where a sandstorm causes severe degradation of the private wireless network (20% packet loss rate, 80ms jitter) and a critical P1 level control command needs to be sent, strategy A (maximum reliability) is {strong FEC (40%), local retransmission, compression: disabled}. With estimated BW_SavingA=0MB, Latency_PenaltyA=5ms, CPU_CostA=15%, RiskA=10, and Strategy B (Balanced Mode) set to {Standard FEC (15%), Compression: Disabled}, and estimated BW_SavingB=0MB, Latency_PenaltyB=1ms, CPU_CostB=5%, RiskB=50, the dynamic coefficient settings are adjusted according to the severe link condition to α=1, β=10, γ=1, λ=20, and δ=1, and P1 service Priority_Bonus=100. The overall score is calculated as follows: =(α·0+δ·100) (β·5+γ·15+λ·10) =(0+100) (50+15+200)=(0+100) (50+15+200) =100 265= 165 = 100 265= 165; =(α·0+δ·100) (β·1+γ·5+λ·50) =(0+100) (10+5+1000)=(0+100) (10+5+1000) =100 1015= 915=100 1015= 915; Since the score of strategy A (-165) is much higher than that of strategy B (-915), the system selects strategy A.

[0055] Furthermore, the business profile also includes business criticality levels. When multiple business traffic competes for resources, the dynamic strategy engine will make the final arbitration based on the business criticality level of the specific business type to ensure that P1 level tasks (such as security shutdown) have absolute priority.

[0056] In one embodiment, when the specific service type is the transmission of critical control commands and periodic data transmission, the extreme reliability guarantee mode is matched. For identified critical write commands (such as SCADA (Supervisory Control And Data Acquisition) control), while applying the highest priority and strong forward error correction (FEC), the gateway will establish a session mirror locally and perform millisecond-level (e.g., 5ms) fast retries. If no remote confirmation is received, it will immediately retransmit from the local copy, completely avoiding the hundreds of milliseconds of end-to-end timeout retransmission of satellite links, and improving the reliability of the control loop from second-level guarantee to millisecond-level guarantee. For massive sensor status query responses, the gateway does not simply forward them, but performs temporary caching and protocol aggregation at the near end, packaging the payloads of multiple small messages and sending them together. This can reduce protocol header overhead and the number of small packets by up to 90%, greatly alleviating link congestion and making way for critical commands.

[0057] When the specific business type is the transmission of data files, the target instruction set covers the data compression mode. Since some data files are already highly compressed, performing secondary compression on already highly compressed images, videos, or encrypted data will not only fail to save bandwidth but will also increase device processing latency and power consumption. Some data files are incompressible, so before consuming resources for sampling compression, a lighter and faster method is used to predict the data stream and classify it into different processing channels to achieve precise resource allocation.

[0058] like Figure 4As shown in the figure, this embodiment provides a method for obtaining a business profile, which specifically includes the following steps: Step 401: Make a judgment based on one or more of the protocol, content feature code or encrypted stream features of the data file to obtain an initial judgment result, wherein the initial judgment result includes clearly uncompressible class, clearly compressible class and tentative compression class.

[0059] In one embodiment, the method for determining whether a data file is explicitly uncompressible is as follows: Based on the protocol and port characteristics of the data file, if the protocol is identified as a known streaming media protocol (such as Real Time Streaming Protocol (RTSP)) or a protocol carrying compressed content (such as a video stream carried via TLS), the data file can be classified as explicitly uncompressible. For unencrypted traffic, the "magic number" in the packet header is checked. For example, FF D8 FF indicates JPEG, 25 50 44 46 indicates PDF, and 1F 8B indicates GZIP. These are clear indicators of compression or high structure. When clear indicators are detected in the data packets of a data file, the data file can be classified as explicitly uncompressible. If the data file is identified as an encrypted stream through a TLS handshake, and its Server Name Indicator (SNI) or certificate information points to a known video service, cloud storage, etc., the data file can be classified as explicitly uncompressible.

[0060] In one embodiment, the method for determining whether a data file is explicitly compressible is as follows: by analyzing the plaintext protocol and content, if the protocol header of the data file is HTTP / 1.1 and the Content-Type is text / plain, application / json, application / xml, etc., then the data file can be classified as explicitly compressible; if the protocol of the data file is identified as some proprietary industrial text protocol (such as an unencrypted specific SCADA report format), then the data file can be classified as explicitly compressible.

