A test regulation system and method for 5G terminals

Abstract

The present application relates to the technical field of test regulation for 5G terminal, and discloses a test regulation system and method for 5G terminal. After the service plane data of UE-base station connection is established, the test parameters and the loop variable are initialized, the throughput rate is collected point by point according to the current sampling interval, and the network state is accurately captured. The mean and standard deviation of the window sample are calculated, the fractional derivative order is adaptively determined according to the mean and standard deviation, the fractional derivative sequence is calculated by using the fractional binomial coefficient, the standard deviation of the sequence is mapped to the local fluctuation, which is used as the closed smooth reconstruction weight, and the smooth throughput rate is quickly output. Based on the smooth result and the fluctuation, the next window sampling interval and starting time are adaptively adjusted: when the network is stable, the interval is lengthened to save resources, and when the fluctuation is severe, the interval is shortened to improve the sampling density; finally, the test is automatically terminated according to the end condition, and the smooth throughput rate and the adaptive sampling interval sequence are output.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of test regulation for 5G terminals, in particular to a test regulation system and method for 5G terminals. BACKGROUND

[0002] The high-speed transmission capability of 5G network data becomes the basis for realizing various innovative applications, including high-definition video streaming, virtual reality (VR), augmented reality (AR), cloud gaming, etc. In the process of 5G network construction and optimization, throughput testing helps operators, device manufacturers and testing agencies evaluate the actual performance of the network to ensure that the network can provide stable and efficient services. The 5G terminal service plane data throughput test method mostly uses fixed time interval sampling and processes the sampling results based on simple moving average or exponential smoothing algorithm. When executing this kind of scheme, the sampling interval and window length need to be set in advance, and the smoothing factor is usually fixed, which cannot be automatically adjusted according to the fluctuations of network state.

[0003] Traditional test systems usually uniformly sample the terminal throughput rate at a fixed, empirically set time interval (e.g. 1 second, 500 milliseconds), ignoring the non-stationary characteristics of throughput rate at peak and trough periods. Fixed interval results in insufficient resolution when fluctuations are intense, while wasting test resources when relatively stable. The processing of sampling data mostly uses N-point moving average or exponential weighted average. Moving average cannot distinguish the intensity of fluctuations in different stages, and exponential smoothing needs to set the smoothing factor in advance, which is difficult to adapt to network sudden changes, and is prone to information lag or excessive suppression of important fluctuations. Some test schemes calculate the difference between adjacent sampling points to estimate the throughput rate change rate, but are limited to first-order or second-order difference, ignoring the long memory characteristics of network fluctuations. Integer-order difference can only reflect the instantaneous change of the recent limited sampling points, and it is difficult to capture the sustained and cumulative small fluctuations. The existing technology often relies on domain experience or repeated trials when setting the sliding window length, smoothing factor, threshold determination and other parameters, and lacks a mechanism for automatically generating parameters based on real-time statistical characteristics. This makes the test system need to be adjusted frequently under different network environments, reducing the deployment efficiency. Fixed sampling and fixed processing logic can easily lead to excessive sampling in the network steady state, wasting test bandwidth and terminal computing resources; while in the network sudden jitter stage, due to insufficient sampling and processing rate, it is difficult to capture key fluctuations in time.

[0004] To this end, the application aims to provide a test regulation system and method for 5G terminals, which first automatically generates window-level parameters in the initialization stage; then dynamically collects the throughput rate in each test window according to the current sampling rhythm; then determines the fractional derivative depth through real-time statistical analysis, and thereby captures the historical memory characteristics and transient fluctuations; then uses a closed smoothing model to weight and reconstruct the original data; finally, the sampling rhythm of the next window is adaptively adjusted according to the smoothing results and local fluctuation feedback until the end condition is met. SUMMARY

[0005] The application provides a test regulation system and method for 5G terminals, which solves the problems mentioned in the background art.

