5G base station energy-saving control method and system

Abstract

The invention discloses a 5G base station energy-saving control method and system, relates to the technical field of communication, and provides a 5G base station energy-saving control method for solving the problem that resources are overused or idle in the prior art, and the method comprises the steps: monitoring a real-time load; analyzing business requirements; intelligent decision making; if it is judged that the component meets the dormancy condition, sending a dormancy instruction to the component to enable the component to enter a dormancy state; if it is judged that the component meets the wake-up condition, a wake-up instruction is sent to the component, so that the component returns to the normal working state. According to the method, whether various hardware components in the base station meet the dormancy or awakening condition or not is accurately judged through the intelligent decision algorithm, dynamic optimal configuration of hardware resources is achieved, and overuse or idling of the hardware resources is effectively avoided. According to different types of hardware components and business requirements, the weight coefficient and the fuzzy membership function of each factor are flexibly adjusted, the prediction accuracy and reliability are improved, and the expandability and the flexibility are high.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and particularly to a 5G base station energy-saving control method and system. Background Art

[0002] With the rapid development of 5G communication technologies, as the core facilities supporting the operation of 5G networks, the energy consumption problem of 5G base stations has become increasingly prominent. 5G base stations not only need to provide higher data transmission rates, lower latencies, and greater connection densities, but also need to achieve energy conservation and emission reduction and reduce operating costs while meeting these high-performance requirements.

[0003] Chinese invention patent CN114362279A discloses a 5G base station energy-saving control method and system. The 5G base station energy-saving control method includes: detecting the charging state of a storage battery, and judging whether the storage battery is in energy-saving charging according to the charging state of the storage battery; if so, obtaining the power of the storage battery; judging whether the power of the storage battery is greater than a first threshold; if so, triggering the storage battery power supply mode. This application can realize the automatic switching of dual power supplies, supply power to the devices of the 5G base station with the electric energy of the storage battery, charge and supply power at the same time during this period, and switch to the commercial power supply mode when the storage battery is lower than the first threshold, which can reduce power consumption and achieve energy conservation, emission reduction, and cost reduction. However, this method still has obvious deficiencies, that is, it is unable to accurately judge the sleep or wake-up conditions of hardware components, resulting in overuse or idleness of resources. Summary of the Invention

[0004] The purpose of the present invention is to make up for the deficiencies of the prior art, and provides a 5G base station energy-saving control method and system. The system realizes the effective reduction of base station energy consumption and the efficient utilization of resources by monitoring the base station load situation in real time, analyzing and predicting service requirements, and accurately judging the sleep or wake-up conditions of hardware components.

[0005] The present application provides the following technical solutions: A 5G base station energy-saving control method includes: S1, real-time load monitoring: Real-time monitoring of the communication data between the 5G base station and the user equipment to obtain the current load situation of the base station. The communication data includes but is not limited to data traffic, the number of connected users, and service type information. By setting high-precision data collectors at key nodes of the communication link between the base station and the user equipment, data samples are collected at fixed time intervals, and the collected data is preprocessed; S2, service demand analysis: Combining factors such as the priority of network services, real-time service request situations, and historical service rules, analyzing and predicting the service demands that may occur currently and in the short term; S3, Intelligent Decision-making: Based on the results of step S1 and step S2, construct an intelligent decision-making algorithm based on fuzzy logic and multi-objective optimization theory, and use this intelligent decision-making algorithm to determine whether various hardware components in the 5G base station meet the sleep condition or wake-up condition; S4, If it is determined that the component meets the sleep condition, send a sleep instruction to the component to make it enter the sleep state; if it is determined that the component meets the wake-up condition, send a wake-up instruction to the component to make it resume normal operation. The instruction is sent through the internal high-speed communication bus of the base station and encoded and transmitted according to the communication protocol format of the hardware component.

[0006] Furthermore, in step S1, a denoising algorithm based on wavelet transform is used to denoise the collected data, and then the denoised data is decomposed at multiple scales by setting the wavelet basis function and the decomposition level. The high-frequency noise components are removed, and the payload information is retained, thereby further improving the accuracy of the payload data.

