A Method and System for Multi-Domain Data Processing in Power Wireless Local Area Networks Based on Trusted Authentication

Abstract

This invention provides a method and system for multi-domain data processing in power wireless local area networks (WLANs) based on trusted authentication. First, a trusted authentication model is invoked to authenticate the trustworthiness of terminal devices, generating a terminal authentication data set containing device identity information. Cross-domain raw data streams are acquired, and device operation features are extracted through multimodal feature extraction. Based on the terminal authentication data set, the trustworthiness of device operation status features is weighted and fused, and dynamically correlated with network transmission quality features to generate a cross-domain correlation feature set. A pre-trained multi-domain collaborative analysis model is used to aggregate the cross-domain correlation feature set and the spatiotemporal context of business operation semantic features, generating a multi-domain collaborative optimization strategy set containing resource scheduling priorities. Finally, based on this multi-domain collaborative optimization strategy set, dynamic resource reconfiguration is performed on business domain nodes, generating a network optimization execution instruction set that is fed back to the terminal device to trigger a trusted communication link upgrade operation, thus achieving effective multi-domain data processing and network optimization in power WLANs.

Description

Technical Field

[0001] This invention relates to the field of network security technology, and more specifically, to a method and system for multi-domain data processing in power wireless local area networks based on trusted authentication. Background Technology

[0002] In the field of power WLANs, with the diversification of services and the rapid increase in the number of connected devices, network security and efficient operation face severe challenges. Existing technologies often rely on single identity identifiers or static passwords for terminal device authentication, lacking dynamic, multi-dimensional trustworthiness assessment mechanisms, making it difficult to effectively resist access by devices with forged identities or abnormal behavior. Simultaneously, cross-domain data processing is often limited to simple data aggregation or rule matching, failing to deeply explore the inherent connections between device operating status, network transmission quality, and service operation semantics, leading to unreasonable resource scheduling and delayed anomaly responses. Furthermore, existing network optimization strategies are mostly based on preset rules or local analysis, lacking the ability to aggregate and collaboratively optimize spatiotemporal contexts from a global perspective, making it difficult to adapt to complex and ever-changing power business scenarios. Therefore, there is an urgent need for an innovative power WLAN data processing method that can comprehensively consider terminal device trustworthiness, cross-domain data characteristics, and service semantics to achieve multi-domain collaborative optimization and dynamic resource reconfiguration. Summary of the Invention

[0003] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a method for multi-domain data processing in a power wireless local area network based on trusted authentication, the method comprising:

[0004] A trusted authentication model is invoked to perform trustworthiness authentication on terminal devices in the power wireless local area network, generating a terminal authentication data set, which includes device identity characteristics, communication protocol compliance characteristics, and real-time behavior trustworthiness score.

[0005] The system acquires cross-domain raw data streams from multiple service domains in a power wireless local area network, performs multimodal feature extraction processing on the cross-domain raw data streams, and obtains device operating status features, network transmission quality features, and service operation semantic features.

[0006] Based on the terminal authentication data set, the device operation status characteristics are subjected to a credibility weighted fusion process to generate a trusted device operation feature set. The trusted device operation feature set is then dynamically correlated with the network transmission quality characteristics to generate a cross-domain correlation feature set.

[0007] A pre-trained multi-domain collaborative analysis model is invoked to perform spatiotemporal context aggregation processing on the cross-domain related feature set and the business operation semantic features to generate a multi-domain collaborative optimization strategy set, which includes resource scheduling priority sequence, abnormal operation interception rules and protocol adaptive adjustment parameters.

[0008] Based on the multi-domain collaborative optimization strategy set, the service domain nodes of the power wireless local area network are dynamically reconfigured to generate a network optimization execution instruction set, which is then fed back to the terminal device to trigger a trusted communication link upgrade operation.

[0009] In another aspect, embodiments of the present invention also provide a multi-domain data processing system for power wireless local area networks based on trusted authentication, including a processor and a machine-readable storage medium connected to the processor. The machine-readable storage medium is used to store programs, instructions, or code, and the processor is used to execute the programs, instructions, or code in the machine-readable storage medium to implement the above-described method.

[0010] Based on the above aspects, the embodiments of this application realize a global closed-loop optimization mechanism from terminal device trust verification to cross-domain data fusion, multi-domain collaborative optimization strategy generation, and network dynamic resource reconfiguration, which significantly improves the security and operating efficiency of power wireless local area networks in complex business scenarios. Specifically, by conducting a comprehensive trustworthiness assessment of terminal devices through a trusted authentication model, not only is the legitimacy of the access devices' identities and the compliance of their communication, but the trustworthiness of the devices' operating status is also dynamically reflected through a real-time behavioral trustworthiness scoring mechanism. Based on this, multimodal feature extraction is performed on the cross-domain raw data stream, effectively integrating multi-dimensional information such as device operating status, network transmission quality, and business operation semantics. Through trustworthiness weighted fusion and dynamic correlation analysis, the trustworthiness of terminal devices is closely linked with their operating status and network transmission quality, generating a set of cross-domain correlation features with high trustworthiness and business relevance. The pre-trained multi-domain collaborative analysis model, through spatiotemporal context aggregation processing, can deeply explore the potential connections between cross-domain correlation features and business operation semantics, generating a set of multi-domain collaborative optimization strategies that includes resource scheduling priorities, abnormal operation interception rules, and protocol adaptive adjustment parameters. This enables intelligent scheduling of power wireless LAN resources and proactive defense against abnormal behavior. Ultimately, based on this set of optimization strategies, the business domain nodes are dynamically reconfigured, and feedback is sent to the terminal devices to trigger trusted communication link upgrades, forming a complete optimization loop from strategy formulation to execution feedback. This not only improves the overall performance and reliability of the network, but also enhances the ability to respond to complex business scenarios and potential security threats. Attached Figure Description

[0011] Figure 1This is a schematic diagram of the execution flow of the multi-domain data processing method for power wireless local area networks based on trusted authentication provided in an embodiment of the present invention.

[0012] Figure 2 This is a schematic diagram of exemplary hardware and software components of a power wireless local area network multi-domain data processing system based on trusted authentication provided in an embodiment of the present invention. Detailed Implementation

[0013] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a method for multi-domain data processing in a power wireless local area network based on trusted authentication, according to an embodiment of the present invention. The following is a detailed description of this method for multi-domain data processing in a power wireless local area network based on trusted authentication.

[0014] Step S110: Call the trusted authentication model to perform trusted authentication processing on the terminal devices in the power wireless local area network, and generate a terminal authentication data set, which includes device identity characteristics, communication protocol compliance characteristics and real-time behavior trusted scores.

[0015] For example, in a specific power production scenario, a power plant uses a power wireless LAN to connect terminal devices distributed throughout the plant area. These terminal devices include various sensors, smart meters, control terminals, etc., used to monitor various parameters of power production and execute control commands in real time. When these terminal devices connect to the power wireless LAN, they need to undergo trust authentication to ensure the security and reliability of the network.

[0016] Step S111: Obtain the device identity certificate and protocol interaction log submitted by the terminal device when accessing the power wireless local area network, perform certificate chain verification processing on the device identity certificate, and obtain the device identity verification result and the trust score of the certificate issuing authority.

[0017] For example, when a terminal device initiates an access request, it submits its device identity certificate and protocol interaction logs to the authentication system. The device identity certificate, issued by an authoritative certificate authority, proves the device's legitimate identity and contains key information such as the device's unique identifier, certificate validity period, and certificate authority information. The protocol interaction logs record the protocol rules and interactions followed by the device during communication with the network.

[0018] Taking a temperature sensor as an example, its device identification certificate may record the sensor's serial number as "TS-00123", the certificate validity period from January 1, 2024 to January 1, 2025, and the certificate issuing authority as "PowerCertAuthority". The protocol interaction log may contain information such as the protocol version and encryption algorithm used when the sensor communicates with the network.

[0019] After receiving the above information, the authentication system performs certificate chain verification on the device's identity certificate. Certificate chain verification is a process that starts from the terminal certificate and verifies it level by level up to the root certificate. First, the authentication system checks whether the format of the device's identity certificate conforms to standard specifications, such as the certificate encoding method and the completeness of fields. Next, it verifies whether the certificate signature is valid, which requires using the certificate authority's public key to decrypt and verify the certificate signature. If the signature verification passes, the authentication system further checks the legitimacy of the certificate authority by querying a pre-configured trust list to determine whether the certificate authority is trusted.

[0020] For each certificate authority, a pre-set trust score range, such as from 0 to 100, can be established based on its historical performance and reputation. Assuming "PowerCert Authority" has a good historical record, its trust score is set to 90. During the verification process, if the entire certificate chain verification passes, the device authentication result is valid, and we then obtain information that the device authentication result is valid and the certificate authority's trust score is 90.

[0021] Step S112: Perform protocol compliance parsing on the protocol interaction log to extract the protocol version matching degree, encryption algorithm compliance identifier and session key update cycle.

[0022] After obtaining the verification result of the device identity certificate, a detailed protocol compliance analysis can be performed on the protocol interaction logs. The protocol interaction logs record a series of protocol interaction information between the device and the network during communication. By analyzing this information, key compliance indicators can be extracted.

[0023] Taking the protocol interaction log of a temperature sensor as an example, we first analyze the protocol version matching degree. Power line wireless LANs typically specify the standard protocol version used. Assuming the current network uses "IEEE 802.11ax", the authentication system extracts the actual protocol version used by the sensor from the protocol interaction log. If the sensor also uses "IEEE 802.11ax", the protocol version matching degree is 100%; if it uses the older version "IEEE 802.11n", the matching degree may be calculated as 30% based on the degree of difference from the standard version.

[0024] Next, the encryption algorithm compliance identifier is extracted. For example, to ensure communication security, the network may specify the use of a set encryption algorithm, such as "AES-256". The authentication system will check the encryption algorithm used by the sensor recorded in the protocol interaction log. If the specified "AES-256" is used, the encryption algorithm compliance identifier is "compliant"; if other unauthorized encryption algorithms are used, such as "DES", the encryption algorithm compliance identifier is "non-compliant".

[0025] Finally, the session key update cycle is extracted. To enhance communication security, the network requires devices to update session keys periodically, assuming a specified update cycle of once every 12 hours. The authentication system analyzes the actual session key update intervals of the sensors from the protocol interaction logs. If the sensor updates its session key every 10 hours, its session key update cycle meets the requirements; if it updates only once every 24 hours, it does not meet the requirements.

[0026] Step S113: Call the trusted authentication model to perform a trust fusion output on the device authentication result, the certificate authority trust score, the protocol version matching degree, the encryption algorithm compliance identifier and the session key update cycle, and generate the real-time behavior trust score.

[0027] After obtaining key information such as device authentication results, certificate authority trust score, protocol version matching degree, encryption algorithm compliance identifier and session key update cycle, a pre-trained trusted authentication model can be called to perform trust fusion output.

