A method and system for 6G WLAN access state open and closed loop control

By collecting and analyzing access status time series in the wireless LAN access architecture, and establishing the mapping weight between access status feature quantities and control actions, the problem of lagging core network status information is solved, and more accurate and real-time access control is achieved.

CN123124053APending Publication Date: 2026-11-17NANJING AIPULU SATELLITE COMMUNICATION TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611604870.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-16
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In the existing wireless LAN access architecture, the access status information obtained by the core network lags behind the actual status of the wireless LAN, resulting in insufficient real-time performance and accuracy of access control.

Method used

By collecting quasi-static access attributes, dynamic link status, and terminal access session status from the wireless LAN side, a WLAN access status timing set is formed. This set is then sent to the NWDAF via an independent out-of-band management plane for time alignment and feature analysis. The effective response timing is calculated, and a mapping weight between access status features and control actions is established to form a closed-loop control process.

Benefits of technology

It improves the accuracy and real-time performance of wireless LAN access control, reduces the impact of service transmission paths, and enhances the targeting and adaptability of control.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123124053A_ABST
    Figure CN123124053A_ABST
Patent Text Reader

Abstract

This invention relates to the field of wireless communication network technology and discloses a method and system for open and closed-loop control of WLAN access status for 6G. The method includes: aligning the time of each state data using the terminal access session as an index to form a WLAN access state timing set; sending the WLAN access state timing set to a Network Data Analysis Function (NWDAF) via an independent out-of-band management plane to form access state feature quantities; registering the differences in access state feature quantities before and after historical access control actions according to the effective response timing sequence using the NWDAF to obtain mapping weights; calculating the state response values ​​of candidate access control actions based on the current access state feature quantities and mapping weights, and converting them into control commands; calculating the mapping residual between the actual state response and the expected state response according to the effective response timing sequence, and recalculating the control priority of candidate access control actions. This invention has the advantage of improving the accuracy of wireless local area network access control.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of wireless communication network technology, specifically to a method and system for controlling the open and closed loop access status of WLAN for 6G. Background Technology

[0002] With the large-scale application of fifth-generation mobile communication technology and the development of sixth-generation mobile communication technology, wireless local area networks (WLANs) are gradually evolving from a supplementary access method to a converged access method for mobile communication networks. Existing technologies typically achieve WLAN-mobile communication core network interoperability through non-3G Partner Program (NDP) interoperability functions or trusted NDP gateway functions, enabling terminals to access core network services via WLAN.

[0003] In the existing wireless LAN access architecture, status information such as wireless access point load, channel utilization, wireless interference, and terminal received signal strength, signal-to-noise ratio, and packet loss rate are mainly collected and managed on the wireless LAN side. The core network usually needs to go through network management or other intermediate processing links to obtain this type of status information. When the terminal access status, wireless link quality, or access point load changes rapidly, the status information obtained by the core network is prone to lag behind the actual status of the wireless LAN. This causes the core network to make access and service quality decisions based on the lagging status, and the status changes after the decision are not easily transmitted back in time to participate in subsequent decisions. This causes the timing of the dynamic status changes of the wireless LAN, the core network access decisions, and the wireless LAN-side control execution to be asynchronous, which in turn affects the real-time performance and accuracy of access control.

[0004] Therefore, it is necessary to design a WLAN access state open and closed loop control method and system for 6G to improve the accuracy of wireless local area network access control. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for 6G WLAN access status open and closed-loop control, which has the advantage of improving the accuracy of wireless local area network access control and solves the problems mentioned in the background technology.

[0006] To achieve the aforementioned goal of improving the accuracy of wireless LAN access control, this invention provides the following technical solution: a WLAN access state open and closed-loop control method for 6G, comprising the following steps: Collect the quasi-static access attributes, dynamic link status, and access session status and service quality status of the corresponding terminals of each wireless access point on the wireless LAN side. Use the terminal access session as an index to align the status data in time and form a WLAN access status time series set. The WLAN access state timing set is sent to the network data analysis function NWDAF through an independent out-of-band management plane. The state transition residuals of dynamic link state changes and access session state changes at different time offsets are calculated. The effective response timing is determined by the time offset corresponding to the minimum residual. Based on this, the quasi-static access attributes and dynamic link states are fused to form access state feature quantities. The NWDAF registers the access state feature differences before and after historical access control actions according to the effective response time sequence, calculates the normalized response coefficient between the access state feature differences and the access control action amplitude, and normalizes the normalized response coefficient according to the access control action to obtain the mapping weight of each access state feature corresponding to each access control action. Calculate the status response value of candidate access control actions based on the current access status characteristics and mapping weights, determine the control priority based on the current service quality status deviation, select access migration, radio resource adjustment or service quality parameter correction actions, convert them into control instructions that can be executed by the radio access point and send them out through the independent out-of-band management plane. The system collects the dynamic link status, access session status, and service quality status after the control command is executed. It calculates the mapping residual between the actual status response and the expected status response according to the effective response sequence. It corrects the corresponding mapping weights based on the mapping residuals and recalculates the control priority of the candidate access control actions based on the corrected mapping weights and access status features.

[0007] A WLAN access state open and closed-loop control system for 6G includes: Access Status Module: Collects the access, link, and service quality status of wireless access points and terminals, performs session-level time alignment, and forms a WLAN access status time sequence set; Timing Opening Module: Sends the WLAN access state timing set to NWDAF through an independent out-of-band management plane, determines the valid response timing, and forms access state characteristic quantities; Response mapping module: Establishes mapping weights between historical control actions and changes in access status; Action orchestration module: Determines control priority based on current access status features and mapping weights, and generates corresponding access control commands; Response correction module: Corrects the mapping weights based on the mapping residual between the actual state response and the expected state response after control, and updates the control priority.

[0008] Compared with existing technologies, the present invention provides a method and system for WLAN access state open and closed-loop control for 6G, which has the following advantages: This invention aligns the WLAN access state with the terminal access session as an index for time synchronization and determines the effective response timing based on the state transition residuals under different time offsets. This eliminates timing deviations between different state data and improves the accuracy of the correspondence between access state changes and control action responses. By sending the WLAN access state to the NWDAF for unified analysis through an independent out-of-band management plane, it opens the WLAN access state to the 6G network analysis and control system and reduces the impact on service transmission paths. By performing time-series registration of historical access control actions and changes in access state characteristics, it establishes mapping weights between access state characteristics and control actions, providing a practical basis for control action selection. Control priorities are determined based on current access state characteristics, mapping weights, and service quality state deviations, allowing for the selection of access migration, radio resource adjustment, or service quality parameter correction actions tailored to the current network state, improving the targeting of access control. After the control command is executed, the mapping weights are corrected based on the mapping residuals between the actual and expected state responses, ensuring that historical control responses are continuously updated with network state changes. This forms a closed-loop control process of state acquisition, control decision-making, execution feedback, and weight correction, improving the real-time performance, accuracy, and adaptability of WLAN access control. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of the method of the present invention; Figure 2 This is a schematic diagram of the system of the present invention. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] Example 1: Please refer to Figure 1 As shown in the figure, a WLAN access state open and closed-loop control method for 6G according to an embodiment of the present invention includes the following steps: S1: Collect the quasi-static access attributes, dynamic link status, and access session status and service quality status of each wireless access point on the wireless LAN side. Use the terminal access session as an index to time-align each status data to form a WLAN access status time series set.

[0012] In S1, the process of forming the WLAN access state timing set is as follows: The system collects the access address, service area, wireless standard, operating channel, frequency band, and resource capabilities of each wireless access point to form quasi-static access attributes. It periodically reads the quasi-static configuration information of each wireless access point through the local management interface, SNMP management protocol, or CAPWAP control plane interface. For the access address, it collects the BSSID, MAC address, and IP address; for the service area, it collects the location name, floor, area code, or geographical location information of the access point; for the wireless standard, it collects the wireless standard types supported and currently enabled by the access point, including 802.11a / b / g / n / ac / ax / be and their operating modes; for the operating channel... The system collects the primary channel number, channel bandwidth, and channel binding status currently used by the access point; for frequency band, it collects the current operating frequency band of the access point, including 2.4GHz, 5GHz, or 6GHz; for resource capabilities, it collects the maximum number of associated terminals supported by the access point, maximum throughput, number of available radio links, number of antennas, and supported QoS levels. The above information is used to build a quasi-static access attribute table according to the access point identifier and is updated when the set collection period arrives; when the quasi-static configuration of an access point changes, the table is updated in an event-triggered manner and the change time is recorded. The quasi-static access attribute table is stored according to the access point identifier and update timestamp for subsequent association using terminal access sessions as an index.

[0013] The system collects channel utilization, received signal strength, signal-to-noise ratio (SNR), link rate, retransmission rate, and packet loss rate to form a dynamic link status. According to a set sampling period, dynamic link parameters are read from the RF statistics interface, driver layer statistics interface, or management information database of each wireless access point. For channel utilization, the busy / idle ratio of the current working channel during the sampling period is collected, including total utilization and the proportion of non-WLAN interference. For received signal strength, the most recently measured RSSI value from the terminal or access point is collected and recorded separately for each terminal access session. For signal-to-noise ratio, the current SNR or SINR of the corresponding link is collected. For link rate, the current negotiated rate, actual transmission rate, and reception rate are collected. For retransmission rate, the ratio of MAC layer retransmissions to total transmissions is collected. For packet loss rate, the packet loss ratio of the link layer or network layer during the sampling period is collected. These parameters are associated with the terminal access session identifier and access point identifier to form a dynamic link status record. Each record includes a sampling time tag and a sampling period identifier and is stored in the dynamic link status table.

[0014] The system collects data on the associated wireless access point of the terminal, access duration, uplink and downlink traffic, and access status changes to form the access session state. It also collects terminal latency, throughput, packet loss rate, jitter, and service level to form the service quality state. Data collection is performed at the terminal access session level. For the access session state, the system collects the currently associated wireless access point identifier, the access duration between the terminal and the access point, uplink and downlink traffic within the sampling period, and access status change events. Access status change events include terminal online / offline, roaming handover, authentication failure, or re-association. The access duration is determined by the difference between the current association establishment time and the current sampling time. Uplink and downlink traffic are calculated from the access point's forwarding count or... The data is obtained from the terminal side statistics. Access status change events are recorded in the form of event logs with timestamps. For service quality status, the latency, throughput, packet loss rate, jitter, and service level of the terminal service flow are collected within the current sampling period. Latency is calculated from the round-trip time of probe messages between the terminal and the gateway or the service flow timestamp. Throughput is calculated from the number of bytes successfully transmitted per unit time. Packet loss rate is calculated from sequence number gaps or duplicates. Jitter is calculated from the difference between adjacent latency samples. Service level is mapped from the service flow identifier, DSCP field, or subscription information. The above access session status and service quality status are all indexed according to the terminal access session identifier and stored separately after being associated with the sampling time.

[0015] Using the terminal access session identifier as a unified index, the quasi-static access attributes, dynamic link states, access session states, and service quality states corresponding to the same terminal access session are associated according to the sampling time. Data from different sampling periods are resampled, and missing state quantities are filled in with the most recent valid sample value to form access state records under a unified time axis. Using the terminal access session identifier as the primary key, the quasi-static access attributes, dynamic link states, access session states, and service quality states corresponding to the same terminal access session are associated. The time alignment unit first sets a unified time axis, and the granularity of the unified time axis is set to the least common multiple or the highest multiple of the sampling intervals of each state data. The sampling interval is an integer multiple of the small sampling interval. Then, each type of state data is resampled according to a unified time axis: for periodic data, linear interpolation or hold-at-the-moment interpolation is used to map it to the unified time axis; for event data, the event nearest neighbor method is used to map it to the unified time axis. For a missing state quantity at a certain moment, the most recent valid sample value is used to fill it in, and the source time of the value and the filling mark are recorded. After alignment, each terminal access session forms an access state record on the unified time axis. This record contains the quasi-static access attributes, dynamic link status, access session status and service quality status at that moment. Each record is stored according to the terminal access session identifier and timestamp.

[0016] The access status records are arranged according to terminal access sessions and wireless access points, and the status change positions corresponding to access point switching, link state changes, and service quality anomalies are marked, forming a WLAN access status time series set that includes static attributes, dynamic links, session changes, and service quality changes. After completing the unified time axis alignment, a time series set construction unit is set up to arrange the access status records under the unified time axis according to two dimensions: first, according to the terminal access session identifier, forming a state sequence from establishment to release for each terminal access session; second, according to the wireless access point identifier, forming a state sequence of all associated terminal access sessions under each wireless access point. Based on the arrangement, the time series set construction unit marks the change positions of the access status records. For access point switching, when the associated wireless access points corresponding to the same terminal access session change, the change positions are marked. When the entry point identifier changes, the moment is marked as the access point switching location, and the access point before and after the switch is recorded. For link state mutations, when the change in channel utilization, received signal strength, signal-to-noise ratio, link rate, retransmission rate, or packet loss rate in the dynamic link state exceeds a certain multiple of the standard deviation of the corresponding historical change, the moment is marked as the link state mutation location, and the parameter type and direction of change are recorded. For service quality anomalies, when the latency, throughput, packet loss rate, or jitter in the service quality state exceeds the allowable range of the corresponding service level, the moment is marked as the service quality anomaly location, and the anomaly index and anomaly degree are recorded. After marking, each access state record, change location mark, and corresponding timestamp, session identifier, and access point identifier are associated and stored to form a WLAN access state time series set.

[0017] S2: The WLAN access state timing set is sent to the network data analysis function NWDAF through the independent out-of-band management plane. The state transition residuals of dynamic link state changes and access session state changes at different time offsets are calculated. The effective response timing is determined by the time offset corresponding to the minimum residual. Based on this, the quasi-static access attributes and dynamic link states are fused to form access state feature quantities.

[0018] In S2, the process of calculating the state transition residuals of dynamic link state changes and access session state changes at different time offsets is as follows: The WLAN access state time series is divided into a dynamic link state sequence and an access session state sequence according to the terminal access session. The state changes between adjacent sampling times are calculated, and the dynamic link state changes are used as the link response sequence and the access session state changes are used as the session change sequence. Access state records are extracted from the WLAN access state time series according to the terminal access session identifier. For each terminal access session, the access state records of the session are arranged in chronological order on a unified time axis to obtain the access state sequence of the session. Then, two types of data, dynamic link state and access session state, are separated from the access state sequence: the channel utilization, received signal strength, signal-to-noise ratio, link rate, retransmission rate, and packet loss rate in the dynamic link state at each sampling time are extracted and arranged in chronological order to form the dynamic link state sequence; the associated wireless access point identifier and access duration in the access session state at each sampling time are extracted. Uplink traffic, downlink traffic, and access state changes are extracted separately and arranged into an access session state sequence in chronological order. Then, the difference between each state quantity at two adjacent sampling times in the dynamic link state sequence is calculated to obtain the dynamic link state change quantity, and the changes of each state quantity are arranged into a link response sequence in chronological order. The difference between each state quantity at two adjacent sampling times in the access session state sequence is calculated to obtain the access session state change quantity, and the changes of each state quantity are arranged into a session change sequence in chronological order. For non-numerical state quantities such as associated wireless access point identifiers, their changes are represented as handover identifiers, for example, a value of 1 when the associated access point changes and a value of 0 when it does not change. For access state change events, their changes are represented as event counts or event type codes. Both the link response sequence and the session change sequence are accompanied by corresponding timestamps, terminal access session identifiers, and access point identifiers, and are stored in the state change sequence table.

[0019] The state change sequence is transmitted to the Network Data Analysis Function (NWDAF) via an independent out-of-band management plane. This process statistically analyzes the time difference distribution between dynamic link state changes and access session state changes in historical access sessions. The middle main interval of this time difference distribution is used as a candidate time offset interval, and time offsets are sequentially set within this interval according to the sampling interval. Sliding registration is then performed on the link response sequence and session change sequence. The independent out-of-band management plane encapsulates the link response sequence and session change sequence from the state change sequence table into out-of-band management messages and sends them to the NWDAF. The independent out-of-band management plane is separate from the service transmission path. The out-of-band management message includes the terminal access session identifier, access point identifier, sampling time, state quantity type, and change value. Upon receiving the message, the NWDAF sets up a time offset statistics unit to extract the dynamic link state changes and access session state changes of the same terminal access session within the same time window from historical access sessions. The time difference between the two is calculated as follows: for the moment when a certain state quantity changes in the dynamic link state change, the time difference between the dynamic link state change and the access session state change is calculated. In the state change process, the moment when the corresponding state variable changes in the same direction or the change amount is closest is found, and the difference between the timestamps of the two is taken as the time difference of the change. If there are multiple corresponding changes in the same session, the time difference is calculated separately and included in the time difference sample set. The time difference samples of all historical access sessions are counted to form a time difference distribution. The middle main interval of the distribution is determined according to the quantile method. For example, the interval between the 5th percentile and the 95th percentile of the time difference distribution is taken, or the interval after removing the maximum and minimum set proportions is taken as the candidate time offset interval. Then, the time offset is set sequentially in the candidate time offset interval according to the sampling interval. Each time offset corresponds to an offset step size. For each time offset, the link response sequence is shifted relative to the session change sequence according to the time offset, so that each change in the link response sequence is paired with the change at the corresponding moment in the session change sequence to complete the sliding registration. During sliding registration, the timestamp, state variable type and change value of each pair are retained and recorded according to the terminal access session identifier to form a candidate registration result set.

[0020] Calculate the absolute value of the difference between the corresponding state changes in the link response sequence and the session change sequence at each time offset, and normalize them according to the standard deviation of the changes in the corresponding state quantities in the historical access session. Sum the normalized differences of each state quantity to obtain the state transition residual corresponding to each time offset. For each candidate registration result set at each time offset, calculate the absolute value of the difference between the corresponding state changes in the link response sequence and the session change sequence pairwise. For numerical state quantities, directly calculate the absolute value of the difference between the two changes. For handover identifiers such as associated radio access point identifiers, calculate the absolute value of the difference between the two handover identifiers. For access state change events, calculate the absolute value of the difference between the two event counts or event type codes. Then, obtain the standard deviation of the changes in each state quantity in the historical access session: For channel utilization, received signal strength, signal-to-noise ratio, link rate, retransmission rate, and packet loss rate in dynamic link states, respectively, calculate the standard deviation of their changes in adjacent sampling times in the historical access session. Standard deviation; for access duration, uplink traffic, and downlink traffic in the access session state, the standard deviation of their changes in adjacent sampling times in the historical access session is calculated respectively; for handover flag and event count, their standard deviation in the historical access session is calculated respectively. The absolute value of the difference between each pair of state changes is divided by the standard deviation of the historical changes of the corresponding state quantity to obtain the normalized difference of the state quantity. If the standard deviation of the historical changes of a certain state quantity is zero or lower than a set threshold, a set small positive number is used as the denominator, or the normalized difference of the state quantity is set to zero. Then, the normalized differences of all state quantities under the same time offset are summed to obtain the state transition residual corresponding to the time offset. According to the time offset from smallest to largest, each time offset and its corresponding state transition residual are combined to form the state transition residual and stored in the residual table. Each record in the residual table includes the terminal access session identifier, access point identifier, time offset, state transition residual, and the type of state quantity involved in the calculation.

[0021] In S2, the process of forming access state characteristic quantities is as follows: For each terminal access session, the time offset corresponding to the smallest state transition residual is selected as the candidate response timing sequence. The frequency of occurrence of each candidate response timing sequence and its corresponding state transition residual are statistically analyzed. From the residual table, each time offset and its state transition residual corresponding to the terminal access session are extracted one by one according to the terminal access session identifier. For each terminal access session, the state transition residual with the smallest value is found in the state transition residual sequence corresponding to the session, and the time offset corresponding to the smallest state transition residual is selected as the candidate response timing sequence for that terminal access session. If multiple time offsets correspond to the same smallest state transition residual in the same terminal access session, the one with the smallest absolute value of the time offset is selected as the candidate response timing sequence. If multiple time offsets still exist, the one with the smallest time offset sequence number is selected as the candidate response. The timing sequence is then determined. The candidate response timing selection unit summarizes each candidate response timing according to the terminal access session identifier and access point identifier. It counts the number of times each candidate response timing occurs in all terminal access sessions, divides this number by the total number of terminal access sessions participating in the statistics, and obtains the frequency of occurrence of the candidate response timing. Simultaneously, it calculates the mean, minimum, maximum, and standard deviation of the state transition residuals corresponding to each candidate response timing. The candidate response timing, frequency of occurrence, and corresponding state transition residual statistics are stored in the candidate response timing table according to the terminal access session identifier and access point identifier. Each record in the candidate response timing table includes the candidate response timing value, frequency of occurrence, mean state transition residual, minimum state transition residual, maximum state transition residual, and the number of sessions participating in the statistics.

[0022] Using the upper boundary of the stable main interval of the historical distribution of state transition residuals as the residual determination criterion, the candidate response sequence that does not exceed the residual determination criterion and has the highest frequency is determined as the valid response sequence. All state transition residuals of historical access sessions are extracted from the residual table to form a historical distribution of state transition residuals. Then, the stable main interval of this historical distribution is determined according to the quantile method. For example, the interval between the 5th and 95th percentiles of the historical distribution of state transition residuals is taken as the stable main interval, or the interval after removing the maximum and minimum set proportions is taken as the stable main interval. The upper boundary value of this stable main interval is used as the residual determination criterion. The residual determination criterion can be set separately according to the access point identifier, service level, or terminal type, or it can be uniformly set according to all historical access sessions. Next, the valid response sequence is determined. The unit selects candidate response timings from the candidate response timing table whose mean state transition residuals do not exceed the residual judgment benchmark, forming a subset of candidate response timings. In the subset of candidate response timings, the candidate response timing with the highest frequency is selected as the valid response timing. If multiple candidate response timings have the same frequency and are all the highest, the one with the smallest mean state transition residual is selected as the valid response timing. If multiple are still present, the one with the smallest absolute value of time offset is selected as the valid response timing. The determined valid response timings are recorded according to the access point identifier, service level, or terminal type, and stored in the valid response timing table. Each record in the valid response timing table includes the valid response timing value, the corresponding frequency, the mean state transition residual, the residual judgment benchmark, and the applicable object identifier.

[0023] The dynamic link state changes are time-registered according to the valid response timing sequence, mapping quasi-static access attributes to the corresponding wireless access point and terminal access session, and associating them with the time-registered dynamic link state. The valid response timing sequence applicable to the current terminal access session is read from the valid response timing sequence table, and the link response sequence and session change sequence for that session are read from the state change sequence table. The time registration unit shifts the link response sequence relative to the session change sequence according to the valid response timing sequence, pairing each change in the link response sequence with the change at the corresponding moment in the session change sequence, thus completing the time registration of the dynamic link state changes. The registered dynamic link state changes are recorded according to the terminal access session identifier, access point identifier, and time on a unified time axis. Simultaneously... The time registration unit reads the quasi-static access attributes of the wireless access point associated with the current terminal access session from the quasi-static access attribute table, including access address, service area, wireless standard, working channel, frequency band, and resource capabilities. It then maps the quasi-static access attributes to all terminal access sessions under that access point according to the access point identifier. Next, it associates the mapped quasi-static access attributes with the time-registered dynamic link status according to the terminal access session identifier and the time on a unified time axis to form an associated status record. Each record in the associated status record contains the terminal access session identifier, access point identifier, time, quasi-static access attributes, the change in the time-registered dynamic link status, and the corresponding session change. The associated status records are stored in the associated status table according to the terminal access session identifier and the time.

[0024] The deviation of the dynamic link state of each terminal access session from the corresponding historical baseline is calculated, concatenated with quasi-static access attributes, and normalized to obtain access state feature quantities including resource capabilities, link quality changes, and access state. Associated state records are extracted from the associated state table according to the terminal access session identifier. For each terminal access session, the historical baseline for each state quantity in its dynamic link state is obtained. The historical baseline is determined as follows: within a set historical time window, the mean or median of each dynamic link state quantity of the terminal access session under the same access point, frequency band, and wireless standard is statistically analyzed, and the mean or median is used as the historical baseline for that state quantity. If the number of historical samples is less than a set threshold, the mean or median of all terminal access sessions under that access point on the same state quantity is used as the historical baseline. Then, the feature quantity construction unit calculates the deviation of each dynamic link state quantity from the corresponding historical baseline at the current moment. The deviation is calculated by dividing the difference between the current state quantity and the historical baseline by the standard deviation of that state quantity within the historical time window to obtain the standardized deviation. If the standard deviation is zero or lower than a set threshold, the deviation is considered normalized. If a small positive number is used as the denominator, or the standardized deviation is set to zero, then the standardized deviation of each dynamic link state quantity, the resource capability parameters in the quasi-static access attributes, and the access duration, uplink and downlink traffic, and access state change identifier in the access session state are concatenated to form the original feature vector. The resource capability parameters include the maximum number of associated terminals, the maximum throughput, the number of available radio links, the number of antennas, and the supported QoS level. The access state change identifier is assigned values ​​according to access point switching, link state mutation, and service quality anomaly. Finally, the original feature vector is normalized according to the value range of each feature quantity in the historical access session. The normalization method is: subtract its historical minimum value from each feature quantity and divide by the difference between the historical maximum value and the historical minimum value; if the historical maximum value and the historical minimum value are equal, then the feature quantity is set to zero. The normalized original feature vector is the access state feature quantity. The access state feature quantity is stored in the access state feature quantity table according to the terminal access session identifier, access point identifier, and time. Each record contains the terminal access session identifier, access point identifier, time, and the normalized value of each feature quantity.

[0025] S3: NWDAF registers the access state feature differences before and after historical access control actions according to the effective response time sequence, calculates the normalized response coefficient between the access state feature differences and the access control action amplitude, and normalizes the normalized response coefficient according to the access control action to obtain the mapping weight of each access state feature corresponding to each access control action.

[0026] In S3, the process of calculating the normalized response coefficient between the access state characteristic difference and the access control action amplitude is as follows: Acquire access migration actions, radio resource adjustment actions, and quality of service parameter correction actions corresponding to historical access control processes, and record the execution time, action object, and action amplitude of each access control action; collect three types of access control actions from historical access control processes. For access migration actions, record their execution time, action object, and action amplitude; the action object includes the source radio access point identifier, the target radio access point identifier, the terminal access session identifier, and the terminal identifier; the action amplitude is determined according to the migration type, for example, a value of 1 for migration within the same frequency band, a value of 2 for migration across frequency bands, and a value of 3 for migration across standards, or a value determined according to the bandwidth change when migration is accompanied by bandwidth adjustment. For radio resource adjustment actions, record their execution time, action object, and action amplitude; the action object includes the radio access point identifier, radio frequency identifier, channel number, frequency band, and terminal access session identifier; the action amplitude is determined according to the resource adjustment type, for example, a value of channel number change for channel switching. The absolute values ​​of the quantities are recorded. For bandwidth adjustment, the value is the change in bandwidth; for transmit power adjustment, the value is the change in power; and for resource block or time slot adjustment, the value is the change in the number of resource blocks or time slots. For service quality parameter correction actions, the execution time, action object, and action amplitude are recorded. The action object includes the terminal access session identifier, service flow identifier, service level, and service quality parameter type. The action amplitude is taken according to the service quality parameter type. For example, the latency upper limit adjustment is the change in the latency upper limit, the throughput lower limit adjustment is the change in the throughput lower limit, the packet loss rate upper limit adjustment is the change in the packet loss rate upper limit, and the jitter upper limit adjustment is the change in the jitter upper limit. All the above records are accompanied by a timestamp of the execution time and are stored in the control action record table according to the access control action type, action object, and execution time. Each record in the control action record table includes the action type, execution time, action object, action amplitude, and the corresponding terminal access session identifier and access point identifier.

[0027] Based on the execution time of the access control action, the response interval of the corresponding access state feature quantity is determined according to the effective response timing. Access state feature quantities before and after the execution of the access control action are extracted. Access control action records are read one by one from the control action record table according to the execution time, and the effective response timing applicable to the action object and terminal access session is read from the effective response timing table. The response interval determination unit uses the execution time of the access control action as a benchmark, determining the pre-action interval as the interval before the execution time and the interval after the execution time with a length equal to the effective response timing as the post-action interval. If the effective response timing is positive, the post-action interval is the interval between the execution time and the execution time plus the effective response timing; if the effective response timing is negative... If the value is specified, the interval after the action is the interval between the execution time plus the effective response time and the execution time. The length of the interval before the action is set to a set multiple of the effective response time, or set to a uniform time length. Then, the response interval determination unit extracts the access state feature quantity of the terminal access session in the interval before the action from the access state feature quantity table, and arranges them in chronological order to form the pre-action feature quantity sequence; it also extracts the access state feature quantity of the terminal access session in the interval after the action, and arranges them in chronological order to form the post-action feature quantity sequence. Both the pre-action feature quantity sequence and the post-action feature quantity sequence contain the normalized value of each feature quantity at each sampling time in the corresponding interval, and store them in association according to the terminal access session identifier, access point identifier, action type and execution time.

[0028] Registration is performed before and after the action according to the effective response time sequence, and the difference between the action and the action of each access state feature quantity is calculated. The pre-action feature quantity sequence and the post-action feature quantity sequence are obtained from the response interval determination unit. For the pre-action feature quantity sequence, the registration unit selects the pre-action reference time according to the effective response time sequence. The selection method for the pre-action reference time is: subtract the effective response time sequence from the access control action execution time to obtain the pre-action reference time. If the pre-action reference time is earlier than the start time of the pre-action interval, the start time of the pre-action interval is taken as the pre-action reference time. For the post-action feature quantity sequence, the registration unit selects the post-action reference time according to the effective response time sequence. The selection method for the post-action reference time is: add the effective response time sequence to the access control action execution time to obtain the post-action reference time. If the post-action reference time is later than the end time of the post-action interval, the end time of the post-action interval is taken as the post-action reference time. The time before and after an action is taken as the reference time after the action. Then, the pre- and post-action registration units search for the most recent valid sampled value in the pre-action feature sequence and use it as the pre-action feature. They also search for the most recent valid sampled value in the post-action feature sequence and use it as the post-action feature. If there is a missing state quantity corresponding to the pre-action or post-action reference time, the most recent valid sampled value before that time is used to fill it in. Then, the pre- and post-action registration units calculate the difference between the post-action feature and the pre-action feature for each access state feature, and obtain the pre- and post-action difference of each access state feature. The pre- and post-action difference retains the sign. A positive sign indicates that the feature increases after the action, and a negative sign indicates that the feature decreases after the action. The pre- and post-action differences of each access state feature are associated and stored according to the terminal access session identifier, access point identifier, action type, execution time, and feature type.

[0029] The difference between each access state feature is divided by the amplitude of the corresponding access control action, and then normalized according to the historical variation range of the access state feature to obtain the normalized response coefficient between the access state feature and the corresponding access control action. The difference between the action and the action of each access state feature is obtained from the pre- and post-registration unit, and the amplitude of the corresponding access control action is obtained from the control action record table. For each access state feature, the response coefficient calculation unit divides its difference between the action and the amplitude of the corresponding access control action to obtain the original response coefficient of the feature relative to the action. If the action amplitude is zero or lower than a set threshold, a set small positive number is used as the denominator, or the original response coefficient is set to zero. Then, the response coefficient calculation unit obtains the access state feature in the historical access session. The range of variation is determined as follows: within a set historical time window, the maximum and minimum values ​​of the feature quantity are statistically analyzed across all terminal access sessions. The difference between the maximum and minimum values ​​is taken as the historical range of variation of the feature quantity. If the historical range of variation is zero or lower than a set threshold, a set small positive number is used as the denominator, or the normalized response coefficient is set to zero. Then, the original response coefficient is divided by the historical range of variation of the feature quantity to obtain the normalized response coefficient. The normalized response coefficient retains the positive and negative signs. A positive sign indicates that the feature quantity increases under the corresponding access control action, and a negative sign indicates that the feature quantity decreases under the corresponding access control action. The normalized response coefficient is stored in association according to the terminal access session identifier, access point identifier, action type, action object, and feature quantity type.

[0030] Normalized response coefficients are recorded according to terminal access sessions, wireless access points, and access control actions, retaining the response direction to form a state response relationship corresponding to historical control actions. Normalized response coefficients are obtained from the response coefficient calculation unit and recorded according to three dimensions. The first dimension is based on terminal access sessions: using the terminal access session identifier as the primary key, the normalized response coefficients and response directions of each access state feature corresponding to each access control action under that session are recorded; the response direction is determined by the sign of the normalized response coefficient, with a positive sign indicating a positive response and a negative sign indicating a negative response. The second dimension is based on wireless access points: using the wireless access point identifier as the primary key, the normalized response coefficients and response directions of each access state feature corresponding to each access control action in all terminal access sessions under that access point are recorded, and the normalized response coefficients and response directions of the same... The three dimensions are: the mean, standard deviation, and sample size of the normalized response coefficients of the feature quantity under the same action; and the access control action record. The access control action type is used as the primary key to record the normalized response coefficients and response directions of each access state feature quantity corresponding to all terminal access sessions and all wireless access points under this action type. The mean, standard deviation, and sample size of the normalized response coefficients of the same feature quantity under this action type are also calculated. The records of the above three dimensions all include the terminal access session identifier, access point identifier, action type, action object, execution time, feature quantity type, normalized response coefficient, response direction, and sample size. They are stored in the state response relationship table according to the primary key and timestamp. The state response relationship table is indexed according to the terminal access session, wireless access point, and access control action to form the state response relationship corresponding to the historical control action.

[0031] In S3, the process of obtaining the mapping weights of each access control action corresponding to each access state feature quantity is as follows: The normalized response coefficients of each access state feature corresponding to the same access control action are extracted from the state response relationship. These are then summarized according to access migration actions, radio resource adjustment actions, and quality of service parameter correction actions. Using the access control action type as the primary key, the normalized response coefficients are read from the state response relationship table. For access migration actions, the normalized response coefficients of each access state feature corresponding to the access migration actions under all terminal access sessions and radio access points are read and categorized into subsets according to same-band migration, cross-band migration, and cross-system migration. For radio resource adjustment actions, the radio resource adjustment actions under all terminal access sessions and radio access points are read. The normalized response coefficients of each access state feature are assigned to subsets according to channel adjustment, bandwidth adjustment, transmit power adjustment, and resource block or time slot adjustment. For service quality parameter correction actions, the normalized response coefficients of each access state feature corresponding to the service quality parameter correction actions under all terminal access sessions and wireless access points are read and assigned to subsets according to latency parameter correction, throughput parameter correction, packet loss rate parameter correction, and jitter parameter correction. Each set is stored in association with action type, action subtype, access state feature type, and terminal access session identifier. Each set contains normalized response coefficients, response direction, sample number, and execution time.

[0032] For each access control action, the sum of the absolute values ​​of the normalized response coefficients corresponding to each access state feature is calculated. The proportion of the absolute value of the normalized response coefficient of each access state feature to the sum of absolute values ​​is used as the response contribution of the corresponding access state feature. A response contribution calculation unit is set up on the NWDAF side. For each access state feature under each access control action, the absolute values ​​of each coefficient in the set of normalized response coefficients of the feature are taken and summed to obtain the sum of absolute values ​​of the feature. When the number of samples is lower than a set threshold, the sum of absolute values ​​is set to zero. The sum of absolute values ​​of all access state features under the same access control action is summed to obtain the total sum of absolute values. When the total sum of absolute values ​​is zero or lower than the set threshold, the response contribution of each access state feature under the action is set to the reciprocal of the total sum of absolute values. Otherwise, the sum of absolute values ​​of each access state feature is divided by the total sum of absolute values ​​to obtain the response contribution of the feature under the access control action. The response contribution is stored in association with the access control action type, action subtype, and access state feature type.

[0033] The consistency of response direction for the same access state feature quantity across different historical access sessions is statistically analyzed, and the corresponding response contribution is weighted based on the consistency of response direction. A response direction consistency calculation unit is set up on the NWDAF side. For each access state feature quantity under each access control action, the response direction is extracted from the normalized response coefficient: a positive sign indicates a positive response, a negative sign indicates a negative response, and zero indicates no direction. The total number of samples, the number of positive response samples, and the number of negative response samples are counted. The larger of the positive and negative response sample counts is divided by the total number of samples to obtain the response direction consistency. When the total number of samples is zero or below a set threshold, the response direction consistency is set to zero. The response contribution of the feature quantity under that action is multiplied by the response direction consistency to obtain the weighted response contribution. The response direction is the direction with the larger sample count. When the number of positive and negative samples is equal, the response direction is recorded as no direction, and the weighted response contribution is set to zero. The weighted response contribution is stored in association with the access control action type, action subtype, access state feature quantity type, and response direction.

[0034] The weighted response contribution is normalized according to the access control action to obtain the mapping weight of each access state feature corresponding to the access control action. A mapping weight generation unit is set on the NWDAF side. For each access control action, the weighted response contribution of all access state features under the same action is summed to obtain the weighted contribution sum. When the weighted contribution sum is zero or lower than a set threshold, the mapping weight of each access state feature under the action is set to zero and the action is marked as temporarily unavailable. Otherwise, the weighted response contribution of each access state feature is divided by the weighted contribution sum to obtain the mapping weight of the feature under the action. The mapping weight takes the value from 0 to 1, and the sum of all mapping weights under the same action is 1. The mapping weight retains the corresponding response direction. When the response direction is directionless, the mapping weight is set to zero. The mapping weight is stored in association according to the access control action type, action subtype, access state feature type, response direction and applicable object identifier, and an index is built according to the access control action to form the mapping weight.

[0035] S4: Calculate the status response value of the candidate access control action based on the current access status characteristics and mapping weight, determine the control priority based on the current service quality status deviation, select access migration, radio resource adjustment or service quality parameter correction action, convert it into a control command that can be executed by the radio access point and send it out through the independent out-of-band management plane.

[0036] In S4, the process of calculating the state response value of the candidate access control action based on the current access state feature quantity and mapping weight is as follows: The system acquires the access state feature corresponding to the current terminal access session and performs a difference operation between each access state feature and its corresponding historical baseline to obtain the current access state offset. Using the current terminal access session identifier and access point identifier as indexes, it reads the access state feature at the current moment from the access state feature table. Simultaneously, it reads the historical baseline for each access state feature corresponding to the terminal access session from the access state feature table. The historical baseline is the mean or median of each access state feature under the same access point, frequency band, and wireless standard within a set historical time window for the terminal access session. When the number of historical samples is lower than a set threshold, the mean or median of the same access state feature for all terminal access sessions under the access point is taken as the historical baseline. The offset calculation unit subtracts the corresponding historical baseline from each access state feature at the current moment to obtain the current access state offset of the access state feature, retaining the positive or negative sign. If the historical baseline is missing, the current access state offset of the access state feature is set to zero. Each current access state offset is stored in association with the terminal access session identifier, access point identifier, feature type, and current moment.

[0037] Candidate access control actions are established for access migration, radio resource adjustment, and quality of service parameter correction. The current access state offset is multiplied by the mapping weight of the corresponding access control action, and then weighted and summed according to the access state features to obtain the predicted state response value for each candidate access control action. For each candidate access control action, the prediction response calculation unit reads the mapping weights of the action type and the corresponding action subtype under the mapping weight table, and then sums the weights according to the access state features. The current access state offset of a feature is multiplied by its mapping weight under the action to obtain the weighted offset of the feature under the action. If the mapping weight of the feature under the action is zero or does not exist in the mapping weight table, the weighted offset of the feature under the action is set to zero. Then, the weighted offsets of all access state features under the same candidate access control action are summed to obtain the predicted state response value corresponding to the candidate access control action, and the positive or negative sign is retained. The predicted state response values ​​of each candidate access control action are stored in association with the terminal access session identifier, access point identifier, action type, action subtype and current time.

[0038] The predicted state response values ​​of each candidate access control action are accumulated according to the effective response sequence, and the response direction is corrected according to the changing direction of latency, throughput, packet loss rate, and jitter in the current service quality state. The effective response sequence applicable to the current terminal access session and access point is read from the effective response sequence table, and the predicted values ​​of access state characteristics after the current time are read from the access state characteristic table, or the access control action records issued within the effective response sequence range are read from the control action record table. The response accumulation and correction unit determines the accumulation interval according to the effective response sequence, which is the interval between the current time and the current time plus the effective response sequence. If the effective response sequence is negative, the accumulation interval is the interval between the current time plus the effective response sequence and the current time. Within the accumulation interval, the same candidate access control action... The predicted state response values ​​at each sampling time are accumulated sequentially to obtain the accumulated predicted state response value for the candidate access control action. Then, the response accumulation and correction unit reads the latency, throughput, packet loss rate, and jitter at the current time from the service quality status and determines the direction of change of each indicator relative to the corresponding historical benchmark. The direction of increase of latency, packet loss rate, and jitter is recorded as negative, and the direction of increase of throughput is recorded as positive. The accumulated predicted state response value is corrected according to the direction of change of each indicator: if the sign of the accumulated predicted state response value is consistent with the direction of improvement of service quality status deviation, the original sign is retained; if they are inconsistent, the accumulated predicted state response value is set to zero or its opposite. The corrected predicted state response value is stored in association with the terminal access session identifier, access point identifier, action type, action subtype, accumulation interval, and correction direction.

[0039] The corrected predicted state response value is associated with the current access state offset to obtain the state response value corresponding to each candidate access control action. The corrected predicted state response value of each candidate access control action is obtained from the response accumulation and correction unit, and the current access state offset is obtained from the offset calculation unit. For each candidate access control action, the state response value generation unit associates the corrected predicted state response value with the current access state offset of each access state feature involved in the calculation under that action, item by item according to the feature type, to form the state response vector of that candidate access control action. Each component in the state response vector includes the feature type, the current access state offset, the mapping weight of the feature under that action, the weighted offset, and the corrected predicted state response value. Then, the state response vector is associated and stored according to the terminal access session identifier, access point identifier, action type, action subtype, and current time to obtain the state response value corresponding to each candidate access control action.

[0040] In S4, the process of converting control commands into executable commands for the wireless access point and distributing them via the independent out-of-band management plane is as follows: The difference between the current service quality status and the corresponding service quality target status is calculated and normalized according to the allowable fluctuation range of the service quality indicators and the service quality status to obtain the service quality status deviation. Using the current terminal access session identifier and service flow identifier as indexes, the latency, throughput, packet loss rate, and jitter at the current moment are read from the service quality status. The upper limit of latency, lower limit of throughput, upper limit of packet loss rate, and upper limit of jitter for the corresponding service level of the terminal access session are read from the service quality target status. The service quality deviation calculation unit calculates the difference for each indicator: the latency difference is the current latency minus the latency upper limit; the throughput difference is the throughput lower limit minus the current throughput; the packet loss rate difference is the current packet loss rate minus the packet loss upper limit; and the jitter difference... The value is the current jitter minus the jitter limit; each difference retains a positive or negative sign, with a positive value indicating that the indicator is worse than the target state and a negative value indicating that the indicator is better than the target state. Then, the allowable fluctuation range of each service quality indicator is read from the service level configuration. The allowable fluctuation range is set separately according to the service level or according to the indicator type. The difference of each indicator is divided by the corresponding allowable fluctuation range to obtain the normalized deviation of the indicator. When the allowable fluctuation range is zero or lower than the set threshold, a small positive number is used as the denominator, or the normalized deviation of the indicator is set to zero. The normalized deviations of each indicator are combined into a service quality status deviation vector according to the indicator type and stored in association with the terminal access session identifier, service flow identifier, service level and current time.

[0041] The status response value of each candidate access control action is multiplied by the service quality status deviation, and constraints are adjusted according to the resource occupancy level and historical execution success rate corresponding to the candidate access control action to obtain the control priority of each candidate access control action. The status response value of each candidate access control action is obtained from the status response value generation unit, and the service quality status deviation vector is obtained from the service quality deviation calculation unit. For each candidate access control action, the control priority calculation unit multiplies each feature component in the status response value of the action with the corresponding index component in the service quality status deviation vector, and sums the products to obtain the initial priority of the candidate access control action. Feature components without corresponding index components in the status response value do not participate in the product. Then, the candidate access control action is read from the resource occupancy level table. The resource occupancy level corresponding to the control action is set separately according to the action type and action subtype, and the value is the normalized proportion of the radio frequency resources, channel resources, bandwidth resources or time slot resources occupied by the action. The historical execution success rate of the candidate access control action within the set historical time window is read from the historical execution success rate table. The historical execution success rate is the ratio of the number of successful executions of the action to the total number of executions. The initial priority is multiplied by the complement of the resource occupancy level, and then multiplied by the historical execution success rate to obtain the control priority of the candidate access control action. When the resource occupancy level is zero or the historical execution success rate is missing, the set default value is used in the calculation. The control priority of each candidate access control action is stored in association with the terminal access session identifier, access point identifier, action type, action subtype and current time.

[0042] The access migration, radio resource adjustment, and service quality parameter correction actions are prioritized according to control priority. The access control action with the highest control priority and whose corresponding state response direction can reduce the service quality state deviation is selected as the target control action. The control priority of each candidate access control action is obtained from the control priority calculation unit, and the access migration action, radio resource adjustment action, and service quality parameter correction action and their respective action subtypes are sorted from high to low control priority. Then, the target action selection unit checks each candidate access control action in turn according to the sorting result: the sign of each feature component in the state response value of the candidate access control action is read from the state response value generation unit, and the sign of each index component in the service quality state deviation vector is read from the service quality deviation calculation unit; if the candidate access control action is in service quality state deviation, the target action is selected. If the status response value corresponding to an indicator with a positive quality status deviation is positive, or the status response value corresponding to an indicator with a negative service quality status deviation is negative, then the status response direction of the candidate access control action is determined to reduce the service quality status deviation, and it is selected as the target control action. If the candidate access control action ranked first does not meet this condition, then the next candidate access control action is checked in order of ranking until the candidate access control action that meets the condition and has the highest control priority is selected as the target control action. If all candidate access control actions do not meet the condition, then the candidate access control action with the highest control priority is selected as the target control action, or the current access control state remains unchanged. The target control action is stored in association with the terminal access session identifier, access point identifier, action type, action subtype, control priority, and current time.

[0043] Based on the action type of the target control action, it is converted into a terminal migration, channel or resource adjustment, transmission parameter or service quality parameter correction instruction executable by the corresponding radio access point, and the control instruction is sent to the corresponding radio access point for execution through an independent out-of-band management plane. The target control action, its action type, action subtype, action object, and action magnitude are obtained from the target action selection unit. When the target control action is an access migration action, the control instruction generation unit generates a terminal migration instruction, the instruction content of which includes the source radio access point identifier, target radio access point identifier, terminal access session identifier, terminal identifier, target frequency band, target channel, and migration execution time. When the target control action is a radio resource adjustment action, the control instruction generation unit generates a channel or resource adjustment instruction, the instruction content of which includes the radio access point identifier, radio frequency identifier, target channel number, target bandwidth, target transmit power, target resource block or time slot allocation, and adjustment execution time. When performing a service quality parameter correction action, the control command generation unit generates a transmission parameter or service quality parameter correction command. The command content includes the terminal access session identifier, service flow identifier, target latency upper limit, target throughput lower limit, target packet loss rate upper limit, target jitter upper limit, and correction execution time. The control command generation unit encapsulates the above commands into an out-of-band management message according to the message format of the independent out-of-band management plane. The message header includes the access point identifier, command type, command sequence number, and check field, and the message payload contains the command content. Then, the out-of-band management message is sent to the corresponding radio access point through the independent out-of-band management plane. After receiving the message, the radio access point parses it and performs terminal migration, channel or resource adjustment, transmission parameter or service quality parameter correction according to the command content. The execution result is then sent back to the NWDAF through the independent out-of-band management plane. The control command and execution result are stored in association according to the terminal access session identifier, access point identifier, command type, command sequence number, and execution time.

[0044] S5: Collect the dynamic link status, access session status and service quality status after the execution of control commands, calculate the mapping residual between the actual status response and the expected status response according to the effective response sequence, correct the corresponding mapping weights according to the mapping residuals, and recalculate the control priority of candidate access control actions based on the corrected mapping weights and access status features.

[0045] In S5, the process of recalculating the control priority of candidate access control actions based on the corrected mapping weights and access state features is as follows: After the target control action is executed, the dynamic link state, access session state, and quality of service state are realigned according to the sampling time of each state data, using the terminal access session identifier as an index, and the actual response interval is determined based on the effective response timing. The execution time, action type, action subtype, and action object of the target control action are read from the control command execution result. Using the terminal access session identifier as an index, the dynamic link state, access session state, and quality of service state after the target control action is executed are re-collected from the WLAN side. The post-execution alignment unit maps the dynamic link state, access session state, and quality of service state to a unified time axis according to the sampling time of each state data. The mapping method is consistent with the access state timing set. Consistent timing: For periodic data, linear interpolation or hold-at-the-moment interpolation is used; for event-based data, the nearest neighbor method is used; for missing state variables, the most recent valid sample value is used to fill in the missing state variables. Then, the valid response timing applicable to the terminal access session and access point is read from the valid response timing table. Based on the execution time of the target control action, the interval after the execution time and with a length equal to the valid response timing is determined as the actual response interval. If the valid response timing is negative, the actual response interval is the interval between the execution time and the valid response timing. The start time, end time, and sampling time list of the actual response interval are stored in association with the terminal access session identifier, access point identifier, action type, and execution time.

[0046] The actual change in access state characteristics within the actual response interval is calculated and registered with the expected state response obtained before the execution of the target control action based on the mapping weight. The absolute value of the difference between the actual state response and the expected state response is calculated and the direction of the difference is retained to obtain the mapping residual. The actual response interval is obtained from the post-execution alignment unit, and the access state characteristics of the terminal access session at each sampling time within the actual response interval are read from the access state characteristic table. The mapping residual calculation unit selects the actual response reference time within the actual response interval according to the same registration method as the difference before and after the action, that is, the target control action execution time plus the effective response time sequence, and finds the effective sampled value closest to the actual response reference time as the actual characteristic after the action. At the same time, the offset calculation unit and the predicted response before the execution of the target control action are read. The calculation unit obtains the pre-action feature quantities and expected state responses. It subtracts the pre-action feature quantities from the actual feature quantities after the action to obtain the actual changes in each access state feature quantity within the actual response interval. Then, it registers the actual changes in each access state feature quantity with the expected state responses obtained before the execution of the target control action according to the mapping weights, item by item according to the feature quantity type. It calculates the absolute value of the difference between the actual change and the expected state response for each feature quantity and retains the direction of the difference. The direction of the difference is determined by the sign of the actual change minus the expected state response. A positive sign indicates that the actual response is greater than the expected response, and a negative sign indicates that the actual response is less than the expected response. The absolute value of the difference and the direction of the difference of each feature quantity form a mapping residual vector, which is stored in association with the terminal access session identifier, access point identifier, action type, action subtype, and execution time.

[0047] The mapping residual is calculated as a percentage of the expected state response. The mapping weight is then adjusted based on the residual direction and the consistency of historical responses. The mapping residual vector is obtained from the mapping residual calculation unit, and the corresponding expected state response is obtained from the predicted response calculation unit. For each access state feature, the mapping weight adjustment unit divides the absolute value of the feature's mapping residual by the absolute value of its expected state response to obtain the residual percentage. When the expected state response is zero or below a set threshold, a small positive number is used as the denominator, or the residual percentage is set to zero. Then, the historical response of the feature under this action type is read from the response direction consistency calculation unit. The degree of consistency is determined by combining the residual direction with the historical response direction: when the residual direction is consistent with the historical response direction, the correction amount is the product of the residual ratio and the degree of consistency with the historical response; when the residual direction is inconsistent with the historical response direction, the correction amount is the negative of the product of the residual ratio and the degree of consistency with the historical response. The correction amount is multiplied by the original mapping weight of the feature quantity under the action to obtain the corrected mapping weight; when the corrected mapping weight is lower than zero, it is set to zero, and when it is higher than the set upper limit, the upper limit is taken. The corrected mapping weight is stored in association with the access control action type, action subtype, access status feature quantity type, response direction and applicable object identifier, and the corresponding record in the mapping weight table is updated.

[0048] The corrected mapping weights are normalized and weighted with the current access state features to obtain the updated candidate access control action state response values. The corrected mapping weights are obtained from the mapping weight correction unit, and for each access control action, the corrected mapping weights of all access state features under the same action are summed to obtain the total corrected weights. If the total corrected weights are zero or below a set threshold, the corrected mapping weights of each access state feature under that action are set to zero, and the action is marked as temporarily unavailable. Otherwise, the corrected mapping weight of each access state feature is divided by the total corrected weights to obtain the normalized corrected mapping weights. Then, the current access state offset is obtained from the offset calculation unit, and each... The current access state offset of the access state feature is multiplied by the normalized correction mapping weight of the feature under the action to obtain the weighted offset. The weighted offsets of all access state features under the same action are summed to obtain the predicted state response value after the action is updated. The updated predicted state response value is accumulated according to the effective response time sequence in the same accumulation and correction method as the predicted response calculation unit. The response direction is corrected according to the change direction of latency, throughput, packet loss rate and jitter in the current service quality state to obtain the updated candidate access control action state response value. The updated state response value is stored in association with the terminal access session identifier, access point identifier, action type, action subtype and current time.

[0049] The control priority of candidate access control actions is recalculated based on the current service quality status deviation. The target access control action is determined according to the updated control priority, and the corresponding control command is issued via an independent out-of-band management plane. The updated candidate access control action status response value is obtained from the weight normalization and response update unit, and the current service quality status deviation vector is obtained from the service quality deviation calculation unit. For each candidate access control action, the priority update and command issuance unit multiplies each feature component in the updated status response value with the corresponding index component in the service quality status deviation vector and sums them to obtain the initial priority. Then, the resource occupancy level and historical execution success rate corresponding to the action are read from the resource occupancy level table and the historical execution success rate table. The initial priority is multiplied by the complement of the resource occupancy level, and then multiplied by the historical execution success rate to obtain the updated control priority. According to the updated control priority, access migration, radio resource adjustment, and service quality parameter correction actions and their sub-types are sorted from high to low. Each candidate access control action is checked sequentially according to the same conditions as the target action selection unit. The candidate access control action with the highest control priority and the state response direction that can reduce the service quality state deviation is selected as the target access control action. According to the same instruction generation method as the control instruction generation unit, the target access control action is converted into a terminal migration, channel or resource adjustment, transmission parameter or service quality parameter correction instruction that can be executed by the corresponding radio access point. It is encapsulated into an out-of-band management message and sent to the corresponding radio access point for execution through an independent out-of-band management plane. The updated control priority, target access control action and sent instructions are stored in association according to the terminal access session identifier, access point identifier, action type, action sub-type and execution time.

[0050] Example 2: Please refer to Figure 2 As shown, a WLAN access state open and closed loop control system for 6G includes: Access Status Module: Collects the access, link, and service quality status of wireless access points and terminals, performs session-level time alignment, and forms a WLAN access status time sequence set; Timing Opening Module: Sends the WLAN access state timing set to NWDAF through an independent out-of-band management plane, determines the valid response timing, and forms access state characteristic quantities; Response mapping module: Establishes mapping weights between historical control actions and changes in access status; Action orchestration module: Determines control priority based on current access status features and mapping weights, and generates corresponding access control commands; Response correction module: Corrects the mapping weights based on the mapping residual between the actual state response and the expected state response after control, and updates the control priority.

[0051] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0052] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A WLAN access state opening and closed loop control method for 6G, characterized in that, Includes the following steps: Collect the quasi-static access attributes, dynamic link status, and access session status and service quality status of the corresponding terminals of each wireless access point on the wireless LAN side. Use the terminal access session as an index to align the status data in time and form a WLAN access status time series set. The WLAN access state timing set is sent to the network data analysis function NWDAF through an independent out-of-band management plane. The state transition residuals of dynamic link state changes and access session state changes at different time offsets are calculated. The effective response timing is determined by the time offset corresponding to the minimum residual. Based on this, the quasi-static access attributes and dynamic link states are fused to form access state feature quantities. The NWDAF registers the access state feature differences before and after historical access control actions according to the effective response time sequence, calculates the normalized response coefficient between the access state feature differences and the access control action amplitude, and normalizes the normalized response coefficient according to the access control action to obtain the mapping weight of each access state feature corresponding to each access control action. Calculate the status response value of candidate access control actions based on the current access status characteristics and mapping weights, determine the control priority based on the current service quality status deviation, select access migration, radio resource adjustment or service quality parameter correction actions, convert them into control instructions that can be executed by the radio access point and send them out through the independent out-of-band management plane. The system collects the dynamic link status, access session status, and service quality status after the control command is executed. It calculates the mapping residual between the actual status response and the expected status response according to the effective response sequence. It corrects the corresponding mapping weights based on the mapping residuals and recalculates the control priority of the candidate access control actions based on the corrected mapping weights and access status features.

2. The WLAN access state open and closed loop control method for 6G of claim 1, wherein, The process of forming a WLAN access state timing set is as follows: Collect the access address, service area, wireless standard, working channel, frequency band and resource capabilities of each wireless access point to form quasi-static access attributes; Collect channel utilization, received signal strength, signal-to-noise ratio, link rate, retransmission rate, and packet loss rate to form a dynamic link status; The data collection terminal associates wireless access point, access duration, uplink and downlink traffic, and access status changes to form the access session status. The system collects terminal latency, throughput, packet loss rate, jitter, and service level data to form a service quality status. Using the terminal access session identifier as a unified index, the quasi-static access attributes, dynamic link status, access session status and service quality status corresponding to the same terminal access session are associated according to the sampling time. Data from different sampling periods are resampled, and the missing status is filled in with the most recent valid sample value to form an access status record under a unified time axis. Each access status record is arranged according to the terminal access session and wireless access point, and the status change positions corresponding to access point switching, link status change and service quality anomaly are marked to form a WLAN access status time series set containing static attributes, dynamic links, session changes and service quality changes.

3. The WLAN access state open and closed loop control method for 6G of claim 2, wherein, The process of calculating the state transition residuals of dynamic link state changes and access session state changes at different time offsets is as follows: The WLAN access state time series is divided into dynamic link state sequence and access session state sequence according to the terminal access session. The state change amount between adjacent sampling times is calculated, and the dynamic link state change amount is used as the link response sequence and the access session state change amount is used as the session change sequence. The state change sequence is transmitted to the network data analysis function NWDAF through an independent out-of-band management plane. The time difference distribution between dynamic link state changes and access session state changes in historical access sessions is statistically analyzed. The middle main interval of the time difference distribution is used as the candidate time offset interval. The time offset is set sequentially in the candidate time offset interval according to the sampling interval. The link response sequence and the session change sequence are then registered in a sliding manner. Calculate the absolute value of the difference between the corresponding state change quantities between the link response sequence and the session change sequence at each time offset, and normalize them according to the standard deviation of the change quantities of the corresponding state quantities in the historical access sessions. Sum the normalized differences of each state quantity to obtain the state transition residual corresponding to each time offset.

4. The WLAN access state open and closed loop control method for 6G of claim 3, wherein, The process of forming access state characteristics is as follows: For each terminal access session, the time offset corresponding to the minimum state transition residual is selected as the candidate response timing, and the occurrence frequency and corresponding state transition residual of each candidate response timing are statistically analyzed. Using the upper boundary of the stable main interval of the historical distribution of state transition residuals as the residual judgment criterion, the candidate response time series that does not exceed the residual judgment criterion and has the highest frequency of occurrence is determined as the effective response time series; The dynamic link state changes are time-registered according to the effective response timing, the quasi-static access attributes are mapped to the corresponding wireless access points and terminal access sessions, and associated with the time-registered dynamic link state. The deviation of the dynamic link state of each terminal access session from the corresponding historical baseline is calculated, concatenated with the quasi-static access attributes and normalized to obtain access state feature quantities that include resource capabilities, link quality changes and access state.

5. The WLAN access state open and closed-loop control method for 6G according to claim 4, characterized in that, The process of calculating the normalized response coefficient between the access state characteristic difference and the access control action amplitude is as follows: Acquire access migration actions, radio resource adjustment actions, and quality of service parameter correction actions corresponding to the historical access control process, and record the execution time, action object, and action magnitude of each access control action; Based on the execution time of the access control action, the response interval of the corresponding access state feature quantity is determined according to the effective response timing, and the access state feature quantity before and after the execution of the access control action is extracted respectively. Perform pre- and post-registration according to the effective response sequence, and calculate the pre- and post-action differences of each access state feature quantity; Divide the difference of each access state feature by the amplitude of the corresponding access control action, and normalize it according to the historical variation range of the access state feature to obtain the normalized response coefficient between the access state feature and the corresponding access control action. Record the normalized response coefficients for each terminal access session, wireless access point, and access control action, and retain the response direction to form the state response relationship corresponding to the historical control actions.

6. The WLAN access state open and closed-loop control method for 6G according to claim 5, characterized in that, The process of obtaining the mapping weights of each access control action corresponding to each access state feature is as follows: The normalized response coefficients of each access state feature quantity corresponding to the same access control action are extracted from the state response relationship, and summarized according to access migration action, radio resource adjustment action and service quality parameter correction action respectively. For each access control action, the sum of the absolute values ​​of the normalized response coefficients corresponding to each access state feature is calculated, and the proportion of the absolute value of the normalized response coefficient of each access state feature to the sum of the absolute values ​​is used as the response contribution of the corresponding access state feature. The consistency of response direction for the same access state feature quantity in different historical access sessions is statistically analyzed, and the contribution of the corresponding response is weighted based on the consistency of response direction. The weighted response contribution is normalized according to the access control action to obtain the mapping weight of each access state feature corresponding to the access control action.

7. A WLAN access state open and closed-loop control method for 6G as described in claim 6, characterized in that, The process of calculating the state response value of candidate access control actions based on the current access state characteristics and mapping weights is as follows: Obtain the access state feature quantity corresponding to the current terminal access session, and then differ each access state feature quantity from the corresponding historical benchmark to obtain the current access state offset. Candidate access control actions are established for access migration, radio resource adjustment and service quality parameter correction. The current access state offset is multiplied by the mapping weight of the corresponding access control action, and the weighted sum is performed according to the access state feature to obtain the predicted state response value corresponding to each candidate access control action. The predicted status response values ​​for each candidate access control action are accumulated according to the effective response sequence, and the response direction is corrected according to the change direction of latency, throughput, packet loss rate and jitter in the current service quality status. The corrected predicted state response value is correlated with the current access state offset to obtain the state response value corresponding to each candidate access control action.

8. A WLAN access state open and closed-loop control method for 6G according to claim 7, characterized in that, The process of converting control commands into executable commands for the wireless access point and distributing them through an independent out-of-band management plane is as follows: Calculate the difference between the current service quality status and the corresponding service quality target status, and normalize it according to the allowable fluctuation range of the business level and service quality indicators to obtain the service quality status deviation. The status response value of each candidate access control action is multiplied by the service quality status deviation, and the constraints are adjusted according to the resource consumption level and historical execution success rate of the candidate access control action to obtain the control priority of each candidate access control action. The access migration, radio resource adjustment, and service quality parameter correction actions are sorted according to control priority. The access control action with the highest control priority and whose corresponding state response direction can reduce the service quality state deviation is selected as the target control action. Based on the action type of the target control action, it is converted into a terminal migration, channel or resource adjustment, transmission parameter or quality of service parameter correction instruction that can be executed by the corresponding wireless access point, and the control instruction is sent to the corresponding wireless access point for execution through an independent out-of-band management plane.

9. A WLAN access state open and closed-loop control method for 6G as described in claim 8, characterized in that, The process of recalculating the control priority of candidate access control actions based on the corrected mapping weights and access state features is as follows: After the target control action is executed, the dynamic link state, access session state and service quality state are re-aligned according to the sampling time of each state data, using the terminal access session identifier as an index, and the actual response interval is determined based on the effective response sequence. Calculate the actual change in the access state characteristic quantity within the actual response interval, register it with the expected state response obtained before executing the target control action based on the mapping weight, calculate the absolute value of the difference between the actual state response and the expected state response and retain the direction of the difference to obtain the mapping residual. Calculate the ratio of the mapping residual to the expected state response residual, and adjust the corresponding mapping weights based on the residual direction and the consistency of historical responses. The corrected mapping weights are normalized and weighted with the current access state features to obtain the updated candidate access control action state response values. The control priority of candidate access control actions is recalculated based on the current service quality status deviation. The target access control action is determined based on the updated control priority, and the corresponding control command is issued through an independent out-of-band management plane.

10. A WLAN access state open and closed-loop control system for 6G, applied to the method described in any one of claims 1-9, characterized in that, include: Access Status Module: Collects the access, link, and service quality status of wireless access points and terminals, performs session-level time alignment, and forms a WLAN access status time sequence set; Timing Opening Module: Sends the WLAN access state timing set to NWDAF through an independent out-of-band management plane, determines the valid response timing, and forms access state characteristic quantities; Response mapping module: Establishes mapping weights between historical control actions and changes in access status; Action orchestration module: Determines control priority based on current access status features and mapping weights, and generates corresponding access control commands; Response correction module: Corrects the mapping weights based on the mapping residual between the actual state response and the expected state response after control, and updates the control priority.