[0007] Furthermore, the specific operations of step S2 include: S2.1, Set the priority weights of different types of services from high to low according to the dependence of the service on network resources, real-time requirements, and factors affecting the user experience: emergency call services with extremely high real-time requirements and significant impact on the user experience, high-definition video streaming services, ordinary web browsing services; S2.2, Capture the request initiation time, request quantity, and requested service type of various services in real time, and record the number of requests of each service per unit time; S2.3, Collect and store the historical data of the base station services. These historical data include the service type distribution, service traffic, and user access quantity information in different time periods. Based on the historical data, construct a service demand prediction model to obtain the predicted value of the historical service pattern; S2.4, Assign the weighted coefficients of the network service priority weight, real-time service request quantity, and historical service pattern predicted value through weighted summation respectively to obtain the comprehensive predicted value of the current and short-term possible service demands.

[0008] Furthermore, in step S2.1, if there is a virtual reality / augmented reality service, set its priority weight value between the emergency call service with extremely high real-time requirements and significant impact on the user experience and the high-definition video streaming service. In the network service priority setting of service demand analysis, for virtual reality (VR) / augmented reality (AR) services, due to their extremely high requirements for bandwidth, latency, and stability, the priority weight is set to P4 = 0.85. In addition to considering the characteristics of the business itself, the determination of this weight also comprehensively takes into account the attention of users to the VR / AR business experience in market research and the evaluation results of the importance of such businesses in future development trends. For VR / AR businesses at different levels, the priority weights can be further subdivided based on their specific resource requirement differences on the basis of P 4 to more accurately reflect the business requirements.

[0009] Furthermore, in step S2.2, analyze the source IP address distribution of the business requests. By performing cluster analysis on the source IP addresses of the business requests, users are divided into different regional groups, and the business request frequencies and business type preferences of each regional group are counted. According to the statistical results, the business demand prediction values are defined in segments.

[0010] Furthermore, in step S2.3, the constructed business demand prediction model is implemented using a time series decomposition and adaptive weighted prediction algorithm or a grey prediction model.

[0011] Furthermore, in step S3, the expression of the intelligent decision-making algorithm is:

[0012] where represents the decision value of the k th hardware component, which is used to determine whether the component meets the sleep or wake-up conditions. When , T s is a preset sleep threshold, and it is determined that the component meets the sleep condition. When , T w is a preset wake-up threshold, and it is determined that the component meets the wake-up condition. represents the number of factors participating in the decision-making, including factors related to the load situation and factors related to business requirements. is the weight coefficient of each factor. represents the fuzzy membership function of the l th factor with respect to the k th hardware component, which is used to map the actual value of each factor to the interval [0, 1].

[0013] Further, in the real-time business request monitoring of the S2 service requirement analysis, in addition to recording the number of business requests, the source IP address distribution of business requests is further analyzed. By performing clustering analysis on the source IP addresses of business requests, users are divided into different regional groups, and the business request frequency and business type preferences of each regional group are statistically analyzed. In this way, when predicting service requirements, the regional characteristics can be combined to more accurately estimate the possible service requirement situations in different regions, providing more targeted service requirement inputs for the intelligent decision-making algorithm.

[0014] Further, in the S3 intelligent decision-making algorithm, for different types of hardware components, the weight coefficients of their respective factors can be readjusted according to the characteristics and functional requirements of the components. For the power management unit, since its main responsibility is to distribute and manage the base station power, the influence of the load value on its decision-making is relatively small, while the influence of the service requirement change rate on its decision-making is relatively large. Therefore, the weight coefficient of the load value can be adjusted to = 0.15, and the weight coefficient of the service requirement change rate is adjusted to = 0.3, and the weight coefficients of other factors are also fine-tuned accordingly to ensure that the intelligent decision-making algorithm can more accurately determine whether the component meets the sleep or wake-up conditions.

[0015] Further, in the S4 hardware component control step, when sending a sleep instruction to the hardware component, in addition to encoding and transmitting it according to the normal communication protocol format, a sleep duration estimated value will also be attached to the instruction. This estimated value is calculated through a linear regression model based on the current service requirement prediction situation and the characteristics of the hardware component. For a certain radio frequency unit, let the service requirement prediction value be Q t , and its rated load is C max , then the sleep duration estimated value T h can be calculated by the following formula:

[0016] Among them, k 1 and k 2 are parameters obtained by fitting a large amount of experimental data. By conducting experiments on different combinations of service requirements and hardware component load conditions, recording the actual sleep duration values of each experiment, and performing linear regression analysis, the values of k 1 and k 2 are obtained. In this way, after receiving the sleep instruction, the hardware component can not only enter the sleep state but also reasonably arrange its low-power mode operation time according to the sleep duration estimated value, improving the energy-saving effect.

[0017] Furthermore, in the S4 hardware component control step, when a wake-up instruction is sent to a hardware component, a wake-up priority identifier will be attached to the instruction. This identifier is determined based on the urgency of the service requirements and the importance of the hardware component in meeting the service requirements. For the case where high-bandwidth service transmission is in progress and will be interrupted due to the dormancy of the hardware component, the wake-up priority identifier of the relevant hardware component is set to high. For general service situations, the wake-up priority identifier is set to medium. For hardware components that can resume normal operation at a later time and have little impact on the service, the wake-up priority identifier is set to low. By setting the wake-up priority identifier, when multiple hardware components need to be woken up simultaneously, they can be woken up in sequence according to the priority order, ensuring the continuity and efficiency of the service.

[0018] A 5G base station energy-saving control system, which includes a real-time load monitoring module, a service demand analysis module, an intelligent decision-making algorithm module, and a hardware component control module; The real-time load monitoring module is used to monitor the communication data between the 5G base station and user equipment in real time to obtain the current load situation of the base station and preprocess the collected data; the communication data includes but is not limited to data traffic, the number of connected users, and service type information. By setting high-precision data collectors at key nodes of the communication link between the base station and user equipment, data samples are collected at fixed time intervals, and real-time data processing algorithms are used to preprocess the collected data to remove outliers and noise interference; The service demand analysis module is used to predict the current and short-term possible service demands based on the priority of network services, real-time service requests, and historical service patterns; The intelligent decision-making algorithm module is used to construct an intelligent decision-making algorithm based on the results of the real-time load monitoring module and the service demand analysis module, and use this intelligent decision-making algorithm to determine whether various hardware components in the 5G base station meet the dormancy conditions or wake-up conditions; The hardware component control module, if it determines that a component meets the dormancy conditions, sends a dormancy instruction to the component to make it enter the dormancy state; if it determines that a component meets the wake-up conditions, sends a wake-up instruction to the component to make it resume the normal working state.

[0019] Compared with the prior art, a 5G base station energy-saving control method and system have the following beneficial effects: 1. This application accurately determines whether various hardware components in the base station meet the dormancy or wake-up conditions through an intelligent decision-making algorithm, realizing the dynamic optimization configuration of hardware resources. This method can effectively avoid the overuse or idle of hardware resources, thereby reducing the energy consumption of the base station and achieving a significant energy-saving effect, which has important practical application value.

[0020] 2. This application constructs the theory of fuzzy logic and multi-objective optimization. According to different types of hardware components and business requirements, it flexibly adjusts the weight coefficients and fuzzy membership functions of various factors to adapt to different application scenarios. This method can also combine the historical business data and real-time business request situations of the base station, and use multiple algorithms such as time series decomposition, adaptive weighting, and grey prediction to predict business requirements, improving the accuracy and reliability of the prediction. This high scalability and flexibility enable this method to adapt to the development and changes of future 5G base stations and have broad market application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a flowchart of the energy-saving control method for 5G base stations; Figure 2 is an operation flowchart of the intelligent decision-making algorithm. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0024] Embodiment 1 An energy-saving control method for 5G base stations, as Figure 1 and Figure 2 shown, includes: S1. Real-time monitor the communication data between the 5G base station and the user equipment to obtain the current load situation of the base station, and preprocess the collected data; use a denoising algorithm based on wavelet transform to denoise the collected data, and then perform multi-scale decomposition on the denoised data by setting the wavelet basis function and the decomposition level.

[0025] S2. Predict the current and short-term possible business requirements based on the priority of network services, real-time business requests, and historical business rules; the specific operations include: S2.1. Set the priority weights of different types of services from high to low according to the dependence of the service on network resources, real-time requirements, and factors affecting the user experience: emergency call services with extremely high real-time requirements and significant impact on the user experience, virtual reality / augmented reality services, high-definition video streaming services, and ordinary web browsing services; for emergency call services with extremely high real-time requirements and significant impact on the user experience, set the priority weight as P1 = 0.9, for high-definition video streaming services, set the priority weight as P2 = 0.7 according to their resolution and frame rate requirements, for ordinary web browsing services, set the priority weight as P3 = 0.5. The determination of these weights is based on a large number of market research and comprehensive evaluation of the importance of different services in actual application scenarios. Given the extremely high requirements of virtual reality / augmented reality services for bandwidth, latency, and stability, set the priority weight as P4 = 0.85. The determination of this weight not only considers the characteristics of the service itself but also synthesizes the attention of users to VR / AR service experience in market research and the evaluation results of the importance of such services in future development trends. For different levels of VR / AR services, the priority weights can be further subdivided based on P4 according to their specific resource requirements differences to more accurately reflect the service needs.

[0026] S2.2. Capture the request initiation time, request quantity, and requested service type of various services in real time, and record the number of requests of each type of service per unit time; deploy a real-time service request monitor at the network access port of the base station, which can capture the request initiation time, request quantity, and requested service type information of various services in real time. For each service type, record the number of requests per unit time, denoted as R i , where i represents different service types; S2.3. Collect and store the historical data of base station services, construct a service demand prediction model based on the historical data, and obtain the predicted values of historical service patterns; these historical data include service type distribution, service traffic, and user access quantity information in different time periods. Through in-depth mining and analysis of these historical data; the formula for constructing the service demand prediction model using time series decomposition and adaptive weighted prediction algorithms is as follows:

[0027] Where, Q t represents the predicted value of service demand at the future time t , n and m are the orders of the trend term and seasonal term in time series decomposition respectively, determined according to the periodic characteristics of historical data, and is the adaptive weight coefficient, represents the seasonal component of the business demand in the historical data at time t-i ; T t-i represents the trend component of the business demand in the historical data at time t-i ; I t-j represents the irregular component of the business demand at time t-j ; The basic formula for constructing a business demand prediction model using the grey prediction model is:

[0028] where represents the original business demand time series data, represents the new sequence generated by performing a first-order accumulation on the original data, a and b are the parameters obtained by fitting through the least squares method. The determination process is as follows: Substitute the original business demand time series data into the basic formula of the grey prediction model, and solve for the parameters a and b through the least squares method to minimize the sum of the squared errors between the predicted values and the actual values. By performing weighted fusion on the results of the two prediction methods, the accuracy of business demand prediction can be further improved.

[0029] S2.4, Assign the network service priority weight, the number of real-time service requests, and the weighted coefficients of the historical service pattern prediction values through weighted summation to obtain the comprehensive prediction value of the current and short-term possible business demands: The weighted coefficients of the network service priority weight, the number of real-time service requests, and the historical service pattern prediction values are respectively set as , , , and , The determination of the weighted coefficients is based on a large number of simulation experiments and the quantitative evaluation of the importance of different factors in business demand prediction. Set = 0.4, = 0.3, = 0.3.

[0030] S3, Based on the results of step S1 and the results of step S2, construct an intelligent decision-making algorithm based on fuzzy logic and multi-objective optimization theory, and use this intelligent decision-making algorithm to determine whether various hardware components in the 5G base station meet the sleep condition or the wake-up condition;

[0031] where represents the kThe decision value of a hardware component, which is used to determine whether the component meets the sleep or wake-up condition. When is T s is the preset sleep threshold, it is determined that the component meets the sleep condition. When is T w is the preset wake-up threshold, it is determined that the component meets the wake-up condition. represents the number of factors involved in the decision-making, including factors related to the load situation and factors related to business requirements. is the weight coefficient of each factor. represents the l th factor's fuzzy membership function with respect to the k th hardware component, which is used to map the actual value of each factor to the interval [0, 1]. represents the l th factor's fuzzy membership function with respect to the k th hardware component.

[0032] The fuzzy membership function serves as the core input factor for the decision value D k ; C k The value of may affect the dynamic allocation of the factor weight D k ; By comparing with the threshold through

[0033] S4, if it is determined that the component meets the sleep condition, a sleep instruction is sent to the component to make it enter the sleep state; if it is determined that the component meets the wake-up condition, a wake-up instruction is sent to the component to make it resume the normal working state.

[0034] When sending a sleep instruction to a hardware component, in addition to encoding and transmitting it according to the normal communication protocol format, a predicted sleep duration will also be attached to the instruction. This predicted value is calculated through a linear regression model based on the current business demand prediction situation and the characteristics of the hardware component. For a certain radio frequency unit, assuming the predicted value of the business demand is Q t and its rated load is C max , then the predicted sleep duration T h can be calculated by the following formula:

[0035] where k 1 and k2 is a parameter obtained by fitting a large amount of experimental data. By conducting experiments on combinations of different service requirements and hardware component load conditions, recording the actual values of the sleep duration for each experiment, and performing linear regression analysis, we obtain k 1 and k the values of 2. In this way, after receiving the sleep instruction, the hardware components can not only enter the sleep state but also reasonably arrange the operation time of their low-power modes according to the estimated sleep duration, improving the energy-saving effect.

[0036] When sending a wake-up instruction to a hardware component, a wake-up priority identifier is attached to the instruction. This identifier is determined based on the urgency of the service requirement and the importance of the hardware component in meeting the service requirement. For the situation where high-bandwidth service transmission is in progress and will be interrupted due to the sleep of the hardware component, the wake-up priority identifier of the relevant hardware component is set to high. For general service situations, the wake-up priority identifier is set to medium. For hardware components that can resume normal operation at a later time and have less impact on the service, the wake-up priority identifier is set to low. By setting the wake-up priority identifier, when multiple hardware components need to be woken up simultaneously, they can be woken up in sequence according to the priority order, ensuring the continuity and efficiency of the service.

[0037] Embodiment 2 A 5G base station energy-saving control system includes: A real-time load monitoring module, which is used to monitor the communication data between the 5G base station and user equipment in real time to obtain the current load situation of the base station and preprocess the collected data; A service requirement analysis module, which is used to predict the current and short-term possible service requirements based on the priority of network services, real-time service requests, and historical service rules; An intelligent decision-making algorithm module, which is used to construct an intelligent decision-making algorithm based on the results of the real-time load monitoring module and the service requirement analysis module, and based on fuzzy logic and multi-objective optimization theory, and use this intelligent decision-making algorithm to determine whether various hardware components in the 5G base station meet the sleep conditions or wake-up conditions; A hardware component control module. If it is determined that a component meets the sleep condition, it sends a sleep instruction to the component to make it enter the sleep state; if it is determined that a component meets the wake-up condition, it sends a wake-up instruction to the component to make it resume the normal working state.

[0038] Application Example 1 This application example describes that in the central area of the city, during the morning and evening rush hours every day (for example, from 7:00 to 9:00 in the morning and from 5:00 to 7:00 in the evening on weekdays), due to a large number of users using the 5G network for high-data-flow activities simultaneously, the load of the 5G base station rises sharply. In order to achieve energy saving while ensuring the user experience, the 5G base station energy-saving control method is adopted.

[0039] Set high-precision data collectors at key nodes of the communication link between the base station and the user equipment to collect data samples in seconds. Assume that at 8 o'clock during the morning rush hour on a certain working day, the collected data is as follows: the data traffic reaches 500 MB within 10 seconds, the number of connected users is 2,000, and the business types mainly concentrate on high-definition video streaming services (accounting for 60%), online meeting services (accounting for 30%), and other services (accounting for 10%). Use a denoising algorithm based on wavelet transform to preprocess the data, removing outliers and noise interference to ensure the accuracy and reliability of the data.

[0040] According to the results of market research, set the priority weight of the high-definition video streaming service to 0.7 and the priority weight of the online meeting service to 0.8 because these services have a significant impact on the user experience and high real-time requirements. Through a real-time service request monitor, capture the request initiation time, request quantity, and type of various services. Within the 10 minutes from 8 o'clock to 8:10, the number of requests for the high-definition video streaming service is 1,500 times, the number of requests for the online meeting service is 800 times, and the number of requests for other services is 200 times.

[0041] Combined with historical business data, adopt a prediction algorithm of time series decomposition and adaptive weighting to construct a business demand prediction model. Assume that by analyzing the historical data of the same period in the past week, determine that the order of the trend term n in time series decomposition = 3, the order of the seasonal term m = 2, after calculation and adjustment, the adaptive weight coefficient = 0.4, = 0.3, = 0.2, = 0.1, = 0.05. Through model prediction, the predicted value of business demand from 8:10 to 8:20 Q t is 1,200 (the predicted value of business demand here is a quantitative value after comprehensively considering various services and can be converted according to actual business volume and other factors). Deeply analyze the source IP address distribution of business requests, and divide users into 5 regional groups through cluster analysis. Among them, the A regional group has the highest business request frequency, accounting for 40% of the total request times, and the business type preference of this regional group is mainly high-definition video streaming services (accounting for 70%).

[0042] Based on fuzzy logic and multi-objective optimization theory, construct an intelligent decision-making algorithm. Assume that the number of factors L participating in the decision-making = 5, l when = 1 is the current load value, l when = 2 is the predicted value of business demand, l when = 3 is the business type priority, lWhen it is 4, it is the service request frequency, l When it is 5, it is the service demand change rate, and the weight coefficients of each factor are respectively: = 0.2 (weight of load value), = 0.3 (weight of service demand prediction value), = 0.15 (weight of service type priority), = 0.15 (weight of service request frequency), = 0.2 (weight of service demand change rate). For a certain hardware component (such as a radio frequency unit), its rated load is 1000 (the unit can be set according to the actual situation, such as Mbps), the current load value is 700, and the minimum load value for normal operation is , and the decision value of this hardware component is calculated through an intelligent decision-making algorithm D k : , among which, the fuzzy membership function of the load value is:

[0043] Substitute C 1k = 700, C 1max = 1000, C 1min = 200 to get .

[0044] Assume that the fuzzy membership values after calculation and mapping of other factors are respectively: F2(C 2k , B k ) = 0.6 (related to service demand prediction value), F3(C 3k , B k ) = 0.7 (related to service type priority, corresponding to the priority weight mapping of high-definition video streaming service), F4(C 4k , B k ) = 0.5 (related to service request frequency), F5(C 5k , B k ) = 0.4 (related to service demand change rate, assumed to be calculated based on recent service demand changes). Through this piecewise definition method, the impact of service demand prediction values in different intervals on the decision-making of hardware components can be more carefully reflected, improving the accuracy of the intelligent decision-making algorithm.

[0045] Then .

[0046] The preset sleep threshold T s = 0.4, the wake-up threshold T w = 0.6, becauseD k = 0.515 > T s and D k < T w The component currently does not meet the sleep or wake-up conditions, which means that the operating state of the component does not need to be adjusted and no processing is performed temporarily.

[0047] If the intelligent decision-making algorithm determines that a component meets the sleep condition, a sleep instruction is sent to the component to make it enter the sleep state. The sleep instruction includes an estimated sleep duration. Suppose for this radio frequency unit, the business demand prediction value Q t = 1200, its rated load is 1000, and the parameters obtained by fitting through a linear regression model k 1 = 0.01, k 2 = 5, then the estimated sleep duration minutes.

[0048] If it is determined that a component meets the wake-up condition, a wake-up instruction is sent to the component to make it resume normal operation. The wake-up instruction includes a wake-up priority flag, which is determined according to the urgency of the business demand and the importance of the hardware component in meeting the business demand. For example, for the case of high-definition video streaming service transmission that is about to be interrupted due to the sleep of the hardware component, the wake-up priority flag of the relevant hardware component is set to high. The instruction is sent through the internal high-speed communication bus of the base station and encoded and transmitted according to the communication protocol format of the hardware component to ensure the accuracy and timeliness of the instruction.

[0049] Application Example 2 This application example describes that in remote areas, due to the sparse distribution of users, the daily load of 5G base stations is usually low. However, in order to meet occasional high-load business demands, the base station needs to maintain a high operating power to ensure business continuity. In order to achieve energy saving while ensuring business continuity, the following 5G base station energy-saving control method is adopted.

[0050] Data collectors are set at key nodes of the communication link between the base station and the user equipment to collect data samples in minutes. Suppose at 14:00 on a certain day, the collected data is as follows: the data traffic is 10 MB within 1 minute, the number of connected users is 50, and the main service types are ordinary web browsing services (accounting for 70%), voice call services (accounting for 20%), and other services (accounting for 10%). Considering the particularity of the communication environment in remote areas, the data collector should have high stability and anti-interference ability. A denoising algorithm based on wavelet transform is used to preprocess the data to remove outliers and noise interference caused by environmental factors or equipment failures, ensuring the accuracy and reliability of the data.