[0006] The application provides the following technical solutions: a test regulation method for 5G terminals, comprising:

[0007] S1, adaptively adjusting the sampling interval and starting time of the next window according to the smoothing reconstruction results and local fluctuation amount;

[0008] S2, sequentially calculating each sampling time point based on the sampling interval of the current window, and synchronously obtaining the throughput rate at the corresponding time;

[0009] S3, calculating the sample mean and sample standard deviation of the sampled samples, and adaptively determining the fractional derivative order according to the coefficient of variation;

[0010] S4, performing derivative operation on the sampled data using the fractional binomial coefficient according to the adaptively determined fractional order, to obtain a fractional derivative sequence;

[0011] S5, performing statistical analysis on the fractional derivative sequence, calculating the sample mean and sample standard deviation thereof, and calculating the local throughput rate fluctuation amount according to the standard deviation;

[0012] S6, constructing a smoothing coefficient according to a predetermined closed smoothing algorithm, weighting and reconstructing the sampled throughput rate sequence, and outputting the smoothed throughput rate sequence;

[0013] S7, updating the sampling interval based on the smoothing reconstruction results and local fluctuation amount, and setting the starting time and sampling interval of the next test window;

[0014] S8, judging whether the preset end condition is met, and if so, outputting the smoothed throughput rate sequence and the adaptive sampling interval sequence and terminating the test.

[0015] Optionally, the adaptive adjustment of the sampling interval and starting time of the next window according to the smoothing reconstruction results and local fluctuation amount specifically comprises:

[0016] setting the total test duration as , the first window sampling interval is ;

[0017] Record the test start time , the test end time ;

[0018] Set the throughput rate collection function to , ; wherein, is a set of real numbers greater than 0;

[0019] Calculate the number of sampling points per window ;

[0020] Initialize the loop variable:

[0021] , , ; wherein, is the window number; is the start time of the first window; is the sampling interval of the first window.

[0022] Optionally, the sampling interval based on the current window is used to calculate each sampling time point in turn, and the throughput rate at the corresponding time is synchronously obtained, specifically including:

[0023] When is true, perform the current window level collection; otherwise, go to step S8 to end;

[0024] The following steps are performed on

[0025] S201, calculate the first sampling time: ;

[0026] S202, synchronously collect the throughput rate: ; wherein, is the sampling point index within the window; is the first sampling time of the first window; the throughput rate obtained by the first sampling of the first window.

[0027] Optionally, the sampled samples are calculated for sample mean and sample standard deviation, and the fractional derivative order is adaptively determined according to the coefficient of variation, specifically including:

[0028] Calculate the original mean and standard deviation of the window: ​

[0029] ; ;

[0030] wherein, is the sample average throughput of the first window; is the sample standard deviation of the first window;

[0031] calculate the coefficient of variation of the current window and the fractional order:

[0032] , , ; wherein, is the coefficient of variation of the first window; is the fractional order derivative order of the first window.

[0033] Optionally, the fractional order determined according to the self-adaption is used to perform derivative operation on the sampling data by using the fractional order binomial coefficient to obtain a fractional order derivative sequence, and specifically includes:

[0034] for each calculate: ;

[0035] wherein, is the fractional order binomial coefficient; is the fractional order derivative of the first sample of the first window; is the summation index.

[0036] Optionally, the statistical analysis is performed on the fractional order derivative sequence to calculate the sample average and the sample standard deviation, and the local throughput fluctuation is calculated according to the standard deviation, and specifically includes:

[0037] calculate the average of the fractional order derivative of the first window ;

[0038] calculate the standard deviation of the fractional order derivative of the first window ;

[0039] set the local throughput fluctuation of the first window as .

[0040] Optionally, the predetermined closed smoothing algorithm is used to construct a smoothing coefficient to perform weighted reconstruction on the sampling throughput sequence to output a smoothed throughput sequence, and specifically includes:

[0041] construct the first Closed smooth coefficient of window ;

[0042] to the window , smooth configuration of the first window of the first Smoothed throughput rate after sampling :

[0043] .

[0044] Optionally, the sampling interval is updated based on the smooth reconstruction result and the local fluctuation amount, and the starting time of the next test window and the sampling interval are set, specifically including:

[0045] If , the global reference average throughput rate ;

[0046] Let the sampling interval of the first window be ;

[0047] Let the sampling time of the first window be , , ;

[0048] Return to step S2.

[0049] Optionally, the judgment whether the preset ending condition is met, if met, the smoothed throughput rate sequence and the adaptive sampling interval sequence are outputted and the test is terminated, specifically including:

[0050] When the sampling entry condition in step S2 is no longer met , that is, when , the sampling loop is terminated, and the total number of effective processing windows ;

[0051] The smoothed throughput rate sequence is outputted , wherein the index is mapped , ; the adaptive sampling interval sequence is ; wherein, is the first smoothed throughput rate sample; is the global sample number of the linear sequence of smoothed throughput rates; is the sampling interval actually used by the first window; is the window index.