[0028] The trusted authentication model is trained on a large amount of historical data and can comprehensively evaluate the trustworthiness of the real-time behavior of terminal devices based on various input information. Taking a temperature sensor as an example, assuming the device authentication result is legitimate, the certificate authority trust score is 90 points, the protocol version matching degree is 100%, the encryption algorithm compliance mark is "compliant", and the session key update cycle meets the requirements.

[0029] The trusted authentication model weights this information, with different items potentially having different weights. For example, device authentication results and certificate authority trust scores may have higher weights, while protocol version matching, encryption algorithm compliance indicators, and session key update cycles may have relatively lower weights. Assuming the weight of the device authentication result is 0.3, the certificate authority trust score is 0.3, the protocol version matching is 0.1, the encryption algorithm compliance indicator is 0.15, and the session key update cycle is 0.15.

[0030] For a valid device authentication result, it can be quantified into a fixed score, such as 100 points; the protocol version matching degree can be directly used as its percentage value; the encryption algorithm compliance indicator is "compliant" and can be quantified into 100 points, and "non-compliant" into 0 points; the session key update cycle meets the requirements and is quantified into 100 points, and does not meet the requirements and is quantified into 0 points.

[0031] The calculation process for the real-time behavioral credibility score is as follows: (100×0.3+90×0.3+100×0.1+100×0.15+100×0.15)=97 points.

[0032] Step S114: Perform dynamic comparison processing based on the real-time behavior trust score and the preset trust threshold. If the real-time behavior trust score is greater than or equal to the trust threshold, generate a device identity legal identifier and a communication protocol compliance identifier, and associate and store the device identity legal identifier, the communication protocol compliance identifier and the real-time behavior trust score as the terminal authentication data set.

[0033] In this embodiment, a confidence threshold, such as 80 points, can be preset. When the real-time behavior confidence score of the temperature sensor is 97 points, the real-time behavior confidence score is compared with the confidence threshold. Since 97 points is greater than 80 points, it indicates that the confidence of the terminal device is high.

[0034] At this point, a device identity legitimacy identifier and a communication protocol compliance identifier can be generated. The device identity legitimacy identifier indicates that the device's identity has been verified as legitimate, and the communication protocol compliance identifier indicates that the protocol followed by the device during communication complies with network regulations.

[0035] Finally, the device's identity legitimacy identifier, communication protocol compliance identifier, and real-time behavior trust score can be associated and stored as a terminal authentication data set. The storage format can be a database record. For example, in a database table named "TerminalAuthenticationData", a new record can be created containing information such as the device's unique identifier "TS-00123", the device's identity legitimacy identifier "legitimate", the communication protocol compliance identifier "compliant", and the real-time behavior trust score "97 points", for subsequent data analysis and processing.

[0036] Step S120: Obtain cross-domain raw data streams from multiple service domains in the power wireless local area network, and perform multimodal feature extraction processing on the cross-domain raw data streams to obtain device operating status features, network transmission quality features, and service operation semantic features.

[0037] In power wireless local area networks, there are multiple different service domains, such as power monitoring service domain, power control service domain, and power dispatch service domain. These service domains generate a large amount of cross-domain raw data streams, including equipment operation logs, network transmission messages, and service operation instructions.

[0038] Taking temperature sensors in the power monitoring business domain and smart switches in the power control business domain as examples, temperature sensors record ambient temperature data in real time and generate equipment operation logs, while smart switches record their on / off status and operation information. During network transmission, network transmission packets containing this data are generated, and operators issue business operation commands to control the sampling frequency of temperature sensors or to control the on / off state of smart switches.

[0039] This allows us to collect these cross-domain raw data streams and perform multimodal feature extraction processing on them.

[0040] Step S121: Perform time-series parsing processing on the device operation log in the cross-domain raw data stream, extract the device power consumption fluctuation sequence, the average CPU load and the peak memory usage, and normalize and concatenate the device power consumption fluctuation sequence, the average CPU load and the peak memory usage to generate the device operation status features.

[0041] Taking the device operation log of a temperature sensor as an example, the device operation log is a log file that records various operating parameters of the device in chronological order. In this embodiment, the log file is processed by time sequence parsing.

[0042] First, the power consumption fluctuation sequence of the device is extracted. The power consumption of the temperature sensor changes under different operating conditions. In this embodiment, power consumption values ​​can be extracted from the device operation log at fixed time intervals (e.g., every minute) to form a power consumption value sequence. Assuming that the power consumption values ​​of the temperature sensor are 20mW, 22mW, 21mW, 23mW, 22mW, 24mW, 23mW, 25mW, 24mW, and 26mW over a consecutive 10 minutes, this constitutes a device power consumption fluctuation sequence.

[0043] Next, the average CPU load is extracted. If the temperature sensor has some computing power, its CPU will have different load conditions. For example, the CPU load value can be recorded every 5 minutes over a period of time (e.g., 1 hour), and then the average of these values ​​can be calculated. Assuming that 12 CPU load values ​​are recorded within 1 hour, at 10%, 12%, 11%, 13%, 12%, 14%, 13%, 15%, 14%, 16%, 15%, and 17%, the average CPU load is (10+12+11+13+12+14+13+15+14+16+15+17)÷12=13.5%.

[0044] Next, extract the peak memory usage. For example, you can monitor the temperature sensor's memory usage over a period of time (e.g., one day) and record the maximum memory usage. Assuming the temperature sensor's memory usage varies throughout the day, with a maximum value reaching 30%, then the peak memory usage is 30%.

[0045] Finally, the device power consumption fluctuation sequence, average CPU load, and peak memory usage are normalized and concatenated. Normalization is performed to unify data from different ranges onto a single scale, facilitating subsequent processing. For the device power consumption fluctuation sequence, assuming its original range is between 20mW and 26mW, it is normalized to a range of 0 to 1. The average CPU load (13.5%) and peak memory usage (30%) are also normalized accordingly. After normalization, these data are concatenated in a predetermined order to form the device operating status feature. For example, the normalized device power consumption fluctuation sequence is placed first, followed by the normalized average CPU load, and finally the normalized peak memory usage, thus generating a device operating status feature containing multi-dimensional data.

[0046] Step S122: Perform protocol parsing processing on the network transmission packets in the cross-domain raw data stream, extract the transmission delay distribution, bandwidth utilization and packet loss rate trend, and perform weighted calculation processing on the transmission delay distribution, bandwidth utilization and packet loss rate trend based on the preset network quality assessment rules to generate the network transmission quality features.

[0047] In this embodiment, the network transmission message is a data unit transmitted in the power wireless local area network, which includes information such as source address, destination address, data content, and transmission time.

[0048] Taking network transmission messages between a temperature sensor and a smart switch as an example, the transmission delay distribution is first extracted. For instance, the transmission time of each message from the sender (temperature sensor) to the receiver (smart switch) can be recorded. Then, the number of messages with different transmission delays within a certain period (e.g., 1 hour) is counted to form the transmission delay distribution. Assuming that within 1 hour, there are 100 messages with transmission delays between 10ms and 20ms, 200 messages between 20ms and 30ms, 150 messages between 30ms and 40ms, and so on, this constitutes the transmission delay distribution.

[0049] Next, extract the bandwidth utilization rate. For example, you can calculate the ratio of the actual bandwidth used by the network to the total network bandwidth over a period of time (e.g., 1 hour). Assuming the total network bandwidth is 100Mbps, and the average actual bandwidth used in 1 hour is 30Mbps, then the bandwidth utilization rate is 30%.

[0050] Next, extract the packet loss rate trend. For example, you can record packet loss data at different times within a period of time (e.g., 1 day), calculate the packet loss rate for each time period, and observe its trend. For example, the packet loss rate might be 2% from 9:00 AM to 10:00 AM, 3% from 10:00 AM to 11:00 AM, and 2.5% from 11:00 AM to 12:00 PM, thus forming a packet loss rate trend.

[0051] Based on preset network quality assessment rules, the transmission delay distribution, bandwidth utilization, and packet loss rate trend are weighted and calculated. The preset network quality assessment rules assign different weights to each indicator; for example, the weight of transmission delay distribution is 0.4, the weight of bandwidth utilization is 0.3, and the weight of packet loss rate trend is 0.3.

[0052] For transmission delay distribution, scores can be quantified based on the importance of different delay intervals. For example, packets with transmission delays between 10ms and 20ms are given higher scores, while packets with delays exceeding 30ms are given lower scores. A weighted average is then calculated. Similarly, bandwidth utilization and packet loss rate trends are quantified based on their impact on network quality. Finally, these quantified scores are weighted and calculated to generate network transmission quality characteristics. For instance, if the quantified score for transmission delay distribution is 80, the quantified score for bandwidth utilization is 70, and the quantified score for packet loss rate trend is 75, then the calculated network transmission quality characteristic score is (80 × 0.4 + 70 × 0.3 + 75 × 0.3) = 76.5.

[0053] Step S123: Perform semantic segmentation on the business operation instructions in the cross-domain raw data stream to obtain a set of operation instruction text fragments. Call a pre-trained semantic analysis model to perform intent recognition processing on the set of operation instruction text fragments to generate a business operation intent vector and a semantic compliance score. Then, perform association encoding processing on the business operation intent vector and the semantic compliance score to generate the business operation semantic features.

[0054] In this embodiment, the operational instructions are typically instructions issued by operators to control equipment or perform set tasks, and they exist in text form.

[0055] Taking the operator's business operation command to the temperature sensor, "Adjust the sampling frequency of the temperature sensor to once per minute," as an example, semantic segmentation is performed first. Semantic segmentation involves dividing the entire operation command text into semantic units, resulting in a set of operation command text fragments. For the above command, the segmented operation command text fragment set might include "will," "temperature sensor," "sampling frequency," "adjust to," and "once per minute."

[0056] Next, a pre-trained semantic analysis model is invoked to perform intent recognition processing on the set of operation instruction text fragments. This pre-trained semantic analysis model is trained on a large amount of business operation instruction data and is capable of understanding the semantics of the text fragments and the operator's intent. Through this semantic analysis model, each operation instruction text fragment is analyzed to identify the operator's intent. For example, the intent of the above instruction might be to adjust the sampling frequency of the temperature sensor.

[0057] Semantic analysis models can transform identified intents into business operation intent vectors. A business operation intent vector is a multi-dimensional vector, with each dimension representing a different intent feature. Assuming a business operation intent vector has five dimensions, representing intent features such as adjusting device parameters, controlling device on / off, querying device status, starting a task, and stopping a task, for example, the instruction to adjust the temperature sensor sampling frequency has a value of 1 in the "adjust device parameters" dimension and 0 in the other dimensions, forming a business operation intent vector [1, 0, 0, 0, 0].

[0058] Simultaneously, the semantic analysis model can assess the semantic compliance of operational instructions and generate a semantic compliance score. Semantic compliance refers to whether operational instructions conform to the business rules and security requirements of the power wireless local area network. For example, if an operational instruction requires adjusting the sampling frequency of a temperature sensor to a value outside its normal operating range, then the semantic compliance of this instruction is low. Assume that, based on the assessment, the semantic compliance score of this instruction is 90 points.

[0059] Finally, the business operation intent vector and the semantic compliance score are correlated and encoded. Correlation encoding combines the business operation intent vector and the semantic compliance score according to predefined rules to generate business operation semantic features. For example, the semantic compliance score can be added as an additional dimension to the business operation intent vector, forming a new vector [1, 0, 0, 0, 0, 90], thus constituting the business operation semantic features.

[0060] Step S130: Based on the terminal authentication data set, perform credibility-weighted fusion processing on the device operation status features to generate a set of trusted device operation features, and perform dynamic correlation analysis processing on the set of trusted device operation features and the network transmission quality features to generate a cross-domain correlation feature set.

[0061] After obtaining the terminal authentication dataset and device operating status characteristics, a trustworthiness-weighted fusion process can be performed. The real-time behavioral trustworthiness score in the terminal authentication dataset reflects the trustworthiness of the terminal device and is used to weight the device operating status characteristics.

[0062] Step S131: Extract real-time behavior credibility score from the terminal authentication data set, and determine the device operation feature weighting coefficient according to the mapping relationship between the real-time behavior credibility score and the preset score interval.

[0063] Taking a temperature sensor as an example, its real-time behavior credibility score can be extracted from the terminal authentication data set, assuming it to be 97 points. This embodiment predefines the mapping relationship between the scoring range and the weighting coefficient of the device's operating characteristics. For example, the weighting coefficient range corresponding to a scoring range of 80-100 points is 0.8-1.0. The scoring range is divided into 20 small intervals, each interval corresponding to an increment of the weighting coefficient. A score of 97 points falls within the scoring range of 80-100 points. Through linear mapping calculation, (97-80)÷(100-80)×(1.0-0.8)+0.8=0.97, meaning the weighting coefficient of the device's operating characteristics is 0.97. This weighting coefficient will be used for subsequent weighting processing of the device's operating status characteristics to reflect the degree to which the credibility of the terminal device affects its operating status characteristics.

[0064] Step S132: Perform sliding window mean calculation processing on the device power consumption fluctuation sequence, CPU load average and memory usage peak in the device operation status characteristics to obtain the power consumption fluctuation average sequence, CPU load average sequence and memory usage average sequence.

[0065] For example, regarding the device operating status characteristics of a temperature sensor, the device power consumption fluctuation sequence, average CPU load, and peak memory usage are important components. Let's first look at the device power consumption fluctuation sequence, assuming the previously extracted sequence represents power consumption values ​​over a continuous 10 minutes: 20mW, 22mW, 21mW, 23mW, 22mW, 24mW, 23mW, 25mW, 24mW, 26mW. A sliding window averaging calculation is used, with the sliding window size set to 3 minutes.

[0066] The first window contains the first three data points (20mW, 22mW, 21mW), with a mean of (20+22+21)÷3=21mW; the second window contains the second to fourth data points (22mW, 21mW, 23mW), with a mean of (22+21+23)÷3=22mW; and so on, ultimately yielding a power consumption fluctuation mean sequence.

[0067] For the average CPU load, assume that the CPU load values ​​recorded every 5 minutes within one hour were previously calculated to be 10%, 12%, 11%, 13%, 12%, 14%, 13%, 15%, 14%, 16%, 15%, and 17%. Using a sliding window of size 3, the average value of the first window (10%, 12%, 11%) is (10+12+11)÷3=11%. Subsequent windows are calculated sequentially to obtain the average CPU load sequence.

[0068] For peak memory usage, assuming that the peak memory usage data is monitored at different times of the day, and divided into a set time interval (such as 1 hour), the average memory usage sequence can also be obtained by using a sliding window to calculate the average.

[0069] Step S133: Based on the weighting coefficient of the device operating characteristics, perform dynamic weighted fusion processing on the average power consumption fluctuation, the average CPU load sequence, and the average memory usage sequence to generate weighted device power consumption characteristics, weighted CPU load characteristics, and weighted memory usage characteristics.

[0070] The weighting factor for the device operating characteristics has already been determined to be 0.97. For the power consumption fluctuation mean sequence, each mean in the sequence is multiplied by this weighting factor. For example, the first mean in the power consumption fluctuation mean sequence is 21mW, which, after weighting, becomes 21 × 0.97 = 20.37mW. By performing this processing on the entire power consumption fluctuation mean sequence, the weighted device power consumption characteristics are obtained.

[0071] For the CPU load mean sequence, each mean in the sequence is also multiplied by 0.97. For example, if the first CPU load mean is 11%, the weighted mean is 11 × 0.97 = 10.67%. After processing the entire sequence, the weighted CPU load feature is obtained.

[0072] For the memory usage mean sequence, the same method is used to multiply each mean by 0.97 to obtain the weighted memory usage feature. Through this dynamic weighted fusion process, the device operating status features are correlated with the trustworthiness of the terminal device. Devices with higher trustworthiness have higher weights for their operating status features in subsequent processing.

[0073] Step S134: Perform time-series alignment and splicing processing on the weighted device power consumption characteristics, the weighted CPU load characteristics, and the weighted memory usage characteristics to generate the trusted device operation characteristic set.

[0074] After obtaining the weighted device power consumption characteristics, weighted CPU load characteristics, and weighted memory usage characteristics, time-series alignment and splicing processing is required. This is because these characteristics are calculated based on time-series data, ensuring they correspond in time.

[0075] Assume there are 10 time points for weighted device power consumption, weighted CPU load, and weighted memory usage characteristics. The weighted device power consumption, weighted CPU load, and weighted memory usage values ​​from the first time point are concatenated sequentially to form a new multidimensional data point. The same concatenation operation is then performed on the data from the second time point, and so on. Finally, these 10 concatenated data points are combined to generate a trusted device operating characteristic set. This trusted device operating characteristic set integrates the device's power consumption, CPU load, and memory usage, and takes into account the trustworthiness of the terminal device.

[0076] Step S135: Perform timestamp alignment processing on the weighted device power consumption characteristics, weighted CPU load characteristics, and weighted memory usage characteristics in the trusted device operation characteristic set to obtain a time-synchronized device operation characteristic sequence.

[0077] After obtaining the set of trusted device operating characteristics, since the weighted device power consumption characteristics, weighted CPU load characteristics, and weighted memory usage characteristics may be collected or calculated at different time scales, timestamp alignment processing is required.

[0078] Taking a temperature sensor as an example, the weighted device power consumption characteristic might be data collected once per minute, the weighted CPU load characteristic might be data collected once every 5 minutes, and the weighted memory usage characteristic might be data collected once every 10 minutes. First, a unified time scale is determined, such as a 1-minute interval. For the weighted CPU load characteristic, the data from each 5-minute interval is evenly distributed across these 5 1-minute time points; for the weighted memory usage characteristic, the data from each 10-minute interval is evenly distributed across these 10 1-minute time points. This ensures that the weighted device power consumption characteristic, weighted CPU load characteristic, and weighted memory usage characteristic are synchronized in time, forming a time-synchronized device operation characteristic sequence. Each time point in this time-synchronized device operation characteristic sequence contains comprehensive information on device power consumption, CPU load, and memory usage.

[0079] Step S136: Perform sliding window statistical processing on the transmission delay distribution, bandwidth utilization and packet loss rate trends in the network transmission quality characteristics to generate the mean transmission delay sequence, the mean bandwidth utilization sequence and the mean packet loss rate sequence.

[0080] In this embodiment, taking transmission delay distribution as an example, it is assumed that the previously obtained data on the distribution of the number of packets in different transmission delay intervals within one hour is divided into minutes. The sliding window size is set to 5 minutes.

[0081] For the first 5-minute window, the number of packets in each transmission delay interval within those 5 minutes is summed, and then divided by 5 to obtain the average number of packets in each transmission delay interval within those 5 minutes, thus forming the average transmission delay for that window. Then, the window is moved forward by 1 minute, and the same calculation is performed on the new 5-minute data, and so on, until the final sequence of average transmission delays is obtained.

[0082] For bandwidth utilization, assuming the bandwidth utilization data obtained previously for every 5 minutes within 1 hour, we also use a sliding window of size 3 to calculate the average bandwidth utilization within each window, thus obtaining a sequence of average bandwidth utilization.

[0083] To analyze packet loss rate trends, assuming we have previously recorded hourly packet loss rate data for a given day, we employ a sliding window approach. For example, with a window size of 6 hours, we calculate the average packet loss rate within each window, resulting in a sequence of average packet loss rates. This sliding window statistical processing smooths out fluctuations in the data, providing a clearer reflection of trends in network transmission quality.

[0084] Step S137: Call the pre-trained correlation analysis model to perform multivariate correlation calculation on the time synchronization device running feature sequence, the transmission delay mean sequence, the bandwidth utilization mean sequence and the packet loss rate mean sequence, and generate the correlation degree between device power consumption and transmission delay, the correlation degree between CPU load and bandwidth utilization, and the correlation degree between memory usage and packet loss rate.

[0085] In this embodiment, the pre-trained correlation analysis model is trained on a large amount of historical data and can analyze the correlation between different variables. For calculating the correlation between device power consumption and transmission delay, the relationship between the weighted device power consumption characteristics and the mean transmission delay sequence in the time-synchronous device operation characteristic sequence can be analyzed. By comparing the changes in device power consumption and transmission delay at different time points, the degree of correlation between them is calculated. For example, if at certain time points, an increase in device power consumption is accompanied by an increase in transmission delay, it indicates a positive correlation; conversely, if an increase in device power consumption is accompanied by a decrease in transmission delay, it indicates a negative correlation. The model will calculate a specific correlation degree value based on this data; assuming the calculated correlation degree between device power consumption and transmission delay is 0.7.

[0086] To calculate the correlation between CPU load and bandwidth utilization, we can analyze the correlation between the weighted CPU load characteristics and the mean bandwidth utilization sequence in the operating characteristic sequence of the timing synchronization device. Similarly, we calculate based on the changes in both at different time points, assuming a correlation of 0.6 between CPU load and bandwidth utilization.

[0087] To calculate the correlation between memory usage and packet loss rate, we can analyze the relationship between the weighted memory usage features and the mean packet loss rate sequence in the operating feature sequence of the time synchronization device. Let's assume that the calculated correlation between memory usage and packet loss rate is 0.5.

[0088] Therefore, the above correlation values ​​reflect the relationship between device operating status and network transmission quality.

[0089] Step S138: Perform multi-dimensional vectorization processing on the correlation between device power consumption and transmission delay, the correlation between CPU load and bandwidth utilization, and the correlation between memory usage and packet loss rate to generate the cross-domain correlation feature set.

[0090] In this embodiment, the correlation between device power consumption and transmission delay, the correlation between CPU load and bandwidth utilization, and the correlation between memory usage and packet loss rate can be combined into a vector in a set order. For example, the correlation between device power consumption and transmission delay (0.7) can be placed in the first dimension, the correlation between CPU load and bandwidth utilization (0.6) in the second dimension, and the correlation between memory usage and packet loss rate (0.5) in the third dimension, forming a three-dimensional vector [0.7, 0.6, 0.5]. This three-dimensional vector is the cross-domain correlation feature set, which integrates the correlation information between device operating status and network transmission quality, and will be used for subsequent multi-domain collaborative analysis.

[0091] Step S140: Call the pre-trained multi-domain collaborative analysis model to perform spatiotemporal context aggregation processing on the cross-domain related feature set and the business operation semantic features to generate a multi-domain collaborative optimization strategy set, which includes resource scheduling priority sequence, abnormal operation interception rules and protocol adaptive adjustment parameters.

[0092] In this embodiment, the pre-trained multi-domain collaborative analysis model can comprehensively consider information from multiple aspects such as device operation, network transmission, and service operation, and mine the spatiotemporal context relationships therein, thereby generating optimization strategies.

[0093] Step S141: Standardize the correlation between device power consumption and transmission delay, CPU load and bandwidth utilization, and memory usage and packet loss rate in the cross-domain correlation feature set to obtain a standardized correlation feature vector.

[0094] The correlations between device power consumption and transmission delay, CPU load and bandwidth utilization, and memory usage and packet loss rate in the cross-domain correlation feature set are standardized. The purpose of standardization is to unify data from different ranges to a common scale, facilitating model processing.

[0095] Assume the correlation coefficient between device power consumption and transmission delay ranges from 0 to 1, the correlation coefficient between CPU load and bandwidth utilization ranges from 0 to 0.8, and the correlation coefficient between memory usage and packet loss rate ranges from 0 to 0.6. Using the min-max normalization method, for a correlation coefficient of 0.7 between device power consumption and transmission delay, the normalized calculation is (0.7-0)÷(1-0)=0.7; for a correlation coefficient of 0.6 between CPU load and bandwidth utilization, the normalized calculation is (0.6-0)÷(0.8-0)=0.75; and for a correlation coefficient of 0.5 between memory usage and packet loss rate, the normalized calculation is (0.5-0)÷(0.6-0)≈0.83. These three normalized values ​​are combined into a vector [0.7, 0.75, 0.83], thus forming the normalized correlation feature vector.

[0096] Step S142: Perform feature dimensionality reduction processing on the business operation intent vector and semantic compliance score in the business operation semantic features to obtain a low-dimensional semantic feature vector.

[0097] In this embodiment, the business operation intent vector may be a high-dimensional vector containing multiple intent features. In order to reduce the dimensionality of the data and improve the processing efficiency of the model, dimensionality reduction processing is required.

[0098] Assume the business operation intent vector is [1, 0, 0, 0, 0, 90], where the first five dimensions represent different intent features, and the sixth dimension is the semantic compliance score. Principal Component Analysis (PCA) is used for dimensionality reduction. First, the covariance matrix of the business operation intent vector is calculated. Then, the eigenvalues ​​and eigenvectors of the covariance matrix are solved, and the eigenvectors with larger eigenvalues ​​are selected to form a projection matrix. Assuming that the first two eigenvectors are selected to form the projection matrix, the business operation intent vector is projected into this low-dimensional space, resulting in a two-dimensional low-dimensional semantic feature vector, assumed to be [0.8, 0.2].

[0099] Step S143: Perform spatiotemporal location encoding on the standardized associated feature vector and the low-dimensional semantic feature vector to generate a spatiotemporal context feature matrix.

[0100] In this embodiment, the spatiotemporal location encoding process is used to associate and integrate different types of features in the spatiotemporal dimension.

[0101] Assume the standardized correlation feature vector is [0.7, 0.75, 0.83] and the low-dimensional semantic feature vector is [0.8, 0.2]. First, assign a timestamp to each feature vector, representing its position in the time series. Then, encode them based on the timestamp and the dimensionality information of the feature vector. For example, arrange the standardized correlation feature vector and the low-dimensional semantic feature vector in chronological order, and add a time-encoded information to the beginning of each vector. Assume the time-encoded information uses a single number to represent the time point, with the first time point being 1 and the second time point being 2. Encode the standardized correlation feature vector [0.7, 0.75, 0.83] as [1, 0.7, 0.75, 0.83], and encode the low-dimensional semantic feature vector [0.8, 0.2] as [2, 0.8, 0.2]. Combine these two encoded vectors to form a matrix [[1, 0.7, 0.75, 0.83], [2, 0.8, 0.2]], which is the spatiotemporal context feature matrix.

[0102] Step S144: Call the multi-domain collaborative analysis model to perform multi-head attention aggregation processing on the spatiotemporal context feature matrix to generate resource scheduling priority weights, abnormal operation detection thresholds, and protocol adjustment sensitivity parameters.

[0103] In this embodiment, the multi-head attention mechanism enables the multi-domain collaborative analysis model to focus on different information in different representation subspaces, thereby capturing the relationships between features more comprehensively.

[0104] When processing the spatiotemporal context feature matrix, the multi-domain collaborative analysis model takes each vector in the matrix as input and performs a weighted summation of these vectors through multiple attention heads to obtain different attention representations. For example, with three attention heads, each attention head calculates the spatiotemporal context feature matrix based on its own weight matrix, resulting in three different attention representations. These three attention representations are then concatenated and linearly transformed to obtain the final aggregated representation.

[0105] Based on this aggregation representation, the multi-domain collaborative analysis model generates resource scheduling priority weights, anomaly detection thresholds, and protocol adjustment sensitivity parameters. Resource scheduling priority weights determine the priority order of different service domains or devices during resource allocation. For example, a calculated resource scheduling priority weight of [0.6, 0.3, 0.1] indicates that the resource scheduling priorities of service domains A, B, and C decrease sequentially. The anomaly detection threshold determines whether a service operation is abnormal. Assuming a generated anomaly detection threshold of 0.8, if a certain characteristic value of a service operation exceeds this threshold, the operation is considered potentially abnormal. The protocol adjustment sensitivity parameter guides the adaptive adjustment of the communication protocol. Assuming a generated protocol adjustment sensitivity parameter of 0.5, it indicates a moderate level of sensitivity to protocol adjustments.

[0106] Step S145: Generate the resource scheduling priority sequence according to the resource scheduling priority weight, generate the abnormal operation interception rule based on the abnormal operation detection threshold, generate the protocol adaptive adjustment parameter according to the protocol adjustment sensitivity parameter, and store the resource scheduling priority sequence, abnormal operation interception rule and protocol adaptive adjustment parameter together as the multi-domain collaborative optimization strategy set.

[0107] Assume the resource scheduling priority weights are [0.6, 0.3, 0.1], corresponding to business domains A, B, and C. Sort the weights from largest to smallest, resulting in a priority order of business domain A > business domain B > business domain C. This is the resource scheduling priority sequence.

[0108] Anomaly detection thresholds are used to generate anomaly interception rules. Assuming the anomaly detection threshold is 0.8, the anomaly interception rule can be set as follows: when a key characteristic value of a business operation exceeds 0.8, the operation is immediately intercepted, and relevant operation information is recorded.

[0109] The protocol adaptive adjustment parameters are generated based on the protocol sensitivity parameter. Assuming the protocol sensitivity parameter is 0.5, the adaptive adjustment parameters may include a heartbeat interval adjustment value, an encryption algorithm switching flag, and a session key rotation period. For example, based on the sensitivity parameter, the heartbeat interval adjustment value might be increased by 50%, the encryption algorithm switching flag might indicate a switch to a more advanced encryption algorithm, and the session key rotation period might be shortened to 80% of its original value.

[0110] Finally, the resource scheduling priority sequence, abnormal operation interception rules, and protocol adaptive adjustment parameters are associated and stored as a multi-domain collaborative optimization strategy set. This information can be stored in a database table, with each record containing fields such as resource scheduling priority sequence, abnormal operation interception rules, and protocol adaptive adjustment parameters, facilitating subsequent queries and use.

[0111] Step S150: Based on the multi-domain collaborative optimization strategy set, perform dynamic resource reconfiguration on the service domain nodes of the power wireless local area network, generate a network optimization execution instruction set, and feed the network optimization execution instruction set back to the terminal device to trigger a trusted communication link upgrade operation.

[0112] After obtaining the set of multi-domain collaborative optimization strategies, it is necessary to perform dynamic resource reconfiguration on the service domain nodes of the power wireless local area network in order to optimize the network and upgrade the trusted communication link.

[0113] Step S151: Extract the resource scheduling priority sequence from the multi-domain collaborative optimization strategy set, sort the computing resource pools of the business domain nodes according to the resource scheduling priority sequence, and generate a resource allocation queue.

[0114] In this embodiment, it is assumed that the resource scheduling priority sequence is business domain A > business domain B > business domain C. The computing resource pool of a business domain node includes computing resources such as CPU computing power and memory space.

[0115] First, calculate the total amount of various resources in the computing resource pool. Assume the computing resource pool has 1000 CPU computing units and 500GB of memory. Based on the resource scheduling priority sequence, allocate more computing resources to business domain A, less to business domain B, and less to business domain C.

[0116] A proportional allocation method can be used, with the allocation ratio determined based on priority. Assume that business domain A is allocated 60% of the resources, business domain B is allocated 30%, and business domain C is allocated 10%. Then, business domain A will receive 1000 × 60% = 600 computing units of CPU power and 500 × 60% = 300GB of memory; business domain B will receive 1000 × 30% = 300 computing units of CPU power and 500 × 30% = 150GB of memory; and business domain C will receive 1000 × 10% = 100 computing units of CPU power and 500 × 10% = 50GB of memory.

[0117] Each business domain and its allocated resource information are arranged in priority order to generate a resource allocation queue. The queue format is as follows: Business Domain A (600 CPU computing units, 300GB memory), Business Domain B (300 CPU computing units, 150GB memory), and Business Domain C (100 CPU computing units, 50GB memory). This resource allocation queue will serve as the basis for subsequent resource allocation operations.

[0118] Step S152: Perform pattern matching processing on the real-time operation instruction stream of the business domain node based on the abnormal operation interception rule. If an operation instruction that matches the abnormal operation interception rule is detected, an operation interception signal is generated and an alarm log is triggered.

[0119] During operation, business domain nodes continuously generate real-time operation command streams, which contain various business operation information. Based on the previously generated abnormal operation interception rules, an operation is judged as abnormal when a key characteristic value exceeds 0.8.

[0120] In this embodiment, real-time monitoring and pattern matching processing can be performed on the real-time operation instruction stream. For example, operation instructions may contain some quantified feature values, such as the complexity score of the operation and the risk assessment value of the operation. Taking the complexity score of the operation as a key feature value as an example, the complexity score of each operation instruction can be extracted in real time and compared with the abnormal operation detection threshold of 0.8.

[0121] Suppose that at a certain moment, a business domain node receives an operation instruction with a calculated complexity score of 0.9. Since 0.9 is greater than 0.8, this indicates that the operation instruction matches the abnormal operation interception rules. At this point, an operation interception signal can be generated immediately to prevent the execution of this abnormal operation. Simultaneously, an alarm log can be triggered, recording detailed information about the abnormal operation, including the content of the operation instruction, the time the operation was initiated, and the operation's complexity score. The alarm log can be stored in a dedicated log file for convenient subsequent auditing and analysis.

[0122] Step S153: Based on the protocol adaptive adjustment parameters, dynamically adjust the communication protocol stack of the service domain node to generate a protocol update instruction set, which includes the heartbeat interval adjustment value, encryption algorithm switching identifier, and session key rotation cycle.

[0123] Based on the previously generated protocol adaptive adjustment parameters, the communication protocol stack of the service domain nodes is dynamically adjusted. These adaptive adjustment parameters include the heartbeat interval adjustment value, encryption algorithm switching flag, and session key rotation cycle.

[0124] Assuming the heartbeat interval adjustment value increases by 50%, the encryption algorithm switching flag indicates a switch to a more advanced encryption algorithm, and the session key rotation cycle is shortened to 80% of its original value.

[0125] Regarding the heartbeat interval, assuming the original interval was 10 seconds, increasing it by 50% results in a new heartbeat interval of 10 × (1 + 50%) = 15 seconds. For the encryption algorithm, the currently used encryption algorithm in the communication protocol stack can be switched to a more advanced algorithm based on the encryption algorithm switching flag, such as switching from AES-128 to AES-256. Regarding the session key rotation period, assuming the original rotation period was 12 hours, shortening it to 80% results in a new session key rotation period of 12 × 80% = 9.6 hours.

[0126] These adjusted parameters are combined into a protocol update instruction set, which takes the following form: the heartbeat interval is adjusted to 15 seconds, the encryption algorithm is switched to AES-256, and the session key rotation cycle is adjusted to 9.6 hours. This protocol update instruction set will be used to update the communication protocol stack of the business domain node.

[0127] Step S154: Perform instruction encoding processing on the resource allocation queue, the operation interception signal, and the protocol update instruction set to generate the network optimization execution instruction set.

[0128] To facilitate transmission and execution within the network, resource allocation queues, operation interception signals, and protocol update instruction sets need to be encoded. Instruction encoding converts this information into data of a defined format so that network devices can recognize and execute it.

[0129] For the resource allocation queue, each business domain and its allocated resource information are encoded according to a set encoding rule. For example, binary encoding is used to convert the business domain's identifier, CPU computing power, and memory space values ​​into binary strings. Assume that the identifier of business domain A is encoded as 001, the CPU computing power of 600 computing units is encoded as 1001011000, and the memory space of 300GB is encoded as 100101100. These codes are combined to form the encoded information of business domain A in the resource allocation queue. The same encoding process is performed on business domains B and C, and then they are concatenated according to priority.

[0130] For operation interception signals, they are encoded into a specified binary identifier, such as 111 representing an operation interception signal. For protocol update instruction sets, the heartbeat interval adjustment value, encryption algorithm switching identifier, and session key rotation period are encoded according to the encoding rules. For example, a heartbeat interval of 15 seconds is encoded as 00001111, an encryption algorithm switch to AES-256 is encoded as 0101, and a session key rotation period of 9.6 hours is converted to 576 minutes and encoded as 1001000000.

[0131] Finally, the encoded resource allocation queue, operation interception signals, and protocol update instruction set are combined in a predetermined order to generate a network optimization execution instruction set. This instruction set contains all the instruction information used for network optimization and will be sent to the terminal device for execution.

[0132] Step S155: Perform instruction parsing processing on the resource allocation queue in the network optimization execution instruction set to generate resource scheduling configuration parameters, and send the resource scheduling configuration parameters to the resource manager of the terminal device to trigger the computing resource reallocation operation.

[0133] In this embodiment, the resource allocation queue in the network optimization execution instruction set is processed by instruction parsing. Instruction parsing involves restoring the encoded resource allocation queue into readable resource allocation information.

[0134] Based on the preceding encoding rules, the encoded information of the resource allocation queue is decoded. For example, the identifier of the business domain, CPU computing power, and memory space values ​​are extracted from the binary encoding. The decoded information is then converted into resource scheduling configuration parameters, in the following format: Business domain A is allocated 600 CPU computing units and 300GB of memory; Business domain B is allocated 300 CPU computing units and 150GB of memory; Business domain C is allocated 100 CPU computing units and 50GB of memory.

[0135] These resource scheduling configuration parameters are sent to the resource manager of the terminal device. The resource manager is the module in the terminal device responsible for managing the allocation of computing resources. After receiving the resource scheduling configuration parameters, it will perform a reallocation operation on computing resources according to these parameters. For example, the resource manager will adjust the runtime slices of each business domain on the CPU, allocating more CPU time to business domain A and 300GB of memory space to it, and so on, to complete the reallocation of computing resources.

[0136] Step S156: Perform signal encoding processing on the operation interception signal to generate an operation interception command, and send the operation interception command to the command filter of the terminal device to enable the real-time operation interception service.

[0137] In this embodiment, the operation interception signal is encoded to make it recognizable by the instruction filter of the terminal device. The operation interception signal is encoded as a binary identifier 111, and this encoding is further encapsulated into an operation interception instruction. The operation interception instruction may include information such as the encoding of the operation interception signal, the source identifier of the instruction, and the validity period of the instruction.

[0138] The operation interception command is sent to the command filter on the terminal device. The command filter is a module in the terminal device used to filter operation commands. Upon receiving the operation interception command, it activates a real-time operation interception service. When a new operation command enters the command filter, the filter checks the command according to the operation interception rules. If it is determined to be an abnormal operation, it will immediately intercept the command and prevent its execution. For example, when an operation command with a complexity score exceeding 0.8 is received, the command filter will reject the command, ensuring the operational security of the terminal device.

[0139] Step S157: Perform protocol encapsulation processing on the heartbeat interval adjustment value, the encryption algorithm switching identifier, and the session key rotation cycle in the protocol update instruction set to generate a protocol update data packet, and send the protocol update data packet to the protocol stack of the terminal device to trigger the communication protocol upgrade operation.

[0140] In this embodiment, the heartbeat interval adjustment value, encryption algorithm switching identifier, and session key rotation cycle in the protocol update instruction set are encapsulated. Protocol encapsulation involves packaging these parameter information according to the communication protocol format to form a complete protocol update data packet.

[0141] First, determine the data packet format of the communication protocol, which typically includes a header and a data portion. The header includes the data packet type, destination address, source address, etc., while the data portion contains the specific parameters for protocol updates. The heartbeat interval adjustment value of 15 seconds, the encryption algorithm switch identifier of AES-256, and the session key rotation cycle of 9.6 hours are filled into the data portion according to the encoding method specified in the protocol.

[0142] For example, the header information is set to the packet type as a protocol update packet, the destination address as the protocol stack address of the terminal device, and the source address as the address of the network optimization system. In the data section, the heartbeat interval adjustment value is encoded as a specified byte sequence, the encryption algorithm switching identifier is encoded as a corresponding code, and the session key rotation period is also encoded accordingly.

[0143] The encapsulated protocol update data packet is sent to the protocol stack of the terminal device. After receiving the protocol update data packet, the protocol stack parses it and upgrades the communication protocol based on the parsing results. For example, the heartbeat interval is adjusted to 15 seconds, the encryption algorithm is switched to AES-256, and the session key rotation cycle is shortened to 9.6 hours, thereby improving the security and stability of communication.

[0144] Step S158: Monitor the network status indicators of the terminal device after performing the computing resource reallocation operation, the real-time operation interception service, and the communication protocol upgrade operation. If the network status indicators reach the preset optimization target threshold, generate a trusted communication link upgrade completion identifier.

[0145] In this embodiment, after the terminal device performs computing resource reallocation, real-time operation interception service, and communication protocol upgrade operations, it is necessary to monitor network status indicators. These indicators include network bandwidth utilization, transmission latency, and packet loss rate.

[0146] In this embodiment, data on these network status indicators can be collected in real time. For example, network bandwidth utilization, transmission latency, and packet loss rate data can be collected every certain period of time (e.g., every minute) using network monitoring tools. The preset optimization target thresholds are set according to the network optimization requirements, such as increasing network bandwidth utilization to over 80%, reducing transmission latency to below 20 milliseconds, and reducing packet loss rate to below 1%.

[0147] The real-time collected network status metrics are compared with preset optimization target thresholds. Assuming that after a period of monitoring, network bandwidth utilization reaches 82%, transmission latency decreases to 18 milliseconds, and packet loss rate decreases to 0.8%, all of these metrics have met the preset optimization target thresholds, a trusted communication link upgrade completion identifier can be generated. This identifier can be a specified signal or flag indicating that the terminal device's communication link has been successfully upgraded, and network security and performance have been effectively improved.

[0148] Furthermore, the method may also include the following steps:

[0149] Step S210: Obtain a set of device identity certificate samples, a set of protocol interaction log samples, and a set of trusted authentication tags for historical terminal devices. The set of trusted authentication tags includes device identity legitimacy identifiers, protocol compliance identifiers, and behavior trust scores.

[0150] In this embodiment, when training the trusted authentication model, it is first necessary to obtain a large amount of historical data as training samples. For example, a set of device identity certificates, a set of protocol interaction log samples, and a set of trusted authentication tags for terminal devices can be collected from the historical records of the power grid wireless local area network.

[0151] The device identity certificate sample set contains device identity certificate information for multiple terminal devices, such as certificates for different sensors and smart meters. Each device identity certificate includes a unique identifier for the device, certificate validity period, and certificate issuing authority information. Assume that 1000 device identity certificate samples have been collected, covering devices of different types and from different periods.

[0152] The protocol interaction log sample set records the protocol interactions of these terminal devices during communication. For example, the logs contain information such as the protocol version, encryption algorithm, and session key update cycle used by the device when communicating with the network. Similarly, 1000 corresponding protocol interaction log samples were collected, each corresponding to a device identity certificate sample.

[0153] The trusted authentication tag set consists of tag information labeled based on historical data, including device identity legitimacy identifiers, protocol compliance identifiers, and behavioral trust scores. Device identity legitimacy identifiers are categorized as "legitimate" or "illegitimate," protocol compliance identifiers are categorized as "compliant" or "non-compliant," and the behavioral trust score is a value between 0 and 100. For example, for a sample device identity certificate and its corresponding protocol interaction log sample, after manual review or historical system judgment, its device identity legitimacy identifier is labeled as "legitimate," its protocol compliance identifier as "compliant," and its behavioral trust score as 90.

[0154] Step S211: Perform certificate chain parsing on the device identity certificate sample set to obtain the certificate authority credibility sample features and certificate validity period sample features.

[0155] In this embodiment, certificate chain parsing is a process that starts from the certificate of the terminal device and verifies it step by step up to the root certificate, while extracting key sample features.

[0156] For each device identity certificate sample, the first step is to check if the certificate format conforms to standard specifications and verify the validity of the certificate signature. The signature is then decrypted and verified using the certificate authority's public key. If the signature verification passes, the legitimacy of the certificate authority is further checked. Based on the certificate authority's historical performance and reputation, each certificate authority is assigned a trust score; this score constitutes the certificate authority trust sample characteristic. For example, "PowerCertAuthority" has a trust score of 90, and "SafeNetCert" has a trust score of 85.

[0157] Simultaneously, certificate validity information is extracted from the device identity certificates and used as a sample feature of certificate validity. Certificate validity typically includes a start date and an end date; for example, a device identity certificate might be valid from January 1, 2024 to January 1, 2025. The certificate validity period can be converted into a numerical value, such as calculating the number of valid days; in this case, the valid days for the device identity certificate are 365 days. After processing 1000 device identity certificate samples, 1000 sets of certificate authority trustworthiness sample features and certificate validity period sample features are obtained.

[0158] Step S212: Perform protocol field extraction processing on the protocol interaction log sample set to obtain protocol version matching degree sample features, encryption algorithm identifier sample features, and session key update cycle sample features.

[0159] In this embodiment, the protocol interaction log contains a large amount of protocol-related information, and sample features are obtained by extracting key fields.

[0160] For each protocol interaction log sample, the protocol version information is first extracted and compared with the standard protocol version specified by the network to calculate the protocol version matching degree. Assuming the standard protocol version specified by the network is "IEEE 802.11ax", if the device recorded in the protocol interaction log also uses "IEEE 802.11ax", the protocol version matching degree is 100%; if the older version "IEEE 802.11n" is used, the protocol version matching degree is calculated based on the degree of difference from the standard version, for example, 30%. This yields the protocol version matching degree sample characteristics.

[0161] Next, the encryption algorithm identification information is extracted. To ensure communication security, the network specifies the use of designated encryption algorithms, such as "AES-256". The encryption algorithm used by the device is checked from the protocol interaction log. If the specified "AES-256" is used, the encryption algorithm identification sample characteristic is "AES-256"; if another unauthorized encryption algorithm is used, such as "DES", the encryption algorithm identification sample characteristic is "DES".

[0162] Finally, session key update cycle information is extracted. The network requires devices to update session keys periodically; assuming the specified update cycle is once every 12 hours, the actual session key update intervals are analyzed from the protocol interaction logs to obtain session key update cycle sample characteristics. For example, a device's session key update cycle might be once every 10 hours. After processing 1000 protocol interaction log samples, 1000 sets of protocol version matching degree sample characteristics, encryption algorithm identifier sample characteristics, and session key update cycle sample characteristics are obtained.

[0163] Step S213: Construct an initial trusted authentication model. Input the trustworthiness sample features of the certificate issuing authority, the certificate validity period sample features, the protocol version matching degree sample features, the encryption algorithm identifier sample features, and the session key update cycle sample features into the initial trusted authentication model for multi-feature fusion processing to generate a predicted behavior trust score.

[0164] In this embodiment, the initial trusted authentication model can adopt a neural network structure, such as a multilayer perceptron (MLP). The initial trusted authentication model includes an input layer, a hidden layer, and an output layer. The number of neurons in the input layer is determined by the number of input features. Here, there are five features: certificate authority trustworthiness sample features, certificate validity period sample features, protocol version matching sample features, encryption algorithm identifier sample features, and session key update cycle sample features. Therefore, the input layer has five neurons.

[0165] The certificate authority trust sample features, certificate validity sample features, protocol version matching sample features, encryption algorithm identifier sample features, and session key update cycle sample features are preprocessed. For example, the certificate authority trust sample features and protocol version matching sample features are normalized to the range of 0 to 1, and the encryption algorithm identifier sample features are encoded and converted into numerical values.

[0166] The preprocessed sample features are input into the initial trusted authentication model. Inside the model, neurons in the input layer pass feature information to the hidden layers, where neurons perform nonlinear transformations and feature fusion. After processing through multiple hidden layers, the information is passed to the output layer, where neurons output a predicted behavior score. For example, for the first sample, its certificate authority credibility is 0.9 (normalized), certificate validity is 365 days (after some numerical conversion), protocol version matching is 1.0 (normalized), encryption algorithm identifier is 2, and session key update cycle is 10 hours (after numerical conversion). These feature values ​​are input into the initial trusted authentication model, processed by weighted summation and nonlinear activation functions in the hidden layers (e.g., using a sigmoid function to map the input values ​​to the range of 0 to 1), and then passed through multiple layers until the output layer finally outputs a predicted behavior credibility score, assuming it is 0.85, which is 85 points (converted to a range of 0 to 100). By performing this process sequentially on 1000 sets of sample features, 1000 confidence scores for predicted behavior are obtained.

[0167] Step S214: Perform loss calculation processing based on the predicted behavior credibility score and the behavior credibility score in the credibility authentication label set to generate a model training loss value. Perform parameter iterative optimization processing on the initial credibility authentication model based on the model training loss value until the model training loss value converges to obtain the trained credibility authentication model.

[0168] In this embodiment, mean squared error (MSE) can be used as the loss function, that is, the square of the difference between the predicted behavior confidence score and the actual behavior confidence score of each sample is calculated, and then the average value is calculated.

[0169] For example, the predicted behavior credibility score of the first sample is 85, and the corresponding real behavior credibility score in the credible authentication label set is 90. The difference between them is 85-90=-5, and the square of the difference is (-5)×(-5)=25. This calculation is performed on all 1000 samples. Then, the squares of the differences between all samples are added together, and the result is divided by the sample size of 1000 to obtain the model training loss value. Let's assume the calculated model training loss value is 10.

[0170] Based on the training loss value of this model, the parameters of the initial trusted authentication model are iteratively optimized. The gradient descent algorithm is adopted, the core idea of ​​which is to calculate the gradient of the loss function with respect to the model parameters, and then update the model parameters in the opposite direction of the gradient to reduce the loss value.

[0171] First, calculate the gradient of the loss function with respect to the model parameters (such as weights and biases in a neural network). Backpropagation can be used to efficiently calculate the gradient. For each parameter, update it based on the magnitude and direction of the gradient. For example, for a weight parameter with a gradient of 0.1 and a learning rate of 0.01, the update amount for that weight parameter would be -0.01 × 0.1 = -0.001. Subtracting this update amount from the weight parameter completes one parameter update.

[0172] The process of sample input, loss calculation, gradient calculation, and parameter update is repeated continuously, with each iteration gradually reducing the model's training loss value. Iteration stops when the model's training loss value decreases to a very small value and no longer shows significant change after multiple iterations, indicating convergence. Assuming that after 1000 iterations, the model's training loss value converges to below 1, a trained trustworthy authentication model is obtained. This model can then be used to authenticate new terminal devices and generate accurate real-time behavioral trustworthiness scores.

[0173] Furthermore, the method may also include the following steps:

[0174] Step S310: Obtain a set of historical cross-domain association feature samples, a set of historical business operation semantic samples, and a set of multi-domain collaborative optimization strategy labels. The set of multi-domain collaborative optimization strategy labels includes resource scheduling priority sequence labels, abnormal operation interception rule labels, and protocol adaptive adjustment parameter labels.

[0175] In this embodiment, a large amount of historical data needs to be collected to train the multi-domain collaborative analysis model. This involves obtaining a set of historical cross-domain association feature samples, a set of historical service operation semantic samples, and a set of multi-domain collaborative optimization strategy labels from the historical records of the power grid wireless local area network.

[0176] The historical cross-domain correlation feature sample set contains correlation information between device operating status and network transmission quality over multiple time periods, such as the correlation between device power consumption and transmission latency, CPU load and bandwidth utilization, and memory usage and packet loss rate. Assume that 500 historical cross-domain correlation feature samples from different time periods have been collected, with each sample containing specific values ​​for these three correlations.

[0177] The historical business operation semantic sample set records historical business operation instructions and their related semantic information, such as business operation intent vectors and semantic compliance scores. For example, 500 business operation instruction samples were collected, and each sample underwent semantic analysis to obtain a corresponding business operation intent vector and semantic compliance score.

[0178] The multi-domain collaborative optimization strategy tag set is annotated with optimization strategy information based on historical realities. It includes resource scheduling priority sequence tags, abnormal operation interception rule tags, and protocol adaptive adjustment parameter tags. The resource scheduling priority sequence tags clarify the resource scheduling priority order for different business domains, for example, business domain A > business domain B > business domain C. The abnormal operation interception rule tags specify the rules for judging abnormal operations, such as an operation complexity score exceeding 0.8 being considered an abnormal operation. The protocol adaptive adjustment parameter tags contain specific protocol adjustment parameters, such as adjusting the heartbeat interval to 15 seconds, switching the encryption algorithm to AES-256, and adjusting the session key rotation cycle to 9.6 hours.

[0179] Step S311: Standardize the historical cross-domain association feature sample set to obtain a standardized cross-domain association feature sample set.

[0180] In this embodiment, the historical cross-domain correlation feature sample set is standardized in order to unify data from different ranges to the same scale, which facilitates model processing.

[0181] For each sample in the historical cross-domain correlation feature sample set, including the correlation between device power consumption and transmission delay, the correlation between CPU load and bandwidth utilization, and the correlation between memory usage and packet loss rate, a min-max normalization method is used to first find the minimum and maximum values ​​of each correlation feature among all samples.

[0182] For example, the correlation between device power consumption and transmission delay in 500 samples ranged from a minimum of 0.2 to a maximum of 0.8; the correlation between CPU load and bandwidth utilization ranged from a minimum of 0.1 to a maximum of 0.7; and the correlation between memory usage and packet loss rate ranged from a minimum of 0.05 to a maximum of 0.6.

[0183] For the first sample, the correlation between device power consumption and transmission delay is 0.3, and the standardized calculation process is (0.3-0.2)÷(0.8-0.2)=0.1÷0.6≈0.17; the correlation between CPU load and bandwidth utilization is 0.2, and the standardized calculation process is (0.2-0.1)÷(0.7-0.1)=0.1÷0.6≈0.17; the correlation between memory usage and packet loss rate is 0.1, and the standardized calculation process is (0.1-0.05)÷(0.6-0.05)=0.05÷0.55≈0.09.

[0184] By performing this standardization process on all 500 samples, we obtain a standardized cross-domain association feature sample set.

[0185] Step S312: Perform dimensionality reduction processing on the historical business operation semantic sample set to obtain a low-dimensional business operation semantic sample set.

[0186] In this embodiment, the historical business operation semantic sample set is subjected to dimensionality reduction processing to reduce the data dimensionality and improve the model's processing efficiency. Principal Component Analysis (PCA) is used for dimensionality reduction.

[0187] First, the business operation intent vector and semantic compliance score from the historical business operation semantic sample set are combined into a high-dimensional vector. Assuming the business operation intent vector has 5 dimensions and the semantic compliance score is the 6th dimension, then each sample is a 6-dimensional vector.

[0188] Calculate the covariance matrix of these 500 6-dimensional vectors. The covariance matrix reflects the correlation between the various dimensions. Then, solve for the eigenvalues ​​and eigenvectors of the covariance matrix. The eigenvalues ​​represent the importance of each eigenvector, and the eigenvectors with the largest eigenvalues ​​are selected to form the projection matrix.

[0189] Assume that, after calculation, the first two feature vectors are selected to form a projection matrix. Each 6-dimensional business operation semantic sample vector is projected onto this 2-dimensional low-dimensional space, resulting in a 2-dimensional low-dimensional business operation semantic sample vector. For example, the 6-dimensional vector of the first sample is [1, 0, 0, 0, 0, 90], which, after projection, becomes the 2-dimensional vector [0.8, 0.2]. Performing this dimensionality reduction process on all 500 samples yields the low-dimensional business operation semantic sample set.

[0190] Step S313: Construct an initial multi-domain collaborative analysis model. Input the standardized cross-domain related feature sample set and the low-dimensional business operation semantic sample set into the initial multi-domain collaborative analysis model for spatiotemporal context aggregation processing to generate a predicted resource scheduling priority sequence, a predicted abnormal operation interception rule, and a predicted protocol adaptive adjustment parameter.

[0191] In this embodiment, the initial multi-domain collaborative analysis model can adopt a deep learning-based architecture, such as the Transformer architecture, which can handle sequence data and perform context aggregation well.

[0192] The standardized cross-domain association feature sample set and the low-dimensional business operation semantic sample set are input into the initial multi-domain collaborative analysis model. First, the input samples are encoded to convert the standardized cross-domain association features and the low-dimensional business operation semantic features into vector representations that the model can process.

[0193] Internally, the model employs a multi-head attention mechanism for spatiotemporal context aggregation. This mechanism allows the model to focus on different information within different representation subspaces, thus capturing the relationships between features more comprehensively. For example, for the correlation between device power consumption and transmission delay in standardized cross-domain correlation features and the business operation intent in low-dimensional business operation semantic features, the model uses a multi-head attention mechanism to calculate the correlation weights between them and aggregate the relevant information.

[0194] After processing through multiple attention layers and feedforward neural network layers, the model outputs a predicted resource scheduling priority sequence, a predicted abnormal operation interception rule, and adaptive adjustment parameters for the prediction protocol. For example, the predicted resource scheduling priority sequence is business domain B > business domain A > business domain C; the predicted abnormal operation interception rule is that an operation with a complexity score exceeding 0.9 is considered an abnormal operation; and the predicted protocol adaptive adjustment parameters are: heartbeat interval adjusted to 20 seconds, encryption algorithm switched to AES-192, and session key rotation cycle adjusted to 10 hours. 500 sets of samples are processed sequentially to obtain 500 prediction results.

[0195] Step S314: Perform weighted loss calculation based on the first difference between the predicted resource scheduling priority sequence and the resource scheduling priority sequence label, the second difference between the predicted abnormal operation interception rule and the abnormal operation interception rule label, and the third difference between the predicted protocol adaptive adjustment parameter and the protocol adaptive adjustment parameter label to generate the total model loss value.

[0196] In this embodiment, a normalization method can be used to handle the first difference. For the predicted resource scheduling priority sequence and the resource scheduling priority sequence label, the Kendall's staunch correlation coefficient is used to measure the similarity between them. The value of this coefficient ranges from -1 to 1. The closer the value is to 1, the more similar the two are, and the closer it is to -1, the greater the difference.

[0197] Assume the predicted resource scheduling priority sequence is business domain B > business domain A > business domain C, and the resource scheduling priority sequence label is business domain A > business domain B > business domain C. The Kendall rank correlation coefficient is calculated to be -0.33 (the specific calculation process is as follows: first, determine all possible business domain pairs, here we have three pairs: business domain A-business domain B, business domain A-business domain C, and business domain B-business domain C. Compare the order consistency of each business domain pair in the predicted sequence and the label sequence; divide the number of inconsistent pairs by the total number of pairs to get a value, then calculate -0.33 according to the Kendall rank correlation coefficient formula). To convert this into a difference measure, subtract the absolute value of the coefficient from 1 to obtain the first difference measure value: 1 - |-0.33| = 0.67.

[0198] Secondly, the prediction of abnormal operation interception rules and their labels mainly involves differences in thresholds. Similarly, a normalization method is used to convert these threshold differences into values ​​between 0 and 1.

[0199] Suppose the predicted abnormal operation interception rule is that an operation with a complexity score exceeding 0.9 is considered an abnormal operation, while the abnormal operation interception rule label is that an operation with a complexity score exceeding 0.8 is considered an abnormal operation. First, calculate the difference in thresholds as |0.9-0.8| = 0.1. Since the threshold range for abnormal operation interception rules is usually between 0 and 1, this difference is directly used as the second difference metric, i.e., 0.1.

[0200] In addition, the protocol adaptively adjusts parameters including heartbeat interval, encryption algorithm, and session key rotation cycle, which require handling the differences between these parameters separately and then integrating them.

[0201] Heart rate interval difference: Assume the predicted heart rate interval is adjusted to 20 seconds, and the protocol adaptively adjusts the parameter label to 15 seconds. To normalize this, a reasonable range of heart rate intervals is first determined, assuming a maximum heart rate interval of 60 seconds and a minimum heart rate interval of 5 seconds. The normalized value of the difference is calculated as (|20-15|)÷(60-5)=5÷55≈0.09.

[0202] Encryption Algorithm Difference: For encryption algorithms, quantization is performed based on factors such as security level and complexity. Assume the quantization value for AES-128 is 1, for AES-192 it's 2, and for AES-256 it's 3. The predicted encryption algorithm is AES-192, the label is AES-256, and the difference quantization value is |2-3|=1. To normalize this, assuming the maximum difference in encryption algorithm quantization values ​​is 3 (from the lowest security level to the highest security level), the normalized value for the encryption algorithm difference is 1÷3≈0.33.

[0203] Session key rotation period difference: Assume the predicted session key rotation period is 10 hours and the tag is 9.6 hours. Determine a reasonable range for the session key rotation period, assuming a maximum period of 24 hours and a minimum period of 1 hour. Calculate the normalized value of the difference as (|10-9.6|)÷(24-1)=0.4÷23≈0.02.

[0204] The third difference metric is obtained by weighted summation of these three normalized difference values. Assuming the weights of the heartbeat interval, encryption algorithm, and session key rotation cycle are 0.2, 0.6, and 0.2 respectively, the third difference metric is 0.09×0.2+0.33×0.6+0.02×0.2=0.018+0.198+0.004=0.22.

[0205] Finally, assign different weights to each difference. Assume the weight of the first difference is 0.3, the weight of the second difference is 0.3, and the weight of the third difference is 0.4. Multiply the first, second, and third differences by their respective weights, and then sum them to obtain the total model loss. That is, 0.67×0.3+0.1×0.3+0.22×0.4=0.201+0.03+0.088=0.319.

[0206] Step S315: Perform backpropagation optimization on the initial multi-domain collaborative analysis model based on the total model loss value until the total model loss value reaches the preset convergence condition, and obtain the trained multi-domain collaborative analysis model.

[0207] In this embodiment, the backpropagation algorithm is an important method in deep learning for calculating gradients and updating model parameters.

[0208] First, calculate the gradient of the loss function with respect to the model parameters (such as the weights and biases in the Transformer architecture) based on the total model loss value. Then, using the chain rule, calculate the gradient layer by layer, starting from the output layer, and backpropagate the gradient information to the parameters of each layer.

[0209] Then, an optimization algorithm (such as the Adam optimization algorithm) is used to update the model parameters based on the calculated gradients. The Adam optimization algorithm combines the ideas of momentum and adaptive learning rate, enabling more efficient parameter updates. For example, for a certain weight parameter, it is updated according to its gradient and the update rules of the Adam optimization algorithm, causing the weight parameter to adjust in the direction of reducing the loss value.

[0210] The process of repeatedly inputting samples, calculating loss, calculating gradients, and updating parameters gradually reduces the total model loss with each iteration. The preset convergence condition can be that the total model loss is less than a set threshold, such as 0.1, or that the change in the total model loss over multiple iterations is less than a very small value. Assuming that after 2000 iterations, the total model loss converges to 0.08, meeting the preset convergence condition, the trained multi-domain collaborative analysis model is obtained. This model can then be used to process new cross-domain correlation features and business operation semantic features to generate accurate multi-domain collaborative optimization strategies.

[0211] Figure 2 The diagram illustrates exemplary hardware and software components of a trusted authentication-based multi-domain data processing system 100 for power WLANs, which can implement the inventive ideas of the present invention, according to some embodiments of the present invention. For example, a processor 120 may be used in the trusted authentication-based multi-domain data processing system 100 for power WLANs and to perform the functions of the present invention.

[0212] The trusted authentication-based multi-domain data processing system 100 for power WLANs can be a general-purpose server or a special-purpose server; both can be used to implement the trusted authentication-based multi-domain data processing method for power WLANs of this invention. Although only one server is shown in this invention, for convenience, the functions described in this invention can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0213] For example, a power WLAN multi-domain data processing system 100 based on trusted authentication may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and various forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the power WLAN multi-domain data processing system 100 based on trusted authentication may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present invention can be implemented according to these program instructions. The power WLAN multi-domain data processing system 100 based on trusted authentication also includes an input / output (I / O) interface 150 between the computer and other input / output devices.

[0214] For ease of explanation, only one processor is described in the trusted authentication-based power WLAN multi-domain data processing system 100. However, it should be noted that the trusted authentication-based power WLAN multi-domain data processing system 100 of this invention may also include multiple processors. Therefore, the steps performed by one processor as described in this invention may also be performed jointly or individually by multiple processors. For example, if the processor of the trusted authentication-based power WLAN multi-domain data processing system 100 performs steps A and B, it should be understood that steps A and B may also be performed jointly by two different processors or individually by one processor. For example, the first processor performs step A, the second processor performs step B, or the first processor and the second processor jointly perform steps A and B.

[0215] Furthermore, this embodiment of the invention also provides a readable storage medium, wherein computer-executable instructions are preset in the readable storage medium, and when the processor executes the computer-executable instructions, the above-mentioned multi-domain data processing method for power wireless local area networks based on trusted authentication is implemented.

[0216] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A power wireless local area network multi-domain data processing method based on trusted authentication, characterized in that, The method comprises: calling a trusted authentication model to perform trustworthiness authentication processing on a terminal device in a power wireless local area network, generating a terminal authentication data set, the terminal authentication data set containing device identity features, communication protocol compliance features and real-time behavior trustworthiness scores; obtaining cross-domain original data streams of multiple service domains in the power wireless local area network, performing multi-modal feature extraction processing on the cross-domain original data streams to obtain device running state features, network transmission quality features and service operation semantic features; based on the terminal authentication data set, performing trustworthiness weighted fusion processing on the device running state features to generate a trusted device running feature set, and performing dynamic correlation analysis processing on the trusted device running feature set and the network transmission quality features to generate a cross-domain correlation feature set; calling a pre-trained multi-domain collaborative analysis model to perform spatio-temporal context aggregation processing on the cross-domain correlation feature set and the service operation semantic features to generate a multi-domain collaborative optimization strategy set, the multi-domain collaborative optimization strategy set containing resource scheduling priority sequences, abnormal operation interception rules and protocol adaptive adjustment parameters; based on the multi-domain collaborative optimization strategy set, performing dynamic resource reconfiguration processing on the service domain nodes of the power wireless local area network to generate a network optimization execution instruction set, and feeding back the network optimization execution instruction set to the terminal device to trigger a trusted communication link upgrade operation.

2. The power wireless local area network multi-domain data processing method based on trusted authentication according to claim 1, characterized in that, The method comprises: obtaining a device identity certificate and a protocol interaction log submitted by the terminal device when accessing the power wireless local area network, performing certificate chain verification processing on the device identity certificate to obtain a device identity verification result and a certificate authority trustworthiness score; performing protocol compliance analysis processing on the protocol interaction log to extract protocol version matching degree, encryption algorithm compliance identifier and session key update period; calling a trusted authentication model to perform trustworthiness fusion output on the device identity verification result, the certificate authority trustworthiness score, the protocol version matching degree, the encryption algorithm compliance identifier and the session key update period to generate the real-time behavior trustworthiness score; performing dynamic comparison processing on the real-time behavior trustworthiness score and a preset trustworthiness threshold, if the real-time behavior trustworthiness score is greater than or equal to the trustworthiness threshold, generating a device identity legal identifier and a communication protocol compliance identifier, and storing the device identity legal identifier, the communication protocol compliance identifier and the real-time behavior trustworthiness score in association as the terminal authentication data set. 3.The power wireless local area network multi-domain data processing method based on trusted authentication of claim 1, wherein, The method comprises: performing time series analysis processing on the device running log in the cross-domain original data stream to extract a device power consumption fluctuation sequence, a CPU load mean value and a memory occupation peak value, and performing normalized splicing processing on the device power consumption fluctuation sequence, the CPU load mean value and the memory occupation peak value to generate the device running state features; The network transmission message in the cross-domain original data stream is subjected to protocol analysis processing, transmission delay distribution, bandwidth utilization rate and packet loss rate trend are extracted, the transmission delay distribution, the bandwidth utilization rate and the packet loss rate trend are subjected to weighted calculation processing based on a preset network quality evaluation rule, and the network transmission quality feature is generated; The service operation instruction in the cross-domain original data stream is subjected to semantic segmentation processing, an operation instruction text segment set is obtained, a pre-trained semantic analysis model is called to perform intent recognition processing on the operation instruction text segment set, a service operation intent vector and a semantic compliance score are generated, and the service operation intent vector and the semantic compliance score are associated and encoded to generate the service operation semantic feature.

4. The power wireless local area network multi-domain data processing method based on trusted authentication according to claim 1, characterized in that, The terminal authentication data set is used to perform credibility weighted fusion processing on the device running state feature to generate a trusted device running feature set, including: An real-time behavior credibility score is extracted from the terminal authentication data set, and a device running feature weighting coefficient is determined according to a mapping relationship between the real-time behavior credibility score and a preset score interval; The device power consumption fluctuation sequence, CPU load mean value and memory occupation peak value in the device running state feature are subjected to sliding window mean value calculation processing respectively to obtain power consumption fluctuation mean value, CPU load mean value sequence and memory occupation mean value sequence; The power consumption fluctuation mean value, the CPU load mean value sequence and the memory occupation mean value sequence are subjected to dynamic weighted fusion processing based on the device running feature weighting coefficient to generate weighted device power consumption feature, weighted CPU load feature and weighted memory occupation feature; The weighted device power consumption feature, the weighted CPU load feature and the weighted memory occupation feature are subjected to time sequence alignment splicing processing to generate the trusted device running feature set.

5. The power wireless local area network multi-domain data processing method based on trusted authentication according to claim 1, characterized in that, The trusted device running feature set and the network transmission quality feature are subjected to dynamic correlation analysis processing to generate a cross-domain correlation feature set, including: The weighted device power consumption feature, the weighted CPU load feature and the weighted memory occupation feature in the trusted device running feature set are subjected to timestamp alignment processing to obtain a time sequence synchronization device running feature sequence; The transmission delay distribution, bandwidth utilization rate and packet loss rate trend in the network transmission quality feature are subjected to sliding window statistical processing to generate a transmission delay mean value sequence, a bandwidth utilization rate mean value sequence and a packet loss rate mean value sequence; A pre-trained correlation analysis model is called to perform multivariate correlation calculation processing on the time sequence synchronization device running feature sequence, the transmission delay mean value sequence, the bandwidth utilization rate mean value sequence and the packet loss rate mean value sequence to generate device power consumption and transmission delay correlation degree, CPU load and bandwidth utilization rate correlation degree, and memory occupation and packet loss rate correlation degree; The device power consumption and transmission delay correlation degree, the CPU load and bandwidth utilization rate correlation degree, and the memory occupation and packet loss rate correlation degree are subjected to multidimensional vectorization processing to generate the cross-domain correlation feature set.

6. The trusted authentication based power wireless local area network multi-domain data processing method of claim 1, wherein, The calling pre-trained multi-domain collaborative analysis model performs spatio-temporal context aggregation processing on the cross-domain association feature set and the business operation semantic feature, and generates a multi-domain collaborative optimization strategy set, including: The device power consumption and transmission delay correlation degree, CPU load and bandwidth utilization correlation degree, and memory occupation and packet loss rate correlation degree in the cross-domain association feature set are standardized to obtain a standardized association feature vector; The business operation intention vector and semantic compliance score in the business operation semantic feature are subjected to feature dimension reduction processing to obtain a low-dimensional semantic feature vector; The standardized association feature vector and the low-dimensional semantic feature vector are subjected to spatio-temporal position coding processing to generate a spatio-temporal context feature matrix; The multi-domain collaborative analysis model is called to perform multi-head attention aggregation processing on the spatio-temporal context feature matrix to generate resource scheduling priority weight, abnormal operation detection threshold, and protocol adjustment sensitivity parameter; The resource scheduling priority sequence is generated according to the resource scheduling priority weight, the abnormal operation interception rule is generated based on the abnormal operation detection threshold, and the protocol adaptive adjustment parameter is generated according to the protocol adjustment sensitivity parameter, and the resource scheduling priority sequence, abnormal operation interception rule and protocol adaptive adjustment parameter are associated and stored as the multi-domain collaborative optimization strategy set. 7.The power wireless local area network multi-domain data processing method based on trusted authentication of claim 1, wherein, The network optimization execution instruction set is generated by performing dynamic resource reconfiguration processing on the business domain node of the power wireless local area network based on the multi-domain collaborative optimization strategy set, and the network optimization execution instruction set is fed back to the terminal device to trigger a trusted communication link upgrade operation, including: The resource scheduling priority sequence is extracted from the multi-domain collaborative optimization strategy set, and the computing resource pool of the business domain node is subjected to priority sorting processing according to the resource scheduling priority sequence to generate a resource allocation queue; Based on the abnormal operation interception rule, the real-time operation instruction stream of the business domain node is subjected to pattern matching processing, and if an operation instruction matching the abnormal operation interception rule is detected, an operation interception signal is generated and an alarm log record is triggered; According to the protocol adaptive adjustment parameter, the communication protocol stack of the business domain node is subjected to parameter dynamic adjustment processing to generate a protocol update instruction set, and the protocol update instruction set includes a heartbeat interval adjustment value, an encryption algorithm switching identifier, and a session key rotation period; The resource allocation queue, the operation interception signal, and the protocol update instruction set are subjected to instruction encoding processing to generate the network optimization execution instruction set; The resource allocation queue in the network optimization execution instruction set is subjected to instruction analysis processing to generate a resource scheduling configuration parameter, and the resource scheduling configuration parameter is sent to the resource manager of the terminal device to trigger a computing resource reallocation operation; The operation interception signal is subjected to signal encoding processing to generate an operation interception instruction, and the operation interception instruction is sent to the instruction filter of the terminal device to enable a real-time operation interception service; Protocol encapsulation processing is performed on the heartbeat interval adjustment value, the encryption algorithm switching identifier and the session key rotation period in the protocol update instruction set to generate a protocol update data packet, and the protocol update data packet is sent to a protocol stack of the terminal device to trigger a communication protocol upgrade operation; A network state indicator of the terminal device after performing the computing resource reallocation operation, the real-time operation interception service and the communication protocol upgrade operation is monitored, and if the network state indicator reaches a preset optimization target threshold, a trusted communication link upgrade completion identifier is generated.

8. The power wireless local area network multi-domain data processing method based on trusted authentication according to any one of claims 1-7, characterized in that, The training method of the trusted authentication model comprises: Obtaining a device identity certificate sample set, a protocol interaction log sample set and a trusted authentication label set of historical terminal devices, wherein the trusted authentication label set comprises device identity legality identifiers, protocol compliance identifiers and behavior trust scores; Certificate chain analysis processing is performed on the device identity certificate sample set to obtain certificate authority trustworthiness sample features and certificate validity period sample features; Protocol field extraction processing is performed on the protocol interaction log sample set to obtain protocol version matching degree sample features, encryption algorithm identifier sample features and session key update period sample features; An initial trusted authentication model is constructed, and the certificate authority trustworthiness sample features, the certificate validity period sample features, the protocol version matching degree sample features, the encryption algorithm identifier sample features and the session key update period sample features are input into the initial trusted authentication model for multi-feature fusion processing to generate a predicted behavior trust score; Loss calculation processing is performed on the predicted behavior trust score and the behavior trust scores in the trusted authentication label set to generate a model training loss value, and parameter iterative optimization processing is performed on the initial trusted authentication model based on the model training loss value until the model training loss value converges, thereby obtaining the trained trusted authentication model.

9. The power wireless local area network multi-domain data processing method based on trusted authentication according to any one of claims 1-7, characterized in that, The training method of the multi-domain collaborative analysis model comprises: Obtaining a historical cross-domain association feature sample set, a historical business operation semantic sample set and a multi-domain collaborative optimization strategy label set, wherein the multi-domain collaborative optimization strategy label set comprises resource scheduling priority sequence labels, abnormal operation interception rule labels and protocol adaptive adjustment parameter labels; Standardization processing is performed on the historical cross-domain association feature sample set to obtain a standardized cross-domain association feature sample set; Dimension reduction processing is performed on the historical business operation semantic sample set to obtain a low-dimensional business operation semantic sample set; An initial multi-domain collaborative analysis model is constructed, and the standardized cross-domain association feature sample set and the low-dimensional business operation semantic sample set are input into the initial multi-domain collaborative analysis model for spatio-temporal context aggregation processing to generate a predicted resource scheduling priority sequence, a predicted abnormal operation interception rule and a predicted protocol adaptive adjustment parameter; According to the first difference between the predicted resource scheduling priority sequence and the resource scheduling priority sequence label, the second difference between the predicted abnormal operation interception rule and the abnormal operation interception rule label, and the third difference between the predicted protocol adaptive adjustment parameter and the protocol adaptive adjustment parameter label, a weighted loss calculation process is performed to generate a model total loss value; Based on the model total loss value, the initial multi-domain collaborative analysis model is subjected to a back propagation optimization process until the model total loss value reaches a preset convergence condition, and a trained multi-domain collaborative analysis model is obtained.

10. A power wireless local area network multi-domain data processing system based on trusted authentication, characterized by, The method comprises a processor and a memory, the memory and the processor are connected, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to realize the power wireless local area network multi-domain data processing method based on trusted authentication in any one of claims 1-9.

Images