[0051] Based on historical service data, deeply analyze the service rules in remote areas. Since the user distribution is sparse and the service demands usually have large fluctuations and uncertainties, a method combining time series decomposition, adaptive weighting, and grey prediction model is adopted to construct a service demand prediction model. Suppose through the analysis of historical data in the past month, the order of the trend term , the order of the seasonal term , the adaptive weight coefficient = 0.5, = 0.3, = 0.2. At the same time, the grey prediction model is used to assist in predicting the service demand. Suppose the original service demand time series data , , (the data here is the quantified value converted according to the service volume, etc.). The parameters a = -0.1, b = 55 are obtained by least squares fitting. Through weighted fusion of the results of the two prediction methods (assuming the weights of the time series decomposition and adaptive weighting prediction results are 0.6, and the weight of the grey prediction model result is 0.4), the predicted value Q t of the service demand from 14:10 to 14:20 is 55 (also the comprehensive quantified value).

[0052] Based on fuzzy logic and multi-objective optimization theory, an intelligent decision-making algorithm is constructed. Suppose the number of factors L participating in the decision-making is 4, l = 1 is the current load value, l = 2 is the predicted value of the service demand, l = 3 is the service type priority, l = 4 is the service request frequency. The weight coefficients of each factor are respectively: = 0.15 (load value weight), = 0.35 (predicted value weight of the service demand), = 0.2 (Business type priority weight), = 0.3 (Business request frequency weight). For a key hardware component of a radio frequency unit, its rated load is 500 (the unit can be set according to the actual situation, such as Mbps), the current load value is 100, and the minimum load value for normal operation is 50. Calculate the decision value of this hardware component through an intelligent decision-making algorithm D k : Among them, the fuzzy membership function of the load value is:

[0053] Substitute C 1k = 100, C 1max = 500, C 1min = 50 into it to get .

[0054] Assume that the fuzzy membership values after calculation and mapping of other factors are respectively: F2(C 2k , B k ) = 0.5 (related to the business demand prediction value), F3(C 3k , B k ) = 0.4 (related to the business type priority, corresponding to the priority weight mapping of the ordinary web browsing business), F4(C 4k , B k ) = 0.3 (related to the business request frequency, assumed to be calculated based on the recent number of business requests).

[0055] Then .

[0056] The preset sleep threshold T s = 0.4, and the wake-up threshold T w = 0.6. Since D k = 0.4785 > T s and D k < T w , this component currently does not meet the sleep or wake-up conditions.

[0057] According to the output of the intelligent decision-making algorithm, send a sleep instruction to the hardware components that meet the sleep conditions, along with an estimated sleep duration. In remote areas, due to the large uncertainty of business requirements, the estimated sleep duration can be set to a relatively short interval, such as 1-2 minutes. Assume that for this radio frequency unit, based on the current business demand prediction and the characteristics of the hardware components, the estimated sleep duration is calculated to be 1.5 minutes through a linear regression model (the calculation process is similar to that of Embodiment 1, and relevant parameter values are assumed here). The sleep instruction should include the sleep mode and wake-up conditions of the hardware components to ensure that the hardware components can maintain a low-power state during sleep and quickly resume the working state when needed. Set the wake-up priority flag according to the urgency of the business requirements. For example, if there is a voice call service request, the wake-up priority flag of the relevant hardware components is set to high. When the business demand resumes, wake up the hardware components in sequence according to the order of the wake-up priority flags to ensure the continuity of critical services.

[0058] After implementing the energy-saving control method, regularly evaluate the energy-saving effect and service continuity of the base station. Assume that before implementing the energy-saving control method, the daily energy consumption of the base station in this remote area is 50 degrees of electricity, and the service interruption rate is 5%. After implementation, through one week of monitoring, the average daily energy consumption is reduced to 40 degrees of electricity, and the service interruption rate is reduced to 3%. By comparing the energy consumption data and service interruption rate indicators before and after implementation, evaluate the effectiveness of the energy-saving control method. According to the evaluation results, optimize and adjust the energy-saving control method. For example, adjust the parameters of the business demand prediction model, optimize the decision rules of the intelligent decision-making algorithm, and adjust the sleep and wake-up thresholds of the hardware components. Keep in communication with the users in the remote area and collect their feedback on the base station performance and energy-saving effect to further improve and optimize the energy-saving control method.

[0059] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present invention. Any reference signs in the claims should not be regarded as limiting the claims involved.

Claims

1. A 5G base station energy-saving control method, characterized in that, Including: S1, real-time monitor the communication data between the 5G base station and the user equipment to obtain the load condition of the current base station, and preprocess the collected data; S2, based on the priority of network services, real-time service requests, and historical service patterns, predict the service demands that may occur currently and in the short term; S3, according to the results of step S1 and step S2, construct an intelligent decision-making algorithm based on fuzzy logic and multi-objective optimization theory, and use this intelligent decision-making algorithm to determine whether various hardware components in the 5G base station meet the sleep condition or the wake-up condition; S4, if it is determined that the component meets the sleep condition, send a sleep instruction to the component to make it enter the sleep state; if it is determined that the component meets the wake-up condition, send a wake-up instruction to the component to make it resume the normal working state.

2. The energy-saving control method for a 5G base station according to claim 1, wherein In step S1, a denoising algorithm based on wavelet transform is used to denoise the collected data, and then the denoised data is decomposed at multiple scales by setting the wavelet basis function and the decomposition level.

3. The energy-saving control method for a 5G base station according to claim 1, wherein, The specific operations of step S2 include: S2.1, set the priority weight values of different types of services from high to low according to the degree of dependence of the service on network resources, real-time requirements, and factors affecting the user experience: emergency call services with extremely high real-time requirements and significant impact on the user experience, high-definition video streaming services, ordinary web browsing services; S2.2, capture the request initiation time, request quantity, and requested service type of various services in real time, and record the number of requests for each type of service per unit time; S2.3, collect and store the historical data of the base station services, construct a service demand prediction model based on the historical data, and obtain the predicted value of the historical service pattern; S2.4, assign the weighted coefficients of the network service priority weight, the real-time service request quantity, and the predicted value of the historical service pattern through weighted summation respectively to obtain the comprehensive predicted value of the service demands that may occur currently and in the short term.

4. The method for energy-saving control of a 5G base station according to claim 3, wherein, In step S2.1, if there is a virtual reality / augmented reality service, set its priority weight value between the priority weight values of the emergency call service with extremely high real-time requirements and significant impact on the user experience and the high-definition video streaming service.

5. The 5G base station energy-saving control method according to claim 3, wherein, In step S2.2, analyze the distribution of the source IP addresses of the service requests, cluster the source IP addresses of the service requests to divide the users into different regional groups, and count the service request frequency and service type preference of each regional group. According to the statistical results, define the service demand prediction value in segments.

6. The method for energy-saving control of a 5G base station according to claim 3, wherein, In step S2.3, the constructed service demand prediction model is implemented using a time series decomposition and adaptive weighted prediction algorithm or a grey prediction model.

7. The energy-saving control method for a 5G base station according to claim 1, characterized in that The expression of the intelligent decision-making algorithm is: Among them, represents the decision value of the k th hardware component, which is used to determine whether the component meets the sleep or wake-up conditions, L represents the number of factors participating in the decision, is the weight coefficient of each factor, represents the l th factor's fuzzy membership function with respect to the k th hardware component.

8. The energy-saving control method for a 5G base station according to claim 1, characterized in that In step S4, when sending a sleep instruction to the hardware component, attach an estimated sleep duration value to the sleep instruction, and this estimated value is calculated through a linear regression model according to the current service demand prediction situation and the characteristics of the hardware component.

9. The energy-saving control method for a 5G base station according to claim 1, characterized in that, In step S4, when a wake-up instruction is sent to a hardware component, the wake-up instruction is attached with a wake-up priority identifier, which is determined according to the urgency of the service requirement and the importance of the hardware component in meeting the service requirement. By setting the wake-up priority identifier, the components are woken up in sequence according to the priority order.

10. A 5G base station energy-saving control system, characterized in that, It includes: A real-time load monitoring module, which is used to monitor the communication data between the 5G base station and the user equipment in real time to obtain the load condition of the current base station and preprocess the collected data; A service requirement analysis module, which is used to predict the current and short-term possible service requirements based on the priority of network services, real-time service requests and historical service rules; An intelligent decision-making algorithm module, which is used to construct an intelligent decision-making algorithm based on the results of the real-time load monitoring module and the service requirement analysis module based on the fuzzy logic and multi-objective optimization theory, and use the intelligent decision-making algorithm to judge whether various hardware components in the 5G base station meet the sleep condition or the wake-up condition; A hardware component control module. If it is determined that a component meets the sleep condition, a sleep instruction is sent to the component to make it enter the sleep state; if it is determined that a component meets the wake-up condition, a wake-up instruction is sent to the component to make it resume the normal working state.

Images