[0061] In one embodiment, if the data files cannot be accurately classified using the above method, they are classified as experimental compression files. The fundamental challenge of experimental compression is the high degree of uncertainty of the information.

[0062] Step 402: Aggregate the file type of the data file and the initial judgment result to obtain the business profile of the data file transmission.

[0063] In one embodiment, a corresponding target instruction set can be matched based on the business profile. The compression mode within the target instruction set varies depending on whether the data file is compressed. For data files explicitly classified as uncompressible, a judgment is made based on network quality and system resource usage. If resources are scarce, the data is directly marked for pass-through without entering any compression process, resulting in zero resource consumption. In real-world networks, already compressed data (such as ZIP and JPEG) has extremely high information entropy and is almost random. Sampling and compressing it will result in a compression rate approaching 100%, offering almost no benefit from compression.

[0064] Furthermore, if network quality and system resources are good, sampling compression will be performed on explicitly incompressible data streams. For example, the first 64KB of the load portion of each stream will be attempted to compress. Only if the measured compression rate is significantly lower than a threshold (e.g., <85%) will the compression be considered effective, and the server will be flagged (reducing its entropy). Subsequent data packets will continue to be compressed, but even if the initial compression is effective, the system will continuously monitor the compression rate. If the compression rate deteriorates subsequently, the system will dynamically revert to pass-through mode.

[0065] For data files that are clearly compressible, a high-efficiency compression algorithm optimized for text (such as Zstd) is directly called, skipping the sampling and probing phase to achieve maximum compression efficiency.

[0066] The fundamental challenge of experimentally compressed data files is the high degree of uncertainty in information. The system cannot know the content of the traffic in advance. Therefore, it is necessary to explore the optimal solution through a progressive, feedback-driven process. The core logic of this process is to obtain the key information for benefit decisions in the shortest possible time and at the lowest possible cost, and to safely perform compression or abandon compression accordingly.

[0067] As shown in Figure 5, this embodiment provides a method for obtaining the final judgment result, which specifically includes the following steps: Step 501: Based on one or more of the connection mode, packet statistical characteristics, TLS handshake information, or protocol context of the data file, assign an information entropy value to the data file.

[0068] In one embodiment, the system first determines the type of traffic based on one or more of the following: connection mode, packet statistics, TLS handshake information, or protocol context. If the traffic is determined to be compressible, a low entropy value is assigned; if it is determined to be compressed, a high entropy value is assigned. The connection mode identifies the traffic type. Persistent, unidirectional, large traffic streams typically correspond to compressed content (such as videos or installation packages) and can be assigned a high entropy value. Interactive, bidirectional, small packet streams, such as database queries and Application Programming Interface (API) calls, often contain structured text, such as JavaScript key-value pairs (JSON) or Extensible Markup Language (XML), which are highly compressible and should be assigned a low entropy value. Secondly, analyzing packet length distribution and arrival intervals reveals that consistently large packets often originate from media or compressed files, while streams with fluctuating packet lengths and irregular intervals are more likely to contain redundant text or metadata and have compression potential. Furthermore, by combining protocol context information (such as the `RETR filename.zip` command in File Transfer Protocol (FTP)) or URL paths (such as ` / api / vs / download / `), the type of content being transmitted can be known in advance. Additionally, in encrypted traffic (such as Hypertext Transfer Protocol Secure (HTTPS), although the payload cannot be parsed, the Server Name Indicator (SNI) and Certificate Authority information exposed during the TLS handshake phase can be used to assist in judgment. For example, accessing `update.example.com` tends to download mixed compressed packages, while `api.example.com` mostly contains compressible text data.

[0069] Step 502: Perform a compression attempt on the data files of the trial compression type to obtain the initial compression ratio.

[0070] In one embodiment, the data file is sampled within a fixed-size sampling window to obtain sampled data. This sampled data is then compressed using a compression algorithm to obtain an initial compression ratio. A fixed-size sampling window is set, which should be large enough to cover representative data samples, yet small enough to avoid excessive time and bandwidth consumption on incompressible data; for example, a sampling window of 64KB is used. Furthermore, the data file is temporarily stored during sampling, introducing minimal and manageable additional latency. The sampled data is compressed once using a high-speed, low-overhead compression algorithm (such as LZ4), and the initial compression ratio is calculated. The compression ratio is the ratio of the compressed file size to the original file size. For example, compressing a 100MB file to 90MB results in a compression ratio of 90 / 100 × 100% = 90%. Generally, a lower compression ratio is better, but a lower ratio leads to longer decompression time.

[0071] Step 503: Based on the combination of the information entropy and the initial compression rate, a final judgment result is obtained, wherein the final judgment result includes compressible, incompressible, or fuzzy.

[0072] In one embodiment, a dual-threshold-based instantaneous decision-making mechanism is employed, and the judgment criteria are not limited to compression ratio. Specifically, a dual-threshold judgment matrix is ​​constructed, combining the information entropy and the initial compression ratio to further classify the files attempting compression into one of the following categories: compressible, incompressible, or ambiguous. Specifically: if the information entropy is less than or equal to the third entropy threshold (e.g., the third entropy threshold is 6.5), and the initial compression ratio is less than or equal to the third compression threshold (e.g., the third compression threshold is 85%), then the data files in the attempting compression category are further classified as compressible; if the information entropy is greater than or equal to the fourth entropy threshold (e.g., the fourth entropy threshold is 7.8), and the initial compression ratio is greater than or equal to the fourth compression threshold (e.g., the fourth compression threshold is 98%), then the data files in the attempting compression category are further classified as incompressible; if the initial compression ratio and information entropy do not meet the above judgment conditions, then the data files in the attempting compression category are further classified as ambiguous. Table 3 is the dual-threshold judgment matrix. Table 3 is only an example and does not constitute a limitation of this embodiment.

[0073] Table 3. Dual Threshold Judgment Matrix

[0074] like Figure 6 As shown, when a data file is classified as ambiguous, this embodiment provides a method for further determining the data file. This method specifically includes the following steps: Step 601: If the information entropy is less than or equal to the first entropy threshold, the initial compression rate is less than or equal to the first compression threshold, and the current gateway's CPU load rate is less than the preset high load threshold, then conservative compression is enabled for the data file and monitoring is performed.

[0075] In one embodiment, when the data file belongs to the "compressible" sub-interval and current resources are sufficient, it ensures that there are idle computing resources available for this exploratory operation with uncertain benefits, without affecting the confirmed high-priority tasks. In this case, an attempt can be made at the lowest cost under the premise that the expected benefits are positive but not maximized. In this case, a low-overhead, high-speed lightweight compression algorithm (such as LZ4 fast mode or Zstd level 1 compression) is enabled to compress the data file to minimize CPU and latency costs, obtain limited bandwidth savings, and immediately start enhanced monitoring, shorten the benefit evaluation window (e.g., change to evaluate once every 256KB of data), and track the changes in compression ratio Rc in real time.

[0076] Specifically, when the information entropy is less than or equal to the first entropy threshold (e.g., the first entropy threshold is 7.0), and the initial compression ratio is less than or equal to the first compression threshold (e.g., the first compression threshold is 0.85), it indicates that the data file belongs to the "prone to compressibility" sub-interval. The characteristics of the "prone to compressibility" sub-interval are that it is closer to the high-efficiency compressibility boundary, the data order is relatively good, and there is a clear compression potential. The current CPU load rate of the gateway is lower than the preset high load threshold (e.g., the high load threshold is 70%), which indicates that the current resources are sufficient.

[0077] In one embodiment, if subsequent monitoring shows that the compression ratio remains stable within an acceptable range (e.g., consistently Rc ≤ 0.9), the conservative compression mode is maintained. If the compression ratio continues to deteriorate (e.g., Rc > 0.95 for two consecutive monitoring windows), the attempt is deemed a failure, and the system immediately triggers a "smooth rollback" mechanism, seamlessly switching to pass-through mode. This "stream feature-result" pair is recorded in the learning library to optimize future initial judgment strategies for similar streams. This refined strategy is entirely built upon the "fuzzy region" defined by the initial decision matrix. By introducing two dimensions—data feature sub-interval division and system resource status—for secondary arbitration, an intelligent transformation from "fuzzy" to "explicit execution instructions" is achieved. This demonstrates the system's ability to make optimal engineering decisions that balance efficiency and robustness even with limited information.

[0078] Step 602: If the information entropy is greater than the second entropy threshold, or the initial compression ratio is greater than the second compression threshold, or the information entropy is greater than the first entropy threshold and less than or equal to the second entropy threshold, and the initial compression ratio is greater than the first compression threshold and less than or equal to the second compression threshold, or the current CPU load rate is greater than or equal to the preset high load threshold, then the data file will be directly passed through.

[0079] In one embodiment, regardless of whether the data file belongs to a predominantly invalid sub-interval, a highly uncertain interval, or when current resources are scarce, investing valuable computing power in a task with uncertain returns would have excessive opportunity costs and would not conform to the principle of maximizing global benefits in the dynamic scoring model. Therefore, such data files are directly passed through, which ensures the lowest latency and zero additional CPU consumption.

[0080] Specifically, when the information entropy is greater than the second entropy threshold (e.g., the second entropy threshold is 7.5) and the initial compression ratio is greater than the second compression threshold (e.g., the second compression threshold is 0.9), the data file belongs to the "prone to invalidity" sub-interval. The characteristics of the "prone to invalidity" sub-interval are that it is closer to the "incompressible" boundary, the data has high randomness, the compression benefit is extremely low, and the risk of attempting compression (such as data expansion) outweighs the potential benefit. When the information entropy is greater than the first entropy threshold and less than or equal to the second entropy threshold, and the initial compression ratio is greater than the first compression threshold and less than or equal to the second compression threshold, the data file belongs to the highly uncertain interval. The data characteristics of the highly uncertain interval are extremely vague, lacking a clear benefit orientation. If the current CPU load rate exceeds the high load threshold (e.g., the high load threshold is 70%), it indicates that the current resource situation is strained.

[0081] For example, for an unknown encrypted HTTPS stream (potentially a document or media), SNI identifies it as SharePoint.com, initially profiling it as "enterprise document collaboration." Sampling analysis shows a moderate entropy value and an LZ4 compression rate of 70%, classifying it as compressible (likely an Office document), and Zstd compression is enabled. However, monitoring reveals that after a few minutes of compression, the compression rate suddenly deteriorates to 95%. The system infers that the stream may have switched to a compressed file (e.g., a user uploaded a ZIP archive), and the dynamic adjustment strategy immediately and smoothly reverts to pass-through. For an unknown file transferred via FTP, where the control channel does not explicitly indicate the file type, the stream characteristics show a stable, large stream, and the sampled entropy value is as high as 7.9. After sampling and compression, the initial compression rate is 98%, immediately identifying it as a highly likely compressed file (e.g., ISO, RAR (Roshal Archive)), abandoning compression, and directly passing it through. For API data that claims to be text but has an unknown format, uses a non-standard port, has varying packet sizes, exhibits request and response patterns, and shows low entropy values ​​(e.g., entropy of 5.1), and has an LZ4 compression rate of up to 40%, it is determined to be strongly compressible. In this case, a high compression rate algorithm is enabled. This "feature-decision-efficiency" result is captured by the learning feedback module, and similar streams from the same destination IP can be directly compressed in the future, skipping the probing.

[0082] In one embodiment, the system also provides a learning and strategy optimization interface, specifically: after all optimized instruction sets are executed, multiple metadata corresponding to the business are recorded and stored in a performance archive. These metadata include traffic characteristics, target instruction sets, link status, and final benefits. The performance archive is periodically analyzed, and the comprehensive benefits generated by each optimized instruction set are evaluated using a benefit model. Optimized instruction sets that are verified to consistently generate the highest comprehensive benefits are promoted to high-priority strategies and incorporated into the automatic matching and scheduling mechanism for subsequent tasks. For example, when it is found that a certain strategy (such as Zstd-4 level compression) consistently generates the highest benefits for a certain type of traffic (such as SQL backup streams sent to a backup server), the system can automatically generate or optimize a high-priority strategy rule, achieving an intelligent evolution from "manual configuration" to "data-driven" approaches.

[0083] Furthermore, during the execution of the target instruction set, the current key performance indicators of the satellite and the wireless private network are detected in real time to obtain the current link state vector. Based on the current link state vector, the target instruction set for each specific service type is changed in real time to obtain an optimized instruction set, and the optimized instruction set is executed.

[0084] Furthermore, during the execution of the target instruction set, the information entropy and compression ratio of the transmitted data file are calculated in real time; based on the currently calculated information entropy and compression ratio, the compression mode of the data file is dynamically adjusted to ensure optimal transmission efficiency and resource utilization.

[0085] Based on the above adaptive flow optimization method, this embodiment takes the offshore drilling platform scenario as an example to describe the solution in detail.

[0086] A deepwater pressurization pump on an offshore drilling platform suddenly shut down malfunctioningly. At this time, the platform faced three urgent tasks. First, the engineers needed to immediately conduct a high-definition video consultation with the equipment manufacturer's experts via satellite link. Second, the control center needed to send adjustment commands to the associated pre-cooling valves to prevent secondary failures. Third, three key data files before and after the failure needed to be transmitted back to the onshore analysis center. At this time, changes in sea state caused the satellite link to gradually deteriorate from relatively stable (stage one) (stage two).

[0087] During the relatively stable phase of the satellite link, the real-time link status probe unit has detected a relatively stable link, with a latency of 700ms, a packet loss rate of 2%, and jitter of ±30ms. The current service types are video conferencing, adjustment commands, and transmission of critical data files. At this point, the deep packet inspection and semantic parsing unit determines through flow feature analysis that the encrypted video conference is an interactive, two-way audio and video stream, marking the service as {Service: Video Conferencing, Critical Level: P1, Feature: Low Latency Sensitive}. The deep parsing unit uses the Modbus protocol for the adjustment commands, identifying function code 0x06 (write to a single register), and the target address corresponding to the pre-cooling valve opening, thus marking the service as {Command: Control Command, Critical Level: P1, Feature: Minimal Packet, Absolutely Reliable}. For the data file 1 to be transmitted, the operation log.txt is parsed. Through analysis of the connection mode (persistent unidirectional large flow) and the initial data packet ASCII code in the operation log, data file 1 is identified as {File: Text Log, Preliminary Judgment: Highly Compressible}. For data file 2 that needs to be transmitted, analyze sensor snapshot.dat. The initial data block entropy value is >7.9 and it has a custom binary header. Mark data file 2 as {File: Unknown binary, Preliminary judgment: Suspected to be compressed, needs to be detected}. For data file 3 that needs to be transmitted, parse core memory dump.bin. No clear characteristics were found. Mark data file 3 as {File: Completely unknown, needs to be detected}.

[0088] The dynamic policy engine makes initial decisions. For P1 level video conferencing, it calls the low-latency interaction mode with the corresponding policy: {Schedule: AF41 (high priority), TCP: Optimized, Compression: Disabled, Jitter Buffer: 30ms}. For P1 control commands, it calls the extreme reliability mode with the corresponding policy: {Schedule: EF, Reliability: Strong FEC (15%) + Local Retransmission, Compression: Disabled}. For data file 1 in the file stream, it calls the high throughput mode with the corresponding policy: {Schedule: BE, Compression: Enable Efficient Compression (Zstd)}. For data file 2 in the file stream, it calls the energy-efficient mode with the corresponding policy: {Schedule: BE, Compression: Enable Detective Compression}. For data file 3 in the file stream, it calls the energy-efficient mode with the corresponding policy: {Schedule: BE, Compression: Enable Detective Compression}. After the above strategy was implemented, the optimized execution and initial results were as follows: smooth video playback, reliable command delivery within milliseconds; file 1 was stably compressed with a compression ratio of 8:1; file 2 was sampled and detected by the intelligent compression and processing engine: the compression rate of the 64KB sample was 99%, and the entropy value was extremely high. It was determined to be "incompressible" and immediately switched to pass-through mode to avoid CPU waste; file 3 was sampled and detected: the compression rate was 60%, and the entropy value was low, so it was determined to be "compressible" and compressed transmission was enabled.

[0089] The link status real-time probe unit detected a deterioration in the current link indicators, namely, a latency of 850ms, a packet loss rate of 8%, and jitter of ±100ms. The link quality of the current satellite and wireless private network has changed from Phase 1 to Phase 2, and the corresponding execution strategy should also be changed.

[0090] Specifically, the deep perception and recognition module dynamically adjusts its strategy to achieve dynamic balance. For example, if video conferencing experiences increased jitter, the anti-jitter buffer is automatically adjusted from 30ms to 80ms, and 10% FEC is enabled to smooth the video. If control commands experience increased packet loss, the FEC strength is automatically increased from 15% to 25%. Due to deteriorating link conditions leading to decreased throughput, the scheduler dynamically reduces the bandwidth weight of all BE queues while ensuring basic availability. In one instance, during the transmission of data file 1, the log content suddenly switched from plain text to a mixture of binary error codes and compressed core dump data blocks. The intelligent compression and processing engine's continuous monitoring window immediately detected a sharp deterioration in the compression ratio from 87% to 102% (i.e., data inflation), triggering a "smooth rollback" mechanism. Within milliseconds, the compression buffer is first cleared and the compressed data is refreshed, then the system seamlessly switches to pass-through mode and records that "the subsequent part of this stream is incompressible." During the transfer of data file 3, as the transfer progressed, the system detected a transition from seemingly random memory data to a large amount of compressible plaintext debugging information strings. The engine's continuous monitoring window showed that the compression ratio improved from the initial 60% to an even better 40%. The system maintains the compression mode and may choose to maintain or fine-tune the compression level based on CPU load to pursue higher efficiency.

[0091] After all tasks are completed, the learning and strategy optimization interface integrates the full-cycle data of this task, updates the performance archive, and adds new rules to the performance database. For example, for streams whose source IP is a log server and whose initial characteristic is text, it is necessary to be wary of the possibility of binary data being mixed in during the middle stage, and it is recommended to adopt a more proactive continuous monitoring strategy. Parameters in the performance database are optimized; for example, it is confirmed that enabling 80ms buffering and 10% FEC for Webex videos is the optimal combination when jitter is >80ms. Device profiling is also performed, such as recording the actual compressibility of core memory dump files, so that compression can be directly enabled for future similar file transfers.

[0092] Example 2: Based on the adaptive traffic optimization gateway method for satellite and wireless private networks provided in the foregoing embodiments, the present invention also provides an apparatus for implementing the above-mentioned method, namely, an adaptive traffic optimization gateway for satellite and wireless private networks. Figure 7 The diagram shown is a schematic representation of the device architecture according to an embodiment of the present invention. The device for an adaptive traffic optimization gateway for satellite and private wireless networks in this embodiment includes one or more processors 21 and a memory 22. Figure 7Take a processor 21 as an example.

[0093] Processor 21 and memory 22 can be connected via a bus or other means. Figure 7 Taking the example of a connection between China and Israel via a bus.

[0094] The memory 22 serves as a non-volatile computer-readable storage medium for an adaptive traffic optimization gateway method for satellite and wireless private networks. It can be used to store non-volatile software programs and non-volatile computer-executable programs, such as the adaptive traffic optimization gateway method for satellite and wireless private networks in the foregoing embodiments. The processor 21 executes various functional applications and data processing of the adaptive traffic optimization gateway device for satellite and wireless private networks by running the non-volatile software programs, instructions, and modules stored in the memory 22, thereby implementing the adaptive traffic optimization gateway method for satellite and wireless private networks as described in the foregoing embodiments.

[0095] Memory 22 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, memory 22 may include memory remotely located relative to processor 21, which can be connected to processor 21 via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0096] The program instructions / modules are stored in memory 22. When executed by one or more processors 21, they perform the adaptive traffic optimization gateway method for satellite and wireless private networks described in the foregoing embodiments, for example, executing the above-described... Figures 1-6 The steps shown.

[0097] This invention also provides a non-volatile computer storage medium storing computer-executable instructions that are executed by one or more processors, for example... Figure 7 One of the processors 21 can enable the aforementioned one or more processors to execute the adaptive traffic optimization gateway method for satellite and wireless private networks in the foregoing embodiments, for example, to perform the above-described... Figures 1-6 The steps shown.

[0098] It is worth noting that the information interaction and execution process between the modules and units in the above-mentioned device and system are based on the same concept as the processing method embodiment of the present invention. For details, please refer to the description in the method embodiment of the present invention, and will not be repeated here.

[0099] Those skilled in the art will understand that all or part of the steps in the various methods of the embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0100] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An adaptive traffic optimization gateway method for satellite and wireless private networks, characterized in that, include: Perform traffic analysis and semantic parsing on service traffic transmitted via satellite and wireless private network to identify at least one specific service type corresponding to the service traffic; Real-time detection of the current key performance indicators of the satellite and wireless private network, and aggregation of the key performance indicators into a link state vector representing the network status; Perform multi-dimensional analysis on each specific business type and generate a business profile for each specific business type; By integrating and analyzing business profiles and link status vectors, an executable target instruction set is generated for each specific business type, and the target instruction set is executed.

2. The adaptive traffic optimization gateway method for satellite and wireless private networks according to claim 1, characterized in that, The fusion analysis of service profiles and link state vectors generates executable target instruction sets for each specific service type, and executes the target instruction sets, including: Based on the business profile of each specific business type, multiple initial strategies are matched from the preset strategy library, where one initial strategy corresponds to one business profile. The link state vector is substituted into the benefit scoring function of each initial strategy for calculation to obtain the final target instruction set to be executed.

3. The adaptive traffic optimization gateway method for satellite and wireless private networks according to claim 2, characterized in that, The benefit scoring function is as follows: You=You Ci; Vi=α·BW_Savingi+δ·Priority_Bonusi; Ci=β·Latency_Penaltyi+γ·CPU_Costi+λ·Riski; Where Si is the overall score of initial strategy i; Vi is the expected benefit of initial strategy i; Ci is the expected cost of strategy i; BW_Savingi is the estimated bandwidth savings brought by initial strategy i; Priority_Bonusi is the business criticality benchmark value; α is the bandwidth economy coefficient; δ is the criticality reward coefficient; Latency_Penaltyi is the estimated additional latency introduced by initial strategy i; CPU_Costi is the estimated CPU resource consumption of initial strategy i; Riski is the adaptation risk estimate of initial strategy i under the current link state; β is the latency penalty coefficient; γ is the CPU cost coefficient; and λ is the risk penalty coefficient.

4. The adaptive traffic optimization gateway method for satellite and wireless private networks according to claim 3, characterized in that, The method further includes: For each initial strategy, the bandwidth savings, service criticality baseline, introduced additional latency, and consumed CPU resources are estimated, wherein the service criticality baseline is determined by the service level. Dynamic parameters are dynamically adjusted based on real-time link status. A comprehensive score for each initial strategy is calculated based on the bandwidth savings, the business criticality benchmark, the introduced additional latency, the consumed CPU resources, and the dynamic parameters. The initial strategy corresponding to the highest comprehensive score is taken as the target instruction set.

5. The adaptive traffic optimization gateway method for satellite and wireless private networks according to claim 1, characterized in that, The specific business type includes the transmission of data files, and the method further includes: The initial judgment result is obtained based on one or more of the protocol, content feature code or encrypted stream features of the data file. The initial judgment result includes clearly uncompressible, clearly compressible and tentatively compressed. By aggregating the file type of the data file and the initial judgment result, a business profile of the data file transmission is obtained.

6. The adaptive traffic optimization gateway method for satellite and wireless private networks according to claim 5, characterized in that, When the initial assessment result of the data file is a trial compression type, the method further includes: Based on one or more of the connection mode, packet statistical characteristics, TLS handshake information, or protocol context of the data file, an information entropy value is assigned to the data file. Perform compression attempts on data files that are intended for trial compression to obtain an initial compression ratio; Based on the combination of the information entropy and the initial compression rate, a final judgment result is obtained, wherein the final judgment result includes compressible, incompressible, or ambiguous.

7. The adaptive traffic optimization gateway method for satellite and wireless private networks according to claim 6, characterized in that, When the final judgment result is ambiguous, the method further includes: If the information entropy is less than or equal to the first entropy threshold, the initial compression rate is less than or equal to the first compression threshold, and the current gateway's CPU load rate is less than the preset high load threshold, then conservative compression is enabled for the data file and monitoring is performed. If the information entropy is greater than the second entropy threshold, or the initial compression ratio is greater than the second compression threshold, or the information entropy is greater than the first entropy threshold and less than or equal to the second entropy threshold, and the initial compression ratio is greater than the first compression threshold and less than or equal to the second compression threshold, or the current CPU load rate is greater than or equal to a preset high load threshold, then the data file will be directly passed through.

8. The adaptive traffic optimization gateway method for satellite and wireless private networks according to claim 1, characterized in that, The method further includes: After all optimization instruction sets have been executed, record multiple metadata related to the business and store these metadata in the performance archive. The performance archive is analyzed periodically to evaluate the overall benefits generated by the execution of each optimized instruction set. Optimized instruction sets that are verified to consistently generate the highest overall benefits are promoted to high-priority strategies and incorporated into the automatic matching and scheduling mechanism of subsequent tasks.

9. An adaptive traffic optimization gateway device for satellite and wireless private networks, characterized in that, include: At least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the processor for performing the adaptive traffic optimization gateway method for satellite and wireless private networks as described in any one of claims 1-8.

10. A non-volatile computer storage medium, characterized in that, The computer storage medium stores computer-executable instructions, which are executed by one or more processors to perform the adaptive traffic optimization gateway method for satellite and wireless private networks as described in any one of claims 1-8.

Images