[0052] A system for implementing the test regulation method for 5G terminals, comprising:

[0053] Initialization module: used for building test environment and defining variables;

[0054] Sampling module: used for collecting 5G terminal throughput rate data according to the current sampling interval;

[0055] Statistical and adaptive module: used for calculating original statistics in the window and adaptively determining the order of fractional derivative;

[0056] Fractional derivative calculation module: used for accurately calculating the fractional derivative sequence based on the determined order;

[0057] Fluctuation amount module: used for calculating the fractional derivative data and determining the local throughput fluctuation amount;

[0058] Smooth reconstruction module: used for reconstructing the throughput rate based on the closed smooth model;

[0059] Interval updating module: used for adaptively updating the sampling interval and setting the next sampling window according to the reconstruction result and the fluctuation amount;

[0060] Termination and output module: used for judging whether the test is ended and outputting the smooth throughput rate sequence and the sampling interval sequence.

[0061] The present application has the following beneficial effects:

[0062] 1. In the test initialization stage, the scheme sets the total test time and the first window sampling interval, builds the measurement environment and defines all loop control variables. The time boundary and the sampling function are taken as part of the overall environment configuration, so that the test process has a clear time planning and calling framework from the beginning. This design avoids the common problems of scattered distribution of environment parameters and difficulty in reproduction in traditional tests, and provides a reliable foundation for subsequent adaptive regulation. This stage proposes a systematic management strategy in terms of resource allocation and environment consistency, solving the problem of difficult comparison and analysis of test results caused by inconsistent test environment configuration.

[0063] 2. In the throughput rate data collection stage, the scheme calculates each sampling time and synchronously obtains the corresponding throughput rate value according to the sampling interval of the current window. The sampling time is automatically mapped to the test timeline, and the precise synchronization is realized through the built-in clock and the collection function without manual intervention. This design is beneficial to improve the timing accuracy of data acquisition, and avoids the measurement deviation caused by inaccurate trigger mechanism in traditional methods. This stage does not depend on fixed trigger or external signal, and can still maintain high timing consistency when the network state changes dramatically, solving the problems of resource waste and measurement blind area caused by invalid or repeated sampling.

[0064] 3、In the statistical analysis and adaptive order determination stage, the scheme calculates the average level and fluctuation amplitude of the throughput rate samples collected in the current window, and adaptively selects the fractional order differentiation order based on the relative fluctuation index. The coefficient of variation is used as the adaptive adjustment basis, so that the fractional order can be dynamically adjusted according to the network fluctuation characteristics. This design is beneficial to reduce the model complexity when the network is stable, and to improve the sensitivity when the network fluctuates violently, and to balance the calculation overhead and response accuracy. Traditional methods often use fixed differentiation order or empirical parameters, and this stage realizes the intelligent and unsupervised update of parameters through real-time fluctuation evaluation, solving the problem of strong subjectivity in parameter selection and difficulty in adapting to changing network environment.

[0065] 4、In the fractional differential sequence calculation stage, the scheme performs fractional differential operation on the sampled data according to the adaptively determined order, to extract the historical long memory characteristics and subtle dynamic changes of the throughput rate sequence. The fractional binomial weight is applied to the data sequence to realize the comprehensive capture of multi-time domain information without additional historical window management. This design is beneficial to enhance the dual perception ability of network sudden events and slow trend changes, and to improve the response sensitivity of the test to transient disturbances. Traditional integer order differentiation only focuses on short-term changes, and this stage considers both short-term fluctuations and long-term trends through fractional order operation, solving the problem of measurement distortion caused by ignoring network memory effect.

[0066] 5、In the local throughput fluctuation evaluation stage, the scheme again calculates the average level and dispersion degree of the fractional differential sequence, and maps the obtained results to a local fluctuation index with the same dimension as the throughput rate. The secondary statistical analysis structure is introduced to convert the fractional differential results into an intuitive index that can be used for smoothing and sampling adjustment. This design is beneficial to present the complex differential fluctuation information in a quantitative form, which is convenient for subsequent algorithm modules to directly call. Traditional methods often use empirical threshold or single fluctuation measure, and this stage enhances the objectivity and robustness of fluctuation evaluation through two-level statistics, solving the problem of previous indicators being easily disturbed by outliers and poor stability.

[0067] 6、In the closed smoothing reconstruