Intelligent self-optimizing mobile communication quality assurance method and system

Abstract

The application discloses a kind of intelligent self-optimizing mobile communication quality guarantee method and system.According to the QoS guarantee package of user subscription, when user is online and uses the application or service subscribed, the user QoS data of last period is obtained;After the user QoS data of last period is preprocessed and integrated calculation, input QoS quality difference comprehensive quantity prediction model, the QoS quality difference comprehensive quantity of next period is obtained by prediction;Real-time acquisition of the user QoS data of next period, calculate real-time user QoS quality difference comprehensive quantity;Real-time QoS quality difference comprehensive quantity is compared with the QoS quality difference comprehensive quantity predicted, whether the user QoS is judged to deteriorate;According to the deterioration of user QoS, combined with RAN side and BN side data, generate QoS adjustment strategy, issue to service domain execution.The application can dynamically perceive user communication quality parameter in real time, and when communication quality parameter deteriorates, trigger the guarantee of user communication quality.

Description

Technical Field

[0001] This invention relates to the fields of ICT (Information & Communication Technology) and 5G / 6G (5th / 6th Generation) mobile communication intelligent networks, and particularly to an intelligent self-optimizing mobile communication quality assurance method and system. Background Technology

[0002] With the rapid development and application of new technologies such as virtual reality, digital twins, autonomous driving, artificial intelligence, and metaverse, human-computer interaction has become closer, and user needs are constantly evolving.

[0003] The rich business forms and diverse business scenarios brought about by new technologies place higher and more stringent demands on the quality of communication networks. In particular, some vertical industries, such as power, industrial control, and telemedicine, require higher bandwidth, lower latency, less jitter and packet loss rate. They need controllable communication network channels and bearers, guaranteed network communication quality, and the ability for the system to automatically detect and optimize communication quality when communication network parameters such as bandwidth, latency, jitter, packet loss rate, and retransmission rate deteriorate, thus ensuring the user's communication experience.

[0004] The existing Quality of Service (QoS) guarantee mechanism for mobile communication network users is rather rigid. Even if users subscribe to a QoS guarantee package, automatic protection and optimization are not supported when QoS deteriorates. It can only guarantee QoS through a "best-effort" approach, resulting in a poor user experience, low satisfaction, and an inability to increase user loyalty and purchasing motivation. Furthermore, the existing QoS guarantee mechanism cannot meet the higher quality communication guarantee requirements of the industrial control industry. Summary of the Invention

[0005] Therefore, mobile communication networks urgently need a new solution and architecture that can automatically detect, intelligently analyze, and proactively generate specific QoS control strategies based on user service scenarios when user QoS deteriorates. This closed-loop dynamic optimization of communication quality ensures user QoS quality, thereby improving user experience and satisfaction, and meeting the high-quality QoS assurance requirements of the industrial control industry.

[0006] To improve and optimize the existing mobile communication network's "best-effort" mechanism for ensuring user communication quality, this invention proposes an intelligent self-optimizing mobile communication quality assurance method and system. This system dynamically senses user communication quality parameters in real time and triggers the assurance of user communication quality when these parameters deteriorate, thus maximizing the protection of user communication quality.

[0007] The technical solution adopted in this invention is as follows:

[0008] On one hand, the present invention provides an intelligent self-optimizing mobile communication quality assurance method, comprising:

[0009] Based on the QoS guarantee package or guarantee policy signed by the user, when the user goes online and uses the signed application or service, the user's QoS data from the previous time period is obtained, and RAN side data and BN side data are collected at the same time.

[0010] Preprocess the user QoS data from the previous time period to obtain the preprocessed user QoS data from the previous time period;

[0011] The preprocessed user QoS data from the previous time period is integrated and calculated to obtain the comprehensive QoS quality difference of the previous time period.

[0012] Input the overall QoS quality of the user in the previous period into the overall QoS quality of the prediction model to predict the overall QoS quality of the next period.

[0013] When the next time period arrives, the user QoS data for the next time period is collected in real time, and the comprehensive QoS quality difference of the user for the next time period is calculated in real time.

[0014] The real-time QoS quality difference of the next time period is compared with the predicted QoS quality difference of the next time period to determine whether the user's QoS has deteriorated.

[0015] Based on the deterioration of user QoS, and combining the collected RAN-side data and BN-side data, a QoS adjustment policy is generated based on the locally pre-configured expert rule tree or experience base, and the QoS adjustment policy is distributed to various devices in the service domain for execution.

[0016] Furthermore, the user QoS data includes at least bandwidth, latency, and packet loss rate, and the overall user QoS quality deterioration is calculated according to the following formula:

[0017] QV=WB×W1+DELAY×W2+PkgLosRat×W3

[0018] Where QV is the overall QoS quality of the user; WB is the bandwidth; DELAY is the latency; PkgLosRat is the packet loss rate; W1, W2, and W3 are the weights of bandwidth, latency, and packet loss rate, respectively.

[0019] Furthermore, the QoS quality deterioration comprehensive prediction model is trained using the following method:

[0020] Obtain the historical user QoS dataset DS1. Each data set in dataset DS1 includes bandwidth, latency, and packet loss rate.

[0021] Preprocess dataset DS1 to obtain dataset DS2;

[0022] The data in each set of dataset DS2 are integrated and calculated to obtain the comprehensive dataset of user QoS quality difference DS3.

[0023] The DS3 dataset was divided into a training set and a test set. The Prophet model was trained using the training set, and the training effect of the model was tested using the test set.

[0024] Furthermore, the step of comparing the real-time QoS quality deterioration of the next time period with the predicted QoS quality deterioration of the next time period to determine whether the user's QoS has deteriorated includes:

[0025] Determine whether the real-time QoS quality difference index for the next time period exceeds the predicted range of the QoS quality difference index for the next time period.

[0026] If the real-time QoS quality difference index for the next time period is within the range of the predicted QoS quality difference index for the next time period, then the user's QoS is considered normal.

[0027] If the real-time QoS quality deterioration value for the next time period exceeds the predicted QoS quality deterioration value for the next time period, then the user's QoS is considered to have deteriorated.

[0028] Furthermore, the generated QoS adjustment policy decision includes:

[0029] Adjustment policies for user use of radio resources implemented on the RAN side include, but are not limited to, QoS level adjustment, channel preemption method adjustment, and non-value user radio resource occupation method adjustment;

[0030] Adjustment strategies for user use of transport network resources implemented on the BN side include, but are not limited to, improving the QoS level of data transmission and switching the transport channel to a higher quality channel;

[0031] Adjustment strategies for user data flow forwarding quality implemented on the CN side include, but are not limited to, improving the QoS level of the data flow, switching the data flow to a specific high-performance transmission network card, and adjusting the data flow offloading method.

[0032] Furthermore, the distribution of the QoS adjustment policy to each device in the service domain for execution includes any of the following methods:

[0033] The generated QoS adjustment policies executed on the RAN side, BN side, or CN side are sent to the SMF via the PCF of the service domain. The SMF then sends the QoS adjustment policies executed on the RAN side to each device on the RAN side for execution via the AMF. The SMF then sends the QoS adjustment policies executed on the BN side or CN side to each device on the BN side or CN side for execution via the UPF.

[0034] The generated QoS adjustment policies, which are executed on the RAN side, BN side, or CN side, are sent to the NEF capability open platform in the service domain. The NEF capability open platform then sends them to the SMF via the PCF. The SMF then sends the QoS adjustment policies executed on the RAN side to each device on the RAN side via the AMF. The SMF then sends the QoS adjustment policies executed on the BN side or CN side to each device on the BN side or CN side via the UPF.

[0035] The generated QoS adjustment policies, which are executed on the RAN side, BN side, or CN side, are distributed to the corresponding RAN management center, BN management center, or CN management center, respectively. The RAN management center, BN management center, or CN management center then maps the corresponding QoS adjustment policies to the operation instructions for the devices in their respective domains and distributes the operation instructions to the devices in their respective domains for execution.

[0036] On the other hand, the present invention provides an intelligent self-optimizing mobile communication quality assurance system, wherein the system is deployed in an Artificial Intelligence Sensing and Control Center (AISACC), which is superimposed on the existing mobile communication network framework, and the system includes:

[0037] The data acquisition module is used to acquire the user's QoS data from the previous time period when the user goes online and uses the contracted application or service, based on the QoS guarantee package or guarantee policy signed by the user, and to collect the user's QoS data for the next time period in real time when the next time period arrives. At the same time, it collects RAN side data and BN side data.

[0038] The data preprocessing module is used to preprocess the user QoS data of the previous time period and the user QoS data of the next time period collected in real time, respectively, to obtain the preprocessed user QoS data of the previous time period and the real-time user QoS data of the next time period.

[0039] The data integration and calculation module is used to integrate and calculate the preprocessed user QoS data of the previous time period and the real-time user QoS data of the next time period to obtain the comprehensive user QoS quality difference of the previous time period and the comprehensive user QoS quality difference of the next time period.

[0040] The QoS quality poor comprehensive quantity prediction module is used to input the user QoS quality poor comprehensive quantity of the previous time period into the QoS quality poor comprehensive quantity prediction model to predict the QoS quality poor comprehensive quantity of the next time period.

[0041] The data analysis module is used to compare the real-time QoS quality difference of the next time period with the predicted QoS quality difference of the next time period to determine whether the user's QoS has deteriorated.

[0042] The strategy decision module is used to generate QoS adjustment policies based on the deterioration of user QoS, combined with the collected RAN-side data and BN-side data, and based on the locally preset expert rule tree or experience base. The QoS adjustment policies are then distributed to various devices in the service domain for execution.

[0043] Optionally, the data acquisition module includes a UPF and a first data acquisition module disposed in the CN, a second data acquisition module disposed in the BN, and a third data acquisition module disposed in the RAN;

[0044] The UPF is used to detect user QoS data when a user goes online and uses a contracted application or service, according to the received data reporting instruction, and send the detected user QoS data to the first data acquisition module.

[0045] The first data acquisition module is used to report the received user QoS data;

[0046] The second data acquisition module is used to acquire data from the BN side;

[0047] The third data acquisition module is used to acquire RAN-side data.

[0048] Optionally, the data acquisition module includes a UPF and a first data acquisition module configured in the CN, and a fourth data acquisition module configured in AISACC.

[0049] The UPF is used to detect user QoS data when a user goes online and uses a contracted application or service, according to the received data reporting instruction, and send the detected user QoS data to the first data acquisition module.

[0050] The first data acquisition module is used to report the received user QoS data;

[0051] The fourth data acquisition module is used to collect RAN-side data and BN-side data from the RAN management center and BN management center, respectively.

[0052] Furthermore, the UPF receives data reporting instructions in the following ways:

[0053] Based on the user's QoS guarantee package subscribed to by the user through OSS to the business domain PCF, the PCF issues PCC policies to the UPF, instructing the UPF to initiate user QoS data awareness reporting when the user goes online and uses the subscribed application or service; or,

[0054] Based on the user protection policy that the user subscribes to through OSS with AISACC, AISACC subscribes to the user QoS data reporting event by subscribing to the first data acquisition module deployed in CN. The first data acquisition module then interacts with UPF to instruct UPF to start user QoS data awareness reporting when the user goes online and uses the subscribed application or service.

[0055] Compared with the prior art, the beneficial technical effects achieved by the present invention are as follows:

[0056] 1) This invention can dynamically sense user communication quality parameters in real time, and trigger the protection of user communication quality when communication quality parameters deteriorate, so as to protect the user's communication quality to the greatest extent. After the user's communication quality in the mobile communication network is optimized, it can continue to sense the user's QoS data and changes, and continue to optimize and protect when QoS deteriorates, thereby forming an intelligent optimization closed loop.

[0057] 2) The components of this invention are flexible in deployment and easy to activate. They can be superimposed on the existing mobile network communication in the business domain without affecting it, or they can be deeply integrated with business domain equipment, which facilitates the selection and implementation of engineering projects.

[0058] 3) This invention can effectively improve the utilization rate of mobile communication network resources, protect the investment value of each bit of network resources, and greatly improve the user's application experience and satisfaction. It has high engineering promotion and application value. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the overall logical framework of the intelligent self-optimizing mobile communication network system proposed in this invention;

[0060] Figure 2 This is a flowchart of the intelligent self-optimizing mobile communication quality assurance method of the present invention;

[0061] Figure 3 This is a schematic diagram of the intelligent self-optimizing mobile communication quality assurance system of the present invention;

[0062] Figure 4 This is a flowchart illustrating the workflow of the intelligent self-optimizing mobile communication quality assurance system according to an embodiment of the present invention. Detailed Implementation

[0063] The present invention will be further described below with reference to specific embodiments. These embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0064] like Figure 1 The diagram illustrates an intelligent self-optimizing mobile communication network system proposed in this invention. It is constructed by overlaying AISACC onto an existing mobile communication network framework. Specifically, this mobile communication network system comprises four domains:

[0065] 1. Business Domain:

[0066] The service domain, also known as the service network domain, is the foundational layer of an intelligent self-optimizing mobile communication network system. It provides basic mobile communication service capabilities and includes several components such as mobile terminal devices, Radio Access Network (RAN), Bearer Network (BN), and Core Network (CN).

[0067] 2. Management Domain:

[0068] The management domain is the management portion of the service domain, comprising the RAN management center, BN management center, and CN management center. Among them:

[0069] The RAN Management Center is responsible for FCAPS management, lifecycle management, and resource (virtual or physical) management and orchestration of radio resources.

[0070] The BN Management Center is responsible for FCAPS management and lifecycle management of the bearer network, as well as the management and orchestration of Overlay and Underlay resources of the bearer network.

[0071] The CN Management Center is responsible for FCAPS management, lifecycle management, and resource (virtual or physical) management and orchestration of core network devices.

[0072] 3. Operations and Maintenance Support Domain:

[0073] The Operations and Maintenance Support Domain is the operational support component of the intelligent self-optimizing mobile communication network system, including user service contracting, package management, billing management, as well as market strategies and planning for the mobile communication system and the promotion of new services.

[0074] 4. Artificial Intelligence Domain:

[0075] The Artificial Intelligence Domain (AISACC) is the intelligent brain of the self-optimizing mobile communication network system, overlaying AISACC onto the existing mobile communication network framework. AISACC is responsible for collecting operational status data from the entire mobile communication network, including performance, alarms, logs, various data records, and dynamic information about service operations, such as session information and traffic information, and analyzing the collected data. Based on the analyzed data, AISACC generates corresponding control policies and, through the management domain or core network's capability open platform, maps and transmits them to the corresponding devices or components, dynamically optimizing the communication network's operational status.

[0076] exist Figure 1 Based on the structure shown, as Figure 2 As shown, a smart self-optimizing mobile communication quality assurance method includes:

[0077] Step S1: Based on the QoS guarantee package or guarantee policy signed by the user, when the user goes online and uses the signed application or service, obtain the user's QoS data from the previous time period, and collect RAN-side data and BN-side data at the same time.

[0078] The user QoS data includes:

[0079] Bandwidth, denoted as WB, is measured in Mb / s (megahertz bits per second).

[0080] Delay, denoted as DELAY, is measured in milliseconds (ms).

[0081] Packet loss rate, denoted as PkgLosRat, is a percentage value.

[0082] The collected BN-side data mainly consists of QoS parameter information for the bearer channel where the user resides. The collected RAN-side data mainly consists of radio resource information, congestion information, and user density information for the base station or cell where the user resides.

[0083] Step S2: Preprocess the user QoS data of the previous time period to obtain the preprocessed user QoS data of the previous time period;

[0084] Preprocessing includes data cleaning and formatting.

[0085] Formatting involves converting user QoS data into the format and content used by the Prohibit model. Specifically, formatting methods include:

[0086] Bandwidth data remains unchanged;

[0087] The formatting method for delay data is DELAY = DELAY * 10;

[0088] The packet loss rate data processing method is PkgLosRat = PkgLosRat * 100 * 100.

[0089] Step S3: Integrate and calculate the preprocessed user QoS data from the previous time period to obtain the comprehensive QoS quality difference of the previous time period.

[0090] Calculate the overall user QoS quality deviation using the following formula:

[0091] QV=WB×W1+DELAY×W2+PkgLosRat×W3

[0092] Where QV is the overall QoS quality of the user; WB is the bandwidth; DELAY is the latency; PkgLosRat is the packet loss rate; W1, W2, and W3 are the weights of bandwidth, latency, and packet loss rate, respectively.

[0093] Step S4: Input the comprehensive QoS quality of the user in the previous time period into the comprehensive QoS quality of the prediction model to predict the comprehensive QoS quality of the next time period.

[0094] The QoS quality deterioration prediction model uses the Prophet model and can be trained using the following method:

[0095] Step ss1: Obtain the historical user QoS dataset DS1;

[0096] Obtain historical user QoS data, including:

[0097] Bandwidth, denoted as WB, is measured in Mb / s (megahertz bits per second).

[0098] Delay, denoted as DELAY, is measured in milliseconds (ms).

[0099] Packet loss rate, denoted as PkgLosRat, is a percentage value.

[0100] Generate vector sets consisting of the above three elements for different time periods, i.e., the historical dataset DS1:

[0101] T1:{WB.t1 DELAY.t1 PkgLosRat.tl}

[0102] T2:{WB.t2 DELAY.t2 PkgLosRat.t2} ......

[0104] Tn:{WB.tn DELAY.tn PkgLosRat.tn}

[0105] Where WB.tn is the bandwidth of time period n, DELAY.tn is the latency of time period n, and PkgLosRat.tn is the packet loss rate of time period n.

[0106] Step ss2: Preprocess the dataset DS1 to obtain the dataset DS2;

[0107] Following the preprocessing method in step S2, each group of data in dataset DS1 is converted into a format and content suitable for the Prophet model. The processed data generates a new historical dataset DS2.

[0108] T1:{WB.tl'DELAY.t1'PkgLosRat.tl'}

[0109] T2:{WB.t2'DELAY.t2'PkgLosRat.t2'} ......

[0111] Tn:{WB.tn'DELAY.tn'PkgLosRat.tn'}.

[0112] Step ss3: Integrate and calculate each group of data in dataset DS2 to obtain the comprehensive dataset DS3 of user QoS quality difference.

[0113] Following the formula in step S3, each group of data in dataset DS2 is integrated, and the three elements in the integration vector are combined into one element. The resulting historical dataset DS3 is shown below:

[0114] T1:{QV.t1}

[0115] T2:{QV.t2} ......

[0117] Tn:{QV.tn}

[0118] Wherein, QV.tn is the comprehensive QoS quality difference of users in time period n.

[0119] Step ss4: Divide the dataset DS3 into a training set and a test set. Use the training set to train the Prophet model and use the test set to test the training effect of the model.

[0120] The first 80% of the data in the DS3 dataset is used as the learning set, and the last 20% is used as the test set.

[0121] Set the parameter values ​​of the Prophet model, use the DS3 learning set as input parameters to learn and train the model, and then use the DS3 test set as validation data to validate the trained model. If the predicted data fits the trend of the actual data in a consistent manner and the fitting error is less than 2%, the Prophet model training is complete.

[0122] After obtaining the QoS quality quality comprehensive quantity prediction model, the model is used to predict the QV data of users in future time periods. By inputting the QoS quality quality quality comprehensive quantity of users in the previous time period into the QoS quality quality quality comprehensive quantity prediction model, the QoS quality quality quality comprehensive quantity of the next time period can be predicted.

[0123] For example, predicting the QV value over the next 10 days, including upper and lower limits. Model output results:

[0124] Tday1:{QV.tday1,QVdw.tdayl,QVup.tday1}

[0125] Tday2:{QV.tday2,QVdw.tday2,QVup.tday2} ......

[0127] Tday10:{QV.tday10,QVdw.tday10,QVup.tday10}

[0128] Where QV.tdayk is the predicted QV value for day k, QVdw.tdayk is the lower limit of QV for day k, and QVup.tdayk is the upper limit of QV for day k.

[0129] Step S5: When the next time period arrives, collect the user QoS data for the next time period in real time and calculate the comprehensive QoS quality difference of the user in the next time period in real time.

[0130] The user QoS data for the next time period is collected in real time. The user QoS data collected in real time is preprocessed according to the aforementioned preprocessing method. Then, the user QoS quality difference for the next time period is calculated based on the QoS quality difference calculation formula.

[0131] Step S6: Compare the real-time QoS quality difference of the next time period with the predicted QoS quality difference of the next time period to determine whether the user's QoS has deteriorated.

[0132] Determine whether the real-time QoS quality difference aggregate value for the next time period exceeds the predicted QoS quality difference aggregate value range [QVdw, QVup] for the next time period.

[0133] If the real-time QoS quality difference index for the next time period is within the predicted range of [QVdw, QVup] for the QoS quality difference index for the next time period, then the user's QoS is considered normal.

[0134] If the real-time QoS quality deterioration value for the next time period exceeds the predicted QoS quality deterioration value range [QVdw, QVup] for the next time period, then the user's QoS is judged to have deteriorated.

[0135] Step S7: Based on the deterioration of user QoS, and combining the collected RAN-side data and BN-side data, a QoS adjustment strategy is generated based on the locally pre-set expert rule tree or experience base, and the QoS adjustment strategy is distributed to each device in the service domain for execution.

[0136] The generated QoS adjustment policy decision may include:

[0137] Adjustment policies for user use of radio resources implemented on the RAN side include, but are not limited to, QoS level adjustment, channel preemption method adjustment, and non-value user radio resource occupation method adjustment.

[0138] Adjustment strategies for user use of transport network resources implemented on the BN side, including but not limited to improving the QoS level of data transmission and switching the transmission channel to a higher quality channel;

[0139] Adjustment strategies for user data flow forwarding quality implemented on the CN side include, but are not limited to, improving the QoS level of the data flow, switching the data flow to a specific high-performance transmission network card, and adjusting the data flow offloading method.

[0140] Among them, combined Figure 1 As shown, distributing QoS adjustment policies to various devices in the service domain for execution can include:

[0141] The generated QoS adjustment policies executed on the RAN side, BN side, or CN side are distributed to the SMF (Session Management Function) via the PCF of the service domain. The SMF then distributes the QoS adjustment policies executed on the RAN side to each device on the RAN side via the AMF (Access Management Function) for execution. The SMF then distributes the QoS adjustment policies executed on the BN side or CN side to each device on the BN side or CN side for execution via the UPF (User Port Function).

[0142] The generated QoS adjustment policies, which are executed on the RAN side, BN side, or CN side, are sent to the NEF capability open platform in the service domain. The NEF capability open platform then sends them to the SMF via the PCF. The SMF then sends the QoS adjustment policies executed on the RAN side to each device on the RAN side via the AMF. The SMF then sends the QoS adjustment policies executed on the BN side or CN side to each device on the BN side or CN side via the UPF.

[0143] The generated QoS adjustment policies, which are executed on the RAN side, BN side, or CN side, are distributed to the corresponding RAN management center, BN management center, or CN management center, respectively. The RAN management center, BN management center, or CN management center then maps the corresponding QoS adjustment policies to the operation instructions for the devices in their respective domains and distributes the operation instructions to the devices in their respective domains for execution.

[0144] like Figure 3 The diagram illustrates an intelligent self-optimizing mobile communication quality assurance system according to an embodiment of the present invention. This system is deployed in AISACC and includes a data acquisition module, a data preprocessing module, a data integration and calculation module, a QoS quality difference comprehensive prediction module, a data analysis module, a strategy decision module, and a data warehouse.

[0145] The data acquisition module is responsible for interacting with devices in the business domain and management domain to collect operational status data of the mobile communication system.

[0146] Data acquisition methods: It supports data acquisition through interaction with management domain devices via management channels, as well as data acquisition through interaction with business domain devices via signaling or data channels; it supports both synchronous and asynchronous data acquisition.

[0147] More specifically, the data acquisition module is used to acquire the user's QoS data for the previous time period when the user goes online and uses the contracted application or service, based on the QoS guarantee package or guarantee policy signed by the user, and to collect the user's QoS data for the next time period in real time when the next time period arrives, while simultaneously collecting RAN-side data and BN-side data.

[0148] The data acquisition module offers flexible deployment options, allowing it to be deployed in the AI ​​domain or independently in the CN / BN / RAN domains. Furthermore, it can be deployed in a simplified, tailored manner alongside service domain devices, such as in UPF devices, eNBs (Evloved Node Bases), or NRs (New Radios).

[0149] Combination Figure 1As shown, in one embodiment, the data acquisition module includes a UPF and a first data acquisition module disposed in the CN, a second data acquisition module disposed in the BN, and a third data acquisition module disposed in the RAN.

[0150] UPF detects user QoS data when a user goes online and uses a subscribed application or service, based on the received data reporting instructions, and sends the detected user QoS data to the first data acquisition module.

[0151] The first data acquisition module is used to report the received user QoS data to the data preprocessing module;

[0152] The second data acquisition module is used to acquire data from the BN side;

[0153] The third data acquisition module is used to acquire data from the RAN side.

[0154] In one implementation, users can subscribe to a QoS guarantee package through the Service Domain PCF via OSS. The PCF issues a PCC (Policy Charge Control) policy to the UPF via the SMF, instructing the UPF to initiate user QoS data awareness reporting when the user goes online and uses the applications or services subscribed to in the package.

[0155] In another implementation, users can also subscribe to a user protection policy through OSS with AISACC. AISACC then subscribes to user QoS data reporting events through subscription messages to the first data acquisition module deployed in the CN. The first data acquisition module interacts with the UPF to instruct the UPF to initiate user QoS data awareness reporting when the user goes online and uses the subscribed application or service.

[0156] Combination Figure 1 As shown, in another embodiment, the data acquisition module includes a UPF and a first data acquisition module disposed in the CN, and a fourth data acquisition module disposed in the AISACC.

[0157] Among them, UPF is used to detect user QoS data when a user goes online and uses the contracted application or service, based on the received data reporting instructions, and send the detected user QoS data to the first data acquisition module;

[0158] The first data acquisition module is used to report the received user QoS data to the data preprocessing module;

[0159] The fourth data acquisition module is used to collect RAN-side data and BN-side data from the RAN management center and BN management center, respectively.

[0160] The data preprocessing module is mainly responsible for cleaning and formatting the collected data.

[0161] Data cleaning refers to the process of re-examining and verifying collected data, removing duplicate information, correcting existing errors, and further providing data consistency verification. The goal is to "wash away" the "dirty" data, discover and correct identifiable errors in data files, including checking data consistency and handling invalid and missing values. It utilizes techniques such as mathematical statistics, data mining, or predefined cleaning rules to transform dirty data into data that meets data quality requirements.

[0162] Data formatting, also known as data normalization or data standardization, refers to standardizing cleaned data according to the format required by AI models in AISACC. This transforms diverse data from acquisition modules into a consistent standard format that the system can recognize and process, which is beneficial for the training, inference, and verification of the system's AI models and improves the overall system efficiency.

[0163] More specifically, the data preprocessing module is used to preprocess the user QoS data of the previous time period and the user QoS data of the next time period collected in real time, respectively, to obtain the preprocessed user QoS data of the previous time period and the real-time user QoS data of the next time period.

[0164] The data integration and calculation module is used to integrate and calculate the preprocessed user QoS data of the previous time period and the real-time user QoS data of the next time period to obtain the comprehensive user QoS quality difference of the previous time period and the comprehensive user QoS quality difference of the next time period.

[0165] The QoS quality deterioration comprehensive quantity prediction module is used to input the user QoS quality deterioration comprehensive quantity of the previous time period into the pre-trained QoS quality deterioration comprehensive quantity prediction model to predict the QoS quality deterioration comprehensive quantity of the next time period.

[0166] The data analysis module is used to compare the real-time QoS quality difference of the next time period with the predicted QoS quality difference of the next time period to determine whether the user's QoS has deteriorated.

[0167] The strategy decision module is used to generate QoS adjustment policies based on the deterioration of user QoS, combined with the collected RAN-side data and BN-side data, and based on the locally preset expert rule tree or experience base. The QoS adjustment policies are then distributed to various devices in the service domain for execution.

[0168] Specifically, the strategy decision module receives analysis results from the data analysis module and, combined with data from the RAN and BN sides, generates specific QoS adjustment strategies for mobile communication services based on locally pre-built expert rule trees or experience bases. For example:

[0169] (1) On the RAN side, the strategy for users to use radio resources is adjusted, such as QoS level adjustment, channel preemption method adjustment, and non-value user radio resource occupation method adjustment.

[0170] (2) On the BN side, the user's adjustment strategy for using transmission network resources, such as improving the QoS level of data transmission, switching the transmission channel to a channel with better quality, etc.

[0171] (3) On the CN side, the adjustment strategy for the forwarding quality of user data streams, such as improving the QoS level of data streams, switching data streams to specific high-performance transmission network cards, and adjusting the data stream offloading method.

[0172] The user QoS adjustment policy can be issued and executed in any of the following three ways:

[0173] Method 1: The policy adjustment is issued and executed by the policy decision module interacting with the PCF of the service domain. The policy is then issued to the SMF via the PCF, and then the SMF issues the QoS adjustment policy executed on the RAN side to the RAN side via the AMF. The SMF then issues the QoS adjustment policy executed on the BN side or CN side via the UPF to the BN side or CN side. The CN side controls the adjustment of the QoS policy for users in the entire mobile communication network. This part follows the mobile communication PCC standard interface and method defined by 3GPP (Third Generation Partnership Project).

[0174] Method 2: Adjustment policy is issued and executed. Similar to Method 1, the difference is that the policy decision module does not interact directly with the PCF of the business domain. Instead, it first interacts with the external capability open interface of the business domain, namely the NEF capability open platform. The NEF capability open platform issues the policy to the SMF via the PCF. The SMF then issues the QoS adjustment policy executed on the RAN side to the RAN side via the AMF. The SMF then issues the QoS adjustment policy executed on the BN side or CN side to the BN side or CN side via the UPF. Finally, the policy is issued to each device in the business domain for execution.

[0175] Method 3: Adjustment policy is issued and executed. The policy decision module interacts with the management centers of RAN, BN, and CN to map the QoS adjustment policy to the corresponding operation instructions of the devices in their respective domains. The management centers of each domain then interact with the devices to complete the execution of the QoS adjustment policy.

[0176] The data warehouse is used to store user QoS data, RAN-side data, and BN-side data collected by the data acquisition module. It is responsible for storing and maintaining the collected data, supporting highly reliable storage of different data formats and high-performance concurrent data I / O, and can trace back the history of data changes when data problems or failures occur, ensuring the security and reliability of the data.

[0177] Through the division of labor and cooperation among different devices and modules in various domains, it is possible to realize real-time dynamic perception of user communication quality parameters, and trigger the protection and optimization of user communication quality when communication quality parameters deteriorate, so as to maximize the protection of user communication quality.

[0178] The following specific embodiments are provided for illustration:

[0179] Users who have signed up for a QoS guarantee package for a certain application or service need to have their communication QoS quality guaranteed when using the application or service, regardless of whether mobile communication network resources are normal or strained.

[0180] like Figure 4 As shown, the workflow for ensuring QoS of user communication using the intelligent self-optimizing mobile communication quality assurance system of the present invention is as follows:

[0181] 400: The user signs up for the QoS guarantee package through the PCF in the business domain via OSS. The PCF issues the PCC policy to the UPF via the SMF, instructing the UPF to start the user QoS data awareness reporting when the user goes online and uses the application or service signed up in the package.

[0182] 401: QoS data awareness and reporting for the service used by the protected user;

[0183] On the CN side, user QoS data is detected by UPF and reported to the first data acquisition module in CN. Then, the first data acquisition module interacts with AISACC and reports the user QoS data to the data preprocessing module in AISACC. The reporting method can be based on management protocol messages from the management plane or notification messages from the signaling plane.

[0184] Data reporting from the BN and RAN sides is handled by AISACC via the fourth data acquisition module when users use services, collecting data from the management centers on both sides. All transmitted data is based on the management plane's management protocols.

[0185] 402: User QoS data preprocessing;

[0186] Data preprocessing includes data cleaning and formatting, which is handled by the data preprocessing module. This module cleans, normalizes, and formats the collected QoS data, and is responsible for the integrity, consistency, and standardization management and maintenance of heterogeneous data. Existing technologies are used for data cleaning.

[0187] 403: User QoS data storage;

[0188] The preprocessed user QoS data, collected BN-side data, and RAN-side data are stored in the data warehouse.

[0189] 404: User QoS data analysis and poor quality identification;

[0190] The system collects QoS data from the data warehouse for the most recent time period. The data integration and calculation module then integrates and calculates the collected user QoS data for the most recent time period to obtain the comprehensive QoS quality quality of the most recent time period.

[0191] Input the QoS quality deterioration comprehensive quantity of the most recent time period into the QoS quality deterioration comprehensive quantity prediction model to predict the QoS quality deterioration comprehensive quantity of the next time period. For example, the user QoS quality deterioration comprehensive quantity QV of time period T1, including QVdw and QVup;

[0192] Then, after the arrival of the T1 period, the user QoS data of the T1 period is collected in real time through the first data acquisition module. After the data is preprocessed by the data preprocessing module, the user QoS quality difference comprehensive quantity of the T1 period is calculated in real time through the data integration and calculation module.

[0193] Finally, the data analysis module compares the real-time QoS quality deterioration value for the next time period with the predicted QoS quality deterioration value for the next time period. It determines whether the real-time QoS quality deterioration value is within the predicted range [QVdw, QVup]. If it is within the range, the user's QoS is normal; if it exceeds the range, it indicates QoS degradation. The data analysis module outputs the normal or degraded result and sends it to the policy decision module.

[0194] 405: User QoS adjustment policy generation;

[0195] The strategy decision module receives the analysis results from the data analysis module, combines the data from the BN side and the RAN side, and generates QoS adjustment strategy decisions based on the locally pre-built expert rule tree or experience base.

[0196] QoS adjustment policies include, but are not limited to, the following information:

[0197] 1. On the RAN side, adjustments to user radio resource usage strategies, such as QoS level adjustments, channel preemption method adjustments, and non-value user radio resource usage method adjustments;

[0198] 2. On the TN side, the adjustment strategy for users' use of transmission network resources, such as improving the QoS level of data transmission and switching the transmission channel to a channel with better quality;

[0199] 3. On the CN side, strategies for adjusting the quality of user data flow forwarding include improving the QoS level of the data flow, switching the data flow to a specific high-performance transmission network card, and adjusting the data flow offloading method.

[0200] 406: User QoS adjustment policy is issued and executed;

[0201] In this embodiment, the policy decision module interacts with the management centers of RAN, TN, and CN to map the QoS adjustment policy into operation instructions corresponding to the devices in their respective domains. The management centers of each domain then interact with the devices to complete the execution of the QoS adjustment policy.

[0202] 407: Closed-loop detection and optimization of QoS adjustment strategy;

[0203] After the user's QoS in the mobile communication network is optimized, the system continues to sense the user's QoS data and changes, and continues to optimize and ensure QoS communication quality when it deteriorates. In other words, this step can directly lead to step 401, forming an intelligent optimization closed loop.

[0204] As can be seen from the above embodiments, the present invention can dynamically sense user communication quality parameters in real time, and trigger the protection of user communication quality when communication quality parameters deteriorate, thereby maximizing the protection of user communication quality; after the user communication quality in the mobile communication network is optimized, it can continue to sense the user's QoS data and changes, and continue to optimize and protect when it deteriorates, thus forming an intelligent optimization closed loop.

[0205] The present invention has been disclosed above with reference to preferred embodiments, but it is not intended to limit the present invention. All technical solutions obtained by adopting equivalent substitutions or equivalent transformations fall within the protection scope of the present invention.

Claims

1. A method for ensuring the quality of mobile communication through intelligent self-optimization, characterized in that, include: Based on the QoS guarantee package or guarantee policy signed by the user, when the user goes online and uses the signed application or service, the user's QoS data from the previous time period is obtained, and RAN side data and BN side data are collected at the same time. Preprocess the user QoS data from the previous time period to obtain the preprocessed user QoS data from the previous time period; The preprocessed user QoS data from the previous time period is integrated and calculated to obtain the comprehensive QoS quality difference of the previous time period. Input the overall QoS quality of the user in the previous period into the overall QoS quality of the prediction model to predict the overall QoS quality of the next period. When the next time period arrives, the user QoS data for the next time period is collected in real time, and the comprehensive QoS quality difference of the user for the next time period is calculated in real time. The real-time QoS quality difference of the next time period is compared with the predicted QoS quality difference of the next time period to determine whether the user's QoS has deteriorated. Based on the deterioration of user QoS, and combining the collected RAN-side data and BN-side data, a QoS adjustment policy is generated based on the locally pre-configured expert rule tree or experience base, and the QoS adjustment policy is distributed to various devices in the service domain for execution.

2. The intelligent self-optimizing mobile communication quality assurance method according to claim 1, characterized in that, The user QoS data includes at least bandwidth, latency, and packet loss rate. The overall user QoS quality score is calculated according to the following formula: QV=WB×W1+DELAY×W2+PkgLosRat×W3 Where QV is the overall QoS quality of the user; WB is the bandwidth; DELAY is the latency; PkgLosRat is the packet loss rate; W1, W2, and W3 are the weights of bandwidth, latency, and packet loss rate, respectively.

3. The intelligent self-optimizing mobile communication quality assurance method according to claim 1, characterized in that, The QoS quality quality comprehensive prediction model is trained using the following method: Obtain the historical user QoS dataset DS1. Each data set in dataset DS1 includes bandwidth, latency, and packet loss rate. Preprocess dataset DS1 to obtain dataset DS2; The data in each set of dataset DS2 are integrated and calculated to obtain the comprehensive dataset of user QoS quality difference DS3. The DS3 dataset was divided into a training set and a test set. The Prophet model was trained using the training set, and the training effect of the model was tested using the test set.

4. The intelligent self-optimizing mobile communication quality assurance method according to claim 1, characterized in that, The step of comparing the real-time QoS quality deterioration of the next time period with the predicted QoS quality deterioration of the next time period to determine whether the user's QoS has deteriorated includes: Determine whether the real-time QoS quality difference index for the next time period exceeds the predicted range of the QoS quality difference index for the next time period. If the real-time QoS quality difference index for the next time period is within the range of the predicted QoS quality difference index for the next time period, then the user's QoS is considered normal. If the real-time QoS quality deterioration value for the next time period exceeds the predicted QoS quality deterioration value for the next time period, then the user's QoS is considered to have deteriorated.

5. The intelligent self-optimizing mobile communication quality assurance method according to claim 1, characterized in that, The generated QoS adjustment policy decision includes: Adjustment policies for user use of radio resources implemented on the RAN side include, but are not limited to, QoS level adjustment, channel preemption method adjustment, and non-value user radio resource occupation method adjustment; Adjustment strategies for user use of transport network resources implemented on the BN side include, but are not limited to, improving the QoS level of data transmission and switching the transport channel to a higher quality channel; Adjustment strategies for user data flow forwarding quality implemented on the CN side include, but are not limited to, improving the QoS level of the data flow, switching the data flow to a specific high-performance transmission network card, and adjusting the data flow offloading method.

6. The intelligent self-optimizing mobile communication quality assurance method according to claim 5, characterized in that, The process of distributing the QoS adjustment policy to each device in the service domain for execution includes any of the following methods: The generated QoS adjustment policies executed on the RAN side, BN side, or CN side are sent to the SMF via the PCF of the service domain. The SMF then sends the QoS adjustment policies executed on the RAN side to each device on the RAN side for execution via the AMF. The SMF then sends the QoS adjustment policies executed on the BN side or CN side to each device on the BN side or CN side for execution via the UPF. The generated QoS adjustment policies, which are executed on the RAN side, BN side, or CN side, are sent to the NEF capability open platform in the service domain. The NEF capability open platform then sends them to the SMF via the PCF. The SMF then sends the QoS adjustment policies executed on the RAN side to each device on the RAN side via the AMF. The SMF then sends the QoS adjustment policies executed on the BN side or CN side to each device on the BN side or CN side via the UPF. The generated QoS adjustment policies, which are executed on the RAN side, BN side, or CN side, are distributed to the corresponding RAN management center, BN management center, or CN management center, respectively. The RAN management center, BN management center, or CN management center then maps the corresponding QoS adjustment policies to the operation instructions for the devices in their respective domains and distributes the operation instructions to the devices in their respective domains for execution.

7. A smart, self-optimizing mobile communication quality assurance system, characterized in that, The system is deployed within AISACC, which is overlaid on an existing mobile communication network framework. The system includes: The data acquisition module is used to acquire the user's QoS data from the previous time period when the user goes online and uses the contracted application or service, based on the QoS guarantee package or guarantee policy signed by the user, and to collect the user's QoS data for the next time period in real time when the next time period arrives. At the same time, it collects RAN side data and BN side data. The data preprocessing module is used to preprocess the user QoS data of the previous time period and the user QoS data of the next time period collected in real time, respectively, to obtain the preprocessed user QoS data of the previous time period and the real-time user QoS data of the next time period. The data integration and calculation module is used to integrate and calculate the preprocessed user QoS data of the previous time period and the real-time user QoS data of the next time period to obtain the comprehensive user QoS quality difference of the previous time period and the comprehensive user QoS quality difference of the next time period. The QoS quality poor comprehensive quantity prediction module is used to input the user QoS quality poor comprehensive quantity of the previous time period into the QoS quality poor comprehensive quantity prediction model to predict the QoS quality poor comprehensive quantity of the next time period. The data analysis module is used to compare the real-time QoS quality difference of the next time period with the predicted QoS quality difference of the next time period to determine whether the user's QoS has deteriorated. The strategy decision module is used to generate QoS adjustment policies based on the deterioration of user QoS, combined with the collected RAN-side data and BN-side data, and based on the locally preset expert rule tree or experience base. The QoS adjustment policies are then distributed to various devices in the service domain for execution.

8. The intelligent self-optimizing mobile communication quality assurance system according to claim 7, characterized in that, The data acquisition module includes a UPF and a first data acquisition module located in the CN, a second data acquisition module located in the BN, and a third data acquisition module located in the RAN. The UPF is used to detect user QoS data when a user goes online and uses a contracted application or service, according to the received data reporting instruction, and send the detected user QoS data to the first data acquisition module. The first data acquisition module is used to report the received user QoS data; The second data acquisition module is used to acquire data from the BN side; The third data acquisition module is used to acquire RAN-side data.

9. The intelligent self-optimizing mobile communication quality assurance system according to claim 7, characterized in that, The data acquisition module includes a UPF and a first data acquisition module located in the CN, and a fourth data acquisition module located in the AISACC. The UPF is used to detect user QoS data when a user goes online and uses a contracted application or service, according to the received data reporting instruction, and send the detected user QoS data to the first data acquisition module. The first data acquisition module is used to report the received user QoS data; The fourth data acquisition module is used to collect RAN-side data and BN-side data from the RAN management center and BN management center, respectively.

10. The intelligent self-optimizing mobile communication quality assurance system according to claim 8 or 9, characterized in that, The UPF receives data reporting instructions in the following ways: Based on the user's QoS guarantee package subscribed to by the user through OSS to the business domain PCF, the PCF issues PCC policies to the UPF, instructing the UPF to initiate user QoS data awareness reporting when the user goes online and uses the subscribed application or service; or, Based on the user protection policy that the user subscribes to through OSS with AISACC, AISACC subscribes to the user QoS data reporting event by subscribing to the first data acquisition module deployed in CN. The first data acquisition module then interacts with UPF to instruct UPF to start user QoS data awareness reporting when the user goes online and uses the subscribed application or service.

Images