A 5G base station computing power dynamic prediction and resource allocation system based on digital twinning

Abstract

The application relates to the technical field of base station computing power allocation, and specifically provides a 5G base station computing power dynamic prediction and resource allocation system based on digital twinning, which comprises the following modules: a platform computing power demand total amount prediction module, which is used for generating a platform computing power prediction demand total amount according to historical operation data sets; a user service computing power demand total amount prediction module, which is used for generating a user service computing power prediction demand total amount according to a historical user computing power demand total amount set; an allocation scheme generation module, which is used for generating a computing power allocation scheme according to base station total computing power information, the platform computing power prediction demand total amount and the user service computing power prediction demand total amount; and a computing power allocation module, which is used for performing computing power allocation on tasks and user services of a digital twinning platform according to the computing power allocation scheme; the system can comprehensively consider total computing power limitation and predicted computing power demands of the digital twinning platform and user services to generate a computing power allocation scheme balancing computing power demands of the digital twinning platform and user services.

Description

Technical Field

[0001] This application relates to the field of base station computing power allocation technology, and more specifically, to a 5G base station computing power dynamic prediction and resource allocation system based on digital twins. Background Technology

[0002] 5G base station systems provide wireless communication services to user terminals within a region. These systems consist of physical base stations comprised of various physical devices (such as antenna units, distributed units, and centralized units), providing wireless coverage, signal processing, and data transmission capabilities. To manage the operation of these physical base stations, existing 5G base station systems deploy management platforms based on digital twin technology. Specifically, existing technologies construct virtual models of physical base stations and their service environments based on digital twin platforms. These models are updated in real-time using various operational data collected from the physical base stations (such as user connections, traffic, device load, power consumption, and wireless environment parameters), allowing the virtual model to synchronously reflect the real-time status of the physical base stations. This enables computing power prediction using the virtual model. Specifically, the virtual model can infer the future computing power required by the physical base station (equivalent to predicting computing power demand) by analyzing historical and real-time operational data, and generate a computing power allocation strategy (equivalent to a resource allocation strategy) based on the predicted computing power demand. This computing power allocation strategy instructs the 5G base station system to adjust the internal computing power configuration of the physical base stations to meet user communication needs and optimize computing power utilization.

[0003] However, the digital twin platform itself does not operate without consumption. Specifically, the various tasks of the digital twin platform (such as data acquisition and processing, model training and updating, state synchronization and maintenance, computing power prediction calculation, and computing power allocation decision generation) all require computing power. This computing power usually comes from the physical base station. That is, the computing power consumption of the digital twin platform itself will occupy the total computing power of the physical base station.

[0004] Furthermore, the computing power consumption of a digital twin platform is not constant. Specifically, the amount of data processed by the virtual model depends on the rate and complexity of data collection by the physical base station. The update frequency of the virtual model and the complexity of the prediction algorithm affect the computing load. The generation frequency and granularity of the computing power allocation strategy also affect the computing power. Therefore, the computing power consumption of a digital twin platform will change dynamically over time. For example, when user traffic patterns change significantly, the digital twin platform may need to collect more data, update the model more frequently, or make more complex predictions, thereby increasing its own computing power requirements.

[0005] In certain specific scenarios, such as in high-density user areas or during large-scale events, physical base stations face a sharp increase in the number of users and service traffic. At this time, physical base stations need to invest a lot of computing power to process user data, perform baseband signal processing, and perform wireless resource scheduling to meet user communication needs. At the same time, in order to cope with rapidly changing network conditions, digital twin platforms may also need to collect data more frequently, update virtual models more quickly, and make more real-time computing power prediction and computing power allocation decisions, which will significantly increase the computing power consumption of the digital twin platform itself.

[0006] In scenarios where user traffic surges simultaneously and the digital twin platform's workload increases, the total computing power within the physical base station faces dual pressure: a portion of the computing power is used to process user traffic, while the other portion is used to run the digital twin platform. Since both the computing power required for the digital twin platform's operation and the computing power needed to process user traffic originate from the physical base station, a competition for computing power arises between the digital twin platform and user traffic. If the digital twin platform's own computing power demands are not properly managed, its consumption of computing power may crowd out the computing power allocated to processing user traffic.

[0007] Current computing power allocation systems based on digital twins primarily function to allocate computing power from physical base stations to serve users based on the computing power demands predicted by the digital twin model. However, when making computing power allocation decisions, these systems often focus on optimizing the processing efficiency or service quality of user services, failing to incorporate the fluctuating computing power demands of the digital twin platform itself into the overall computing power planning and allocation. For example, when the computing power allocation system predicts that a user service requires a large amount of computing power and attempts to allocate it, if the digital twin platform also needs to consume a large amount of computing power for model updates or computing power demand predictions, the actual resources available for user services will be lower than expected. Conversely, to ensure user services, the computing power allocation system may restrict the computing power usage of the digital twin platform, resulting in the virtual model not being updated in a timely manner or a decrease in the accuracy of computing power demand predictions, thus affecting subsequent computing power allocation decisions.

[0008] The current lack of dynamic perception and integrated management of the computing power consumption of the digital twin platform itself makes it difficult to achieve globally optimal resource allocation during periods of high user traffic load or rapid changes. In other words, the computing power allocation system may be caught in a dilemma between meeting user needs and ensuring the normal operation of the digital twin platform. Therefore, existing technologies have problems such as insufficient computing power allocated to the digital twin platform, resulting in the virtual model not being able to be updated in a timely manner or a decrease in the accuracy of computing power demand prediction, which affects the accuracy and reliability of computing power allocation decisions; and excessive computing power allocated to the digital twin platform, resulting in a large gap between the actual computing power allocated to user business processing and the expected computing power, which seriously affects the quality of user service.

[0009] Currently, there is no effective technical solution to the above-mentioned problems. It should be noted that the information disclosed in this section is only for understanding the background of the present invention and therefore may include information that does not constitute prior art. Summary of the Invention

[0010] The purpose of this application is to provide a 5G base station computing power dynamic prediction and resource allocation system based on digital twins. It can effectively solve the problems of insufficient computing power allocated to the digital twin platform, which leads to the virtual model not being updated in a timely manner or the prediction accuracy of computing power demand decreasing, affecting the accuracy and reliability of computing power allocation decisions; and excessive computing power allocated to the digital twin platform, which leads to a large gap between the actual computing power allocated to user business processing and the expected computing power, seriously affecting the quality of user services.

[0011] This application provides a 5G base station computing power dynamic prediction and resource allocation system based on digital twins, which includes:

[0012] The platform computing power demand prediction module is used to generate the predicted total computing power demand of the platform based on the historical operating dataset. The historical operating dataset is a collection of all internal operating data of the digital twin platform within the first preset time period before the current time node.

[0013] The user business computing power demand total prediction module is used to generate the user business computing power predicted total demand based on the historical user computing power demand total set. The historical user business computing power demand total set is the set of all user business computing power demand totals within the second preset time period before the current time node.

[0014] The base station total computing power acquisition module is used to acquire pre-set base station total computing power information;

[0015] The allocation scheme generation module is used to generate a computing power allocation scheme that includes the total computing power allocation for user service processing and the total computing power allocation for the digital twin platform itself, based on the total computing power information of the base station, the total predicted computing power demand of the platform, and the total predicted computing power demand for user services.

[0016] The computing power allocation module is used to allocate computing power to the tasks and user services of the digital twin platform according to the computing power allocation scheme.

[0017] This application provides a dynamic prediction and resource allocation system for 5G base station computing power based on digital twins. First, it predicts the computing power requirements of the digital twin platform and user services separately. Then, it comprehensively considers the total computing power limitations of the 5G base station and the predicted computing power requirements of the digital twin platform and user services to generate a computing power allocation scheme that balances the computing power requirements of the digital twin platform and user services. Finally, it allocates computing power to the digital twin platform and user services according to this scheme. Because this application explicitly incorporates the computing power requirements of the digital twin platform itself into the computing power prediction and allocation process, and because it can allocate appropriate computing power to the digital twin platform and user services based on the total computing power information of the base station, the total predicted computing power requirements of the platform, and the total predicted computing power requirements of user services, this application can effectively solve the problems of insufficient computing power allocated to the digital twin platform leading to untimely updates of the virtual model or decreased prediction accuracy of computing power requirements, affecting the accuracy and reliability of computing power allocation decisions; and excessive computing power allocated to the digital twin platform leading to a large gap between the actual and expected computing power allocated to user services, severely impacting user service quality.

[0018] Optionally, the process of generating a computing power allocation scheme that includes the total computing power allocation for user service processing and the total computing power allocation for the digital twin platform itself, based on the total computing power information of the base station, the total predicted computing power demand of the platform, and the total predicted computing power demand for user services, includes:

[0019] A1. Analyze whether the sum of the predicted total computing power demand of the platform and the predicted total computing power demand of user services is less than or equal to the total computing power information of the base station. If yes, proceed to step A2; otherwise, proceed to step A3.

[0020] A2. Take the total predicted computing power demand of the platform as the total computing power allocation of the digital twin platform itself, and take the total predicted computing power demand of user business as the total computing power allocation of user business processing, so as to obtain the computing power allocation scheme.

[0021] A3. Reduce the total predicted computing power demand of the platform and / or the total predicted computing power demand of user services until the sum of the total predicted computing power demand of the platform and the total predicted computing power demand of user services is less than or equal to the total computing power information of the base station. Then, use the total predicted computing power demand of the platform at this time as the total computing power allocation of the digital twin platform itself and use the total predicted computing power demand of user services at this time as the total computing power allocation of user services to obtain the computing power allocation scheme.

[0022] Optionally, step A3 includes:

[0023] A31. Calculate the reduction in computing power based on the sum of the total predicted computing power demand of the platform and the total predicted computing power demand of user services, and the total computing power information of the base station.

[0024] A32. Obtain a first user service set and a first platform task set, wherein the first user service set includes at least one first service type and the first platform task set includes at least one first task type;

[0025] A33. Based on the first business type, query the pre-built mapping relationship table between the first business type and the priority score to determine the business priority score corresponding to each first business type; and based on the first task type, query the pre-built mapping relationship table between the first task type and the priority score to determine the task priority score corresponding to each first task type.

[0026] A34. Calculate the total business priority score based on all business priority scores, and calculate the total task priority score based on all task priority scores.

[0027] A35. Determine the user business computing power reduction ratio and the platform task computing power reduction ratio based on the total score of business priority and the total score of task priority. The sum of the user business computing power reduction ratio and the platform task computing power reduction ratio is 1.

[0028] A36. Calculate the reduction in computing power for user services based on the reduction amount and the reduction ratio of computing power for user services, and calculate the reduction in computing power for the platform based on the reduction amount and the reduction ratio of computing power for platform tasks.

[0029] A37. Reduce the total predicted computing power demand of the platform by the amount of platform computing power reduction and reduce the total predicted computing power demand of user services by the amount of user service computing power reduction. Then, use the total predicted computing power demand of the platform at this time as the total computing power allocation of the digital twin platform itself and use the total predicted computing power demand of user services at this time as the total computing power allocation of user service processing to obtain a computing power allocation scheme.

[0030] Optionally, the first platform task set further includes at least one first task quantity, each first task quantity corresponding to a first task type; the first user service set further includes at least one first service quantity, each first service quantity corresponding to a first service type; step A34 includes:

[0031] A341. Calculate the first preliminary priority score for each first business type based on the business priority score and the number of first businesses, and then calculate the total business priority score based on all the first preliminary priority scores.

[0032] A342. Calculate the second preliminary priority score corresponding to each first task type based on the task priority score corresponding to the first task type and the number of first tasks, and then calculate the total task priority score based on all the second preliminary priority scores.

[0033] Optionally, step A342 includes:

[0034] A3421. Obtain the operating status of the digital twin platform, and then query the pre-built mapping relationship table of platform operating status and scoring compensation coefficient combination based on the operating status of the digital twin platform to determine the scoring compensation coefficient corresponding to different first task types.

[0035] A3422. Calculate the second preliminary priority score corresponding to each first task type based on the task priority score, the number of first tasks, and the score compensation coefficient. Then, calculate the total task priority score based on all the second preliminary priority scores.

[0036] Optionally, step A31 includes:

[0037] A311. Calculate the computing power reduction amount based on the sum of the platform's predicted total computing power demand and the user's predicted total computing power demand, the preset computing power buffer, and the base station's total computing power information.

[0038] Optionally, the steps of allocating computing power to the tasks and user services of the digital twin platform according to the computing power allocation scheme include:

[0039] B1. Obtain the second user service set and the second platform task set. The second user service set includes at least one set of second service types and their corresponding number of second services. The second platform task set includes at least one set of second task types and their corresponding number of second tasks.

[0040] B2. Based on the second business type, query the pre-built mapping table of the second business type and priority score to determine the business priority score corresponding to each second business type, and based on the second task type, query the pre-built mapping table of the second task type and priority score to determine the task priority score corresponding to each second task type.

[0041] B3. Calculate the third preliminary priority score for each second business type based on the business priority score and the number of second business types. Then, normalize all the third preliminary priority scores to determine the first weighted weight for each second business type.

[0042] B4. Calculate the fourth preliminary priority score corresponding to each second task type based on the task priority score and the number of second tasks. Then, normalize all the fourth preliminary priority scores to determine the second weighted weight corresponding to each second task type.

[0043] B5. Allocate computing power to user services based on the total computing power allocation for user business processing and all first weighted weights, and allocate tasks to the digital twin platform based on the total computing power allocation for the digital twin platform itself and all second weighted weights.

[0044] Optionally, the step of generating the predicted total platform computing power demand based on historical operational datasets includes:

[0045] C1. Use the interquartile range method to identify and remove outliers from historical running datasets;

[0046] C2. Generate the total predicted computing power demand of the platform based on historical operating datasets.

[0047] Optionally, the total computing power demand of all users in the historical user service computing power demand set corresponds to the same time period.

[0048] Optionally, the internal operating data of the digital twin platform includes processing unit occupancy, memory usage, and data processing volume.

[0049] As can be seen from the above, the 5G base station computing power dynamic prediction and resource allocation system based on digital twins provided in this application first predicts the computing power requirements of the digital twin platform and user services respectively. Then, it comprehensively considers the total computing power limit of the 5G base station and the predicted computing power requirements of the digital twin platform and user services to generate a computing power allocation scheme that balances the computing power requirements of the digital twin platform and user services. Finally, it allocates computing power to the digital twin platform and user services according to the computing power allocation scheme. Since this application explicitly incorporates the computing power requirements of the digital twin platform itself into the computing power prediction and allocation process, and this application can allocate appropriate computing power to the digital twin platform and user services based on the total computing power information of the base station, the total predicted computing power requirements of the platform, and the total predicted computing power requirements of user services, this application can effectively solve the problems of insufficient computing power allocated to the digital twin platform leading to the virtual model not being updated in a timely manner or the prediction accuracy of computing power requirements decreasing, affecting the accuracy and reliability of computing power allocation decisions, and excessive computing power allocated to the digital twin platform leading to a large gap between the actual computing power allocated to user service processing and the expected computing power, seriously affecting the quality of user services. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the structure of a 5G base station computing power dynamic prediction and resource allocation system based on digital twin, provided in an embodiment of this application.

[0051] Attached reference numerals: 1. Platform computing power demand prediction module; 2. User service computing power demand prediction module; 3. Base station total computing power acquisition module; 4. Allocation scheme generation module; 5. Computing power allocation module. Detailed Implementation

[0052] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0053] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0054] like Figure 1 As shown, this application provides a 5G base station computing power dynamic prediction and resource allocation system based on digital twins, which includes:

[0055] Platform computing power demand prediction module 1 is used to generate the predicted total computing power demand of the platform based on the historical running dataset. The historical running dataset is a collection of all internal running data of the digital twin platform within the first preset time period before the current time node.

[0056] User business computing power demand total prediction module 1 is used to generate the predicted total user business computing power demand based on the historical user computing power demand total set. The historical user business computing power demand total set is the set of all user business computing power demand totals within the second preset time period before the current time node.

[0057] Base station total computing power acquisition module 3 is used to acquire pre-set base station total computing power information;

[0058] The allocation scheme generation module 4 is used to generate a computing power allocation scheme that includes the total computing power allocation for user service processing and the total computing power allocation for the digital twin platform itself, based on the total computing power information of the base station, the total predicted computing power demand of the platform, and the total predicted computing power demand for user services.

[0059] The computing power allocation module 5 is used to allocate computing power to the tasks and user services of the digital twin platform according to the computing power allocation scheme.

[0060] The present application provides a 5G base station computing power dynamic prediction and resource allocation system based on digital twins, which aims to dynamically predict the computing power required for the operation of the digital twin platform and the computing power required for user services in the 5G base station environment, so as to reasonably allocate computing power to the digital twin platform and user services under the total computing power limit of the 5G base station, thereby minimizing computing power conflicts and various problems caused by unreasonable computing power allocation.

[0061] The platform computing power demand prediction module 1 in this embodiment can generate the predicted total computing power demand of the platform based on historical operating datasets. The working principle of the platform computing power demand prediction module 1 is as follows: The digital twin platform will continue to run, and under normal operating conditions, the various tasks inside the digital twin platform (such as data acquisition, model updates, state synchronization, etc.) will be executed periodically or triggered according to specific events. That is, all the internal operating data of the digital twin platform within the first preset time period before the current time node (equivalent to the operating data of the digital twin platform in the past period) can reflect the computing power consumption and computing power consumption pattern of the digital twin platform under a specific workload. The workload of the digital twin platform usually does not change abruptly. Therefore, the platform computing power demand prediction module 1 in this embodiment can discover the regularity of the computing power demand of the digital twin platform changing over time by analyzing all the internal operating data of the digital twin platform within the first preset time period before the current time node. Then, based on this regularity, the future computing power demand of the digital twin platform is predicted to obtain the predicted total computing power demand of the platform. Specifically, the platform computing power demand prediction module 1 in this embodiment can utilize existing time series prediction models (such as moving average prediction models, exponential smoothing prediction models, ARIMA prediction models, etc.) to extract features such as the changing trends and cycles of the internal operating data of the digital twin platform from historical operating datasets, and use these features to predict the future computing power demand of the digital twin platform. It should be understood that the effectiveness of time series prediction methods depends on the quality of historical data and the selection of models. Specifically, the higher the quality of historical data, the more accurate the extracted features, and selecting a suitable time series prediction model can better capture the regularity in historical data. The first preset time period in this embodiment is a value preset by those skilled in the art based on experience or actual needs. It should be understood that since the internal operating data of the digital twin platform will also increase sharply when user business volume surges, the value of the first preset time period should not be too large in order to ensure the accuracy and reliability of the predicted total computing power demand of the platform.

[0062] The user service computing power demand prediction module 1 in this embodiment can generate the predicted total computing power demand for user services based on the historical set of total user computing power demand. The working principle of the user service computing power demand prediction module 1 is as follows: the total computing power demand of all user services within the second preset time period before the current time node can reflect the computing power consumption and computing power consumption pattern of user services in the past period. A surge in user service volume will be reflected in the trend of a rapid increase in computing power consumption of user services at a certain time node in the past. Therefore, the user service computing power demand prediction module 1 in this embodiment can discover the regularity of the computing power demand of user services changing over time by analyzing the total computing power demand of all user services within the second preset time period before the current time node, and then predict the future computing power demand of user services based on this regularity to obtain the predicted total computing power demand of user services. Specifically, the user service computing power demand prediction module 1 in this embodiment can extract features such as the changing trend and cycle of the total user service computing power demand from the historical set of total user computing power demand using existing time series prediction models (such as moving average prediction models, exponential smoothing prediction models, ARIMA prediction models, etc.), and use these features to predict the future computing power demand of user services. The second preset time period in this embodiment is a value preset by those skilled in the art based on experience or actual needs. It should be understood that, since user service volume may surge, the value of the second preset time period should not be too large in order to ensure the accuracy and reliability of the predicted total user service computing power demand.

[0063] In this embodiment, the total computing power information of the base station is a pre-set value. Preferably, this embodiment can obtain the pre-set total computing power information of the base station by querying a pre-built mapping table of base station representation and total computing power using the unique identifier of the 5G base station. Since the maximum computing power (equivalent to the available computing power of the 5G base station) is designed when the 5G base station is designed, and the maximum computing power of the 5G base station is measured after the construction of the 5G base station is completed, the total computing power information of the base station in this embodiment can be the design value when designing the 5G base station or the measured value when measuring the 5G base station.

[0064] The allocation scheme generation module 4 in this embodiment can generate a computing power allocation scheme that includes the total computing power allocation for user service processing and the total computing power allocation for the digital twin platform itself, based on the total computing power information of the base station, the total predicted computing power demand of the platform, and the total predicted computing power demand for user services. The total computing power allocation for user service processing is the total computing power allocated to user service processing, and the total computing power allocation for the digital twin platform itself is the total computing power allocated to the digital twin platform itself. It should be understood that the sum of the total computing power allocation for user service processing and the total computing power allocation for the digital twin platform itself in this embodiment is less than or equal to the total computing power information of the base station. That is, this embodiment is equivalent to generating a computing power allocation scheme that balances the computing power demand of the digital twin platform and the computing power demand of user services, taking into account the total computing power limitations of the 5G base station and the predicted computing power demand of the digital twin platform and user services.

[0065] The computing power allocation module 5 in this embodiment can use existing computing power allocation methods to allocate computing power to the tasks and user services of the digital twin platform according to the computing power allocation scheme.

[0066] This application provides a dynamic prediction and resource allocation system for 5G base station computing power based on digital twins. First, it predicts the computing power requirements of the digital twin platform and user services separately. Then, it comprehensively considers the total computing power limitations of the 5G base station and the predicted computing power requirements of the digital twin platform and user services to generate a computing power allocation scheme that balances the computing power requirements of the digital twin platform and user services. Finally, it allocates computing power to the digital twin platform and user services according to this scheme. Because this application explicitly incorporates the computing power requirements of the digital twin platform itself into the computing power prediction and allocation process, and because it can allocate appropriate computing power to the digital twin platform and user services based on the total computing power information of the base station, the total predicted computing power requirements of the platform, and the total predicted computing power requirements of user services, this application can effectively solve the problems of insufficient computing power allocated to the digital twin platform leading to untimely updates of the virtual model or decreased prediction accuracy of computing power requirements, affecting the accuracy and reliability of computing power allocation decisions; and excessive computing power allocated to the digital twin platform leading to a large gap between the actual and expected computing power allocated to user services, severely impacting user service quality.

[0067] In some preferred embodiments, the process of generating a computing power allocation scheme that includes the total computing power allocation for user service processing and the total computing power allocation for the digital twin platform itself, based on the total computing power information of the base station, the total predicted computing power demand of the platform, and the total predicted computing power demand for user services, includes:

[0068] A1. Analyze whether the sum of the predicted total computing power demand of the platform and the predicted total computing power demand of user services is less than or equal to the total computing power information of the base station. If yes, proceed to step A2; otherwise, proceed to step A3.

[0069] A2. The total predicted computing power demand of the platform is used as the total computing power allocation for the digital twin platform itself, and the total predicted computing power demand of user services is used as the total computing power allocation for user service processing, to obtain the computing power allocation scheme v.

[0070] A3. Reduce the total predicted computing power demand of the platform and / or the total predicted computing power demand of user services until the sum of the total predicted computing power demand of the platform and the total predicted computing power demand of user services is less than or equal to the total computing power information of the base station. Then, use the total predicted computing power demand of the platform at this time as the total computing power allocation of the digital twin platform itself and use the total predicted computing power demand of user services at this time as the total computing power allocation of user services to obtain the computing power allocation scheme.

[0071] Step A1 is equivalent to comparing the predicted total computing power requirements of the digital twin platform and user services with the maximum computing power of the 5G base station. Specifically, if the sum of the predicted total computing power requirements of the platform and the predicted total computing power requirements of user services is less than or equal to the total computing power information of the base station, it means that the available computing power of the 5G base station can simultaneously meet the predicted computing power requirements of the digital twin platform and the predicted computing power requirements of user services. In this case, step A2 can be executed to directly use the total predicted computing power requirements of the platform as the total computing power allocation for the digital twin platform itself and the total predicted computing power requirements of user services as the total computing power allocation for user service processing. If the sum of the predicted total computing power requirements of the platform and the predicted total computing power requirements of user services is greater than the total computing power information of the base station, it means that the available computing power of the 5G base station can meet the predicted computing power requirements of both the digital twin platform and user services. The method simultaneously satisfies the predicted computing power requirements of the digital twin platform and the predicted computing power requirements of user services. In this case, step A2 needs to be executed to reduce the total predicted computing power requirements of the platform and / or the total predicted computing power requirements of user services, and to use the adjusted total predicted computing power requirements of the platform as the total computing power allocation for the platform itself and the adjusted total predicted computing power requirements of user services as the total computing power allocation for user services. This ensures that the total computing power allocation (the sum of the total computing power allocation for user services and the total computing power allocation for the digital twin platform itself) does not exceed the maximum computing power of the 5G base station. This effectively avoids the situation where the computing power allocation scheme cannot be executed due to the total computing power allocation exceeding the maximum computing power of the 5G base station, thus ensuring the executability of the computing power allocation scheme.

[0072] In some preferred embodiments, step A3 includes:

[0073] A31. Calculate the reduction in computing power based on the sum of the total predicted computing power demand of the platform and the total predicted computing power demand of user services, and the total computing power information of the base station.

[0074] A32. Obtain a first user service set and a first platform task set, wherein the first user service set includes at least one first service type and the first platform task set includes at least one first task type;

[0075] A33. Based on the first business type, query the pre-built mapping relationship table between the first business type and the priority score to determine the business priority score corresponding to each first business type; and based on the first task type, query the pre-built mapping relationship table between the first task type and the priority score to determine the task priority score corresponding to each first task type.

[0076] A34. Calculate the total business priority score based on all business priority scores, and calculate the total task priority score based on all task priority scores.

[0077] A35. Determine the user business computing power reduction ratio and the platform task computing power reduction ratio based on the total score of business priority and the total score of task priority. The sum of the user business computing power reduction ratio and the platform task computing power reduction ratio is 1.

[0078] A36. Calculate the reduction in computing power for user services based on the reduction amount and the reduction ratio of computing power for user services, and calculate the reduction in computing power for the platform based on the reduction amount and the reduction ratio of computing power for platform tasks.

[0079] A37. Reduce the total predicted computing power demand of the platform by the amount of platform computing power reduction and reduce the total predicted computing power demand of user services by the amount of user service computing power reduction. Then, use the total predicted computing power demand of the platform at this time as the total computing power allocation of the digital twin platform itself and use the total predicted computing power demand of user services at this time as the total computing power allocation of user service processing to obtain a computing power allocation scheme.

[0080] Step A31 can calculate the computing power reduction by subtracting the sum of the predicted total demand for platform computing power and the predicted total demand for user service computing power from the base station's total computing power information. Since different types of user services use different applications, services, or traffic, step A32 can determine the service type corresponding to the user service by analyzing the various applications or services (e.g., video streaming, online games, web browsing) used by the user through the 5G base station. Step A32 can also determine the service type corresponding to the user service by analyzing the user service's traffic usage. Since there are usually multiple user services, the first user service set obtained in step A32 includes at least one first service type. Because the tasks running within the digital twin platform are pre-divided into different task types, and different types of tasks have different resource consumption (e.g., CPU utilization, memory usage, data processing volume), the first task type in step A32 can be a pre-set value. Step A32 can also determine the task type corresponding to the task by analyzing the task's resource consumption. Since the digital twin platform usually executes multiple tasks, the first platform task set obtained in step A32 includes at least one first task type. Step A34 calculates the total business priority score by summing all business priority scores, and calculates the total task priority score by summing all task priority scores. Step S35 determines the user business computing power reduction ratio and the platform task computing power reduction ratio based on the total business priority score and the total task priority score, and the sum of the user business computing power reduction ratio and the platform task computing power reduction ratio is 1. Specifically, the process for determining the user business computing power reduction ratio and the platform task computing power reduction ratio in step S35 can be as follows: normalize the total business priority score and the total task priority score; use the normalized total business priority score as the platform task computing power reduction ratio, and use the normalized total task priority score as the user business computing power reduction ratio. For example, if the total business priority score is 60 points and the total task priority score is 40 points, then the user business computing power reduction ratio is 0.4 and the platform task computing power reduction ratio is 0.6. Step A36 can calculate the reduction in computing power for user services by multiplying the reduction in computing power by the reduction ratio of computing power for user services, and calculate the reduction in computing power for the platform by multiplying the reduction in computing power by the reduction ratio of computing power for platform tasks. Since this embodiment is equivalent to dynamically adjusting the total allocation of computing power for user service processing and the total allocation of computing power for the digital twin platform itself based on the priority of user services and the priority of tasks when the available computing power of the 5G base station cannot simultaneously meet the predicted computing power requirements of the digital twin platform and the predicted computing power requirements of user services, this embodiment can allocate more computing power to higher-priority tasks (user services or tasks) to avoid excessive impact on high-priority tasks, thereby effectively improving the rationality and reliability of computing power allocation.

[0081] In some preferred embodiments, the first platform task set further includes at least one first task quantity, each first task quantity corresponding to a first task type; the first user service set further includes at least one first service quantity, each first service quantity corresponding to a first service type; step A34 includes:

[0082] A341. Calculate the first preliminary priority score for each first business type based on the business priority score and the number of first businesses, and then calculate the total business priority score based on all the first preliminary priority scores.

[0083] A342. Calculate the second preliminary priority score corresponding to each first task type based on the task priority score corresponding to the first task type and the number of first tasks, and then calculate the total task priority score based on all the second preliminary priority scores.

[0084] Since the number of user services is related to the computing power required by the user services, and the number of tasks is related to the computing power required by the tasks, this embodiment is equivalent to introducing the consideration of the number of services and tasks when calculating the total score of service priority and the total score of task priority. Therefore, this embodiment can effectively avoid the situation where the evaluation of service priority and task priority is inaccurate due to ignoring the quantitative factors, and the accuracy of the total score of service priority and the total score of task priority decreases, thereby further improving the rationality and reliability of computing power allocation.

[0085] In some preferred embodiments, step A342 includes:

[0086] A3421. Obtain the operating status of the digital twin platform, and then query the pre-built mapping relationship table of platform operating status and scoring compensation coefficient combination based on the operating status of the digital twin platform to determine the scoring compensation coefficient corresponding to different first task types.

[0087] A3422. Calculate the second preliminary priority score corresponding to each first task type based on the task priority score, the number of first tasks, and the score compensation coefficient. Then, calculate the total task priority score based on all the second preliminary priority scores.

[0088] This embodiment can utilize existing operational status monitoring tools or technologies to obtain the operational status of the digital twin platform. Alternatively, it can obtain the operational status by analyzing the platform's operational logs. The mapping table between platform operational status and scoring compensation coefficient combinations stores different platform operational statuses and their corresponding scoring compensation coefficient combinations. Each scoring compensation coefficient combination includes multiple scoring compensation coefficients, and each coefficient corresponds to a first task type. Step S3422 calculates the second preliminary priority score corresponding to each first task type by multiplying the task priority score corresponding to the first task type, the number of first tasks, and the scoring compensation coefficient. Because the relative importance of various tasks within a digital twin platform changes under different operating states—for example, when the platform is under high load or unstable—the priority of tasks such as data synchronization and status detection needs to be appropriately increased to ensure stable operation and data consistency, while the priority of tasks such as model updates and optimizations can be appropriately reduced to free up more computing power for critical tasks—this embodiment, by introducing the operating state of the digital twin platform as a factor influencing task priority calculation, enables the calculated total task priority score to more accurately reflect the actual needs and importance of various tasks within the platform under the current state. This prioritizes tasks that are intuitively important to the current operating state of the digital twin platform when computing power is limited, thereby effectively improving the overall performance and stability of the digital twin platform and further enhancing the rationality and reliability of computing power allocation.

[0089] In some preferred embodiments, step A31 includes:

[0090] A311. Calculate the computing power reduction amount based on the sum of the platform's predicted total computing power demand and the user's predicted total computing power demand, the preset computing power buffer, and the base station's total computing power information.

[0091] The calculation formula for the computing power reduction amount in this embodiment is: Computing power reduction amount = Total predicted platform computing power demand + Total predicted user service computing power demand + Preset computing power buffer amount - Total base station computing power information. The preset computing power buffer amount in this embodiment can be a fixed value set by those skilled in the art based on experience or actual needs. Preferably, the preset computing power buffer amount in this embodiment is determined based on the total predicted platform computing power demand and the total predicted user service computing power demand. Specifically, this embodiment can determine the preset computing power buffer amount by querying a pre-built mapping table of platform computing power demand, user service computing power demand, and computing power buffer amount based on the total predicted platform computing power demand and the total predicted user service computing power demand. That is, this embodiment is equivalent to dynamically determining the preset computing power buffer amount based on the predicted computing power demand, so as to reserve sufficient computing power to cope with emergencies while meeting the computing power needs of user services and digital twin platform tasks as much as possible. Because this embodiment calculates the computing power reduction based on the sum of the platform's predicted total computing power demand and the user service's predicted total computing power demand, the preset computing power buffer, and the base station's total computing power information, this embodiment ensures that the final total computing power allocation for user service processing and the total computing power allocation for the digital twin platform itself will not reach the 5G base station's total computing power limit. Instead, it reserves a preset computing power buffer to improve the ability of the 5G base station's dynamic computing power prediction and resource allocation system based on digital twins to cope with emergencies.

[0092] In some preferred embodiments, the step of allocating computing power to the tasks and user services of the digital twin platform according to the computing power allocation scheme includes:

[0093] B1. Obtain the second user service set and the second platform task set. The second user service set includes at least one set of second service types and their corresponding number of second services. The second platform task set includes at least one set of second task types and their corresponding number of second tasks.

[0094] B2. Based on the second business type, query the pre-built mapping table of the second business type and priority score to determine the business priority score corresponding to each second business type, and based on the second task type, query the pre-built mapping table of the second task type and priority score to determine the task priority score corresponding to each second task type.

[0095] B3. Calculate the third preliminary priority score for each second business type based on the business priority score and the number of second business types. Then, normalize all the third preliminary priority scores to determine the first weighted weight for each second business type.

[0096] B4. Calculate the fourth preliminary priority score corresponding to each second task type based on the task priority score and the number of second tasks. Then, normalize all the fourth preliminary priority scores to determine the second weighted weight corresponding to each second task type.

[0097] B5. Allocate computing power to user services based on the total computing power allocation for user business processing and all first weighted weights, and allocate tasks to the digital twin platform based on the total computing power allocation for the digital twin platform itself and all second weighted weights.

[0098] Steps B1 and B2 are similar to steps A31-A33 and A341 and A342 in the above embodiments, and will not be discussed in detail here. The specific process of step B5 can be as follows: calculate the first computing power allocation corresponding to each second business type based on the total computing power allocation for user business processing and the first weighted weight, and calculate the second computing power allocation corresponding to each second task type based on the total computing power allocation for the digital twin platform itself and the second weighted weight; for each second business type, calculate the computing power allocation for the user business corresponding to the second business type based on the first computing power allocation and the number of businesses, and then allocate computing power of the size of the computing power allocation to each user business corresponding to the second business type; for each second task type, calculate the computing power allocation for the task corresponding to the second task type based on the second computing power allocation and the number of tasks, and then allocate computing power of the size of the computing power allocation to each task corresponding to the second task type. This embodiment can prioritize the allocation of computing power to higher-priority user services or tasks by weighting the allocation of computing power to various types of user services based on service priority scores and service quantity, and by weighting the allocation of computing power to various types of tasks based on task priority scores and task quantity. Therefore, this embodiment can effectively improve the effectiveness and flexibility of computing power allocation.

[0099] In some preferred embodiments, the step of generating the predicted total platform computing power demand based on historical operational datasets includes:

[0100] C1. Use the interquartile range method to identify and remove outliers from historical running datasets;

[0101] C2. Generate the total predicted computing power demand of the platform based on historical operating datasets.

[0102] The interquartile range (ICM) method in this embodiment is an existing algorithm. This ICM method determines the distribution range of the data by calculating the quartiles of the dataset and marks data points outside this range as outliers. Therefore, this embodiment can use the ICM method to identify and remove outliers from historical running datasets. Since this embodiment first uses the ICM method to identify and remove outliers from historical running datasets before generating the total predicted platform computing power demand, it makes the data in the historical running datasets more representative, thereby effectively improving the accuracy and reliability of the total predicted platform computing power demand. Preferably, the process of generating the total predicted user business computing power demand based on the historical user computing power demand set in this embodiment is as follows: identifying and removing outliers from the historical user computing power demand set using the ICM method; generating the total predicted user business computing power demand based on the historical user computing power demand set; this embodiment can effectively improve the accuracy and reliability of the total predicted user business computing power demand.

[0103] In some preferred embodiments, all user service computing power demands in the historical user service computing power demand set correspond to the same time period. This embodiment achieves this by marking the total user service computing power demand with time periods during collection or storage, and then filtering the total user service computing power demand with the corresponding time period markings based on the target prediction time period when constructing the historical user service computing power demand set. This ensures that all user service computing power demands in the historical user service computing power demand set correspond to the same time period. This embodiment is equivalent to maintaining consistency in the time period of all user service computing power demands in the historical user service computing power demand set by making them correspond to the same time period. Since user service computing power demands within the same time period are correlated and similar, this embodiment effectively avoids the situation where the accuracy and reliability of the predicted user service computing power demand decreases because the total user service computing power demands in the historical user service computing power demand set come from different time periods, and the user service computing power demands in different time periods are not correlated and are not similar. Preferably, in this embodiment, all internal operating data of the digital twin platform in the historical operating dataset corresponds to the same time period.

[0104] In some preferred embodiments, the internal operating data of the digital twin platform includes processing unit occupancy, memory usage, and data processing volume.

[0105] As can be seen from the above, the 5G base station computing power dynamic prediction and resource allocation system based on digital twins provided in this application first predicts the computing power requirements of the digital twin platform and user services respectively. Then, it comprehensively considers the total computing power limit of the 5G base station and the predicted computing power requirements of the digital twin platform and user services to generate a computing power allocation scheme that balances the computing power requirements of the digital twin platform and user services. Finally, it allocates computing power to the digital twin platform and user services according to the computing power allocation scheme. Since this application explicitly incorporates the computing power requirements of the digital twin platform itself into the computing power prediction and allocation process, and this application can allocate appropriate computing power to the digital twin platform and user services based on the total computing power information of the base station, the total predicted computing power requirements of the platform, and the total predicted computing power requirements of user services, this application can effectively solve the problems of insufficient computing power allocated to the digital twin platform leading to the virtual model not being updated in a timely manner or the prediction accuracy of computing power requirements decreasing, affecting the accuracy and reliability of computing power allocation decisions, and excessive computing power allocated to the digital twin platform leading to a large gap between the actual computing power allocated to user service processing and the expected computing power, seriously affecting the quality of user services.

[0106] In the embodiments provided in this application, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of the above units is only a logical functional division, and there may be other division methods in actual implementation. Furthermore, multiple units or components may be combined or integrated into another robot, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interface; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0107] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0108] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0109] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A 5G base station computing power dynamic prediction and resource allocation system based on digital twins, characterized in that, The 5G base station computing power dynamic prediction and resource allocation system based on digital twins includes: The platform computing power demand prediction module is used to generate the predicted total platform computing power demand based on the historical operating dataset. The historical operating dataset is a collection of all internal operating data of the digital twin platform within a first preset time period before the current time node. The user business computing power demand total prediction module is used to generate the user business computing power predicted total demand based on the historical user computing power demand total set, wherein the historical user computing power demand total set is the set of all user business computing power demand totals within the second preset time period before the current time node. The base station total computing power acquisition module is used to acquire pre-set base station total computing power information; The allocation scheme generation module is used to generate a computing power allocation scheme that includes the total computing power allocation for user service processing and the total computing power allocation for the digital twin platform itself, based on the total computing power information of the base station, the total predicted computing power demand of the platform, and the total predicted computing power demand for user services. The computing power allocation module is used to allocate computing power to the tasks of the digital twin platform and the user services according to the computing power allocation scheme. The process of generating a computing power allocation scheme that includes the total computing power allocation for user service processing and the total computing power allocation for the digital twin platform itself, based on the total computing power information of the base station, the total predicted computing power demand of the platform, and the total predicted computing power demand for user services, includes: A1. Analyze whether the sum of the predicted total computing power demand of the platform and the predicted total computing power demand of the user services is less than or equal to the total computing power information of the base station. If yes, proceed to step A2; otherwise, proceed to step A3. A2. Take the total predicted computing power demand of the platform as the total computing power allocation of the digital twin platform itself, and take the total predicted computing power demand of user services as the total computing power allocation of user service processing, so as to obtain a computing power allocation scheme. A3. Reduce the total predicted computing power demand of the platform and / or the total predicted computing power demand of the user services until the sum of the total predicted computing power demand of the platform and the total predicted computing power demand of the user services is less than or equal to the total computing power information of the base station. Then, use the total predicted computing power demand of the platform at this time as the total computing power allocation of the digital twin platform itself and use the total predicted computing power demand of the user services at this time as the total computing power allocation of the user services to obtain a computing power allocation scheme.

2. The 5G base station computing power dynamic prediction and resource allocation system based on digital twins according to claim 1, characterized in that, Step A3 includes: A31. Calculate the computing power reduction amount based on the sum of the predicted total computing power demand of the platform and the predicted total computing power demand of user services, and the total computing power information of the base station; A32. Obtain a first user service set and a first platform task set, wherein the first user service set includes at least one first service type and the first platform task set includes at least one first task type; A33. Based on the first business type, query the pre-built mapping relationship table of the first business type and priority score to determine the business priority score corresponding to each of the first business types, and based on the first task type, query the pre-built mapping relationship table of the first task type and priority score to determine the task priority score corresponding to each of the first task types. A34. Calculate the total business priority score based on all the business priority scores, and calculate the total task priority score based on all the task priority scores. A35. Determine the user business computing power reduction ratio and the platform task computing power reduction ratio based on the total score of the business priority and the total score of the task priority, wherein the sum of the user business computing power reduction ratio and the platform task computing power reduction ratio is 1. A36. Calculate the user service computing power reduction amount based on the computing power reduction amount and the user service computing power reduction ratio, and calculate the platform computing power reduction amount based on the computing power reduction amount and the platform task computing power reduction ratio; A37. Reduce the total predicted computing power demand of the platform by the amount of platform computing power reduction and reduce the total predicted computing power demand of user services by the amount of user service computing power reduction. Then, use the total predicted computing power demand of the platform at this time as the total computing power allocation of the digital twin platform itself and use the total predicted computing power demand of user services at this time as the total computing power allocation of user service processing to obtain a computing power allocation scheme.

3. The 5G base station computing power dynamic prediction and resource allocation system based on digital twins according to claim 2, characterized in that, The first platform task set further includes at least one first task quantity, each first task quantity corresponding to one first task type; the first user service set further includes at least one first service quantity, each first service quantity corresponding to one first service type; step A34 includes: A341. Calculate the first preliminary priority score corresponding to each of the first business types based on the business priority score corresponding to the first business type and the number of first businesses, and then calculate the total business priority score based on all the first preliminary priority scores. A342. Calculate the second preliminary priority score corresponding to each of the first task types based on the task priority score corresponding to the first task type and the number of first tasks, and then calculate the total task priority score based on all the second preliminary priority scores.

4. The 5G base station computing power dynamic prediction and resource allocation system based on digital twins according to claim 3, characterized in that, Step A342 includes: A3421. Obtain the operating status of the digital twin platform, and then query a pre-built mapping table of platform operating status and scoring compensation coefficient combination based on the operating status of the digital twin platform to determine the scoring compensation coefficient corresponding to different first task types. A3422. Calculate the second preliminary priority score corresponding to each of the first task types based on the task priority score, the number of first tasks, and the score compensation coefficient, and then calculate the total task priority score based on all the second preliminary priority scores.

5. The 5G base station computing power dynamic prediction and resource allocation system based on digital twins according to claim 2, characterized in that, Step A31 includes: A311. Calculate the computing power reduction amount based on the sum of the total predicted computing power demand of the platform and the total predicted computing power demand of the user services, the preset computing power buffer amount, and the total computing power information of the base station.

6. The 5G base station computing power dynamic prediction and resource allocation system based on digital twins according to claim 1, characterized in that, The step of allocating computing power to the tasks of the digital twin platform and the user services according to the computing power allocation scheme includes: B1. Obtain a second user service set and a second platform task set. The second user service set includes at least one set of second service types and their corresponding number of second services. The second platform task set includes at least one set of second task types and their corresponding number of second tasks. B2. Based on the second business type, query the pre-built mapping relationship table of the second business type and priority score to determine the business priority score corresponding to each of the second business types, and based on the second task type, query the pre-built mapping relationship table of the second task type and priority score to determine the task priority score corresponding to each of the second task types. B3. Calculate the third preliminary priority score corresponding to each of the second business types based on the business priority score and the number of second business types, and then normalize all the third preliminary priority scores to determine the first weighted weight corresponding to each of the second business types. B4. Calculate the fourth preliminary priority score corresponding to each of the second task types based on the task priority score and the number of second tasks, and then normalize all the fourth preliminary priority scores to determine the second weighted weight corresponding to each of the second task types. B5. Allocate computing power to the user service based on the total computing power allocation for user service processing and all the first weighted weights, and allocate tasks of the digital twin platform based on the total computing power allocation for the digital twin platform itself and all the second weighted weights.

7. The 5G base station computing power dynamic prediction and resource allocation system based on digital twins according to claim 1, characterized in that, The step of generating the predicted total platform computing power demand based on historical operational datasets includes: C1. Use the interquartile range method to identify and remove outliers from historical running datasets; C2. Generate the total predicted computing power demand of the platform based on the historical operating dataset.

8. The 5G base station computing power dynamic prediction and resource allocation system based on digital twins according to claim 1, characterized in that, The total computing power demand of all users in the historical user computing power demand set corresponds to the same time period.

9. The 5G base station computing power dynamic prediction and resource allocation system based on digital twins according to claim 1, characterized in that, The internal operating data of the digital twin platform includes processing unit occupancy, memory usage, and data processing volume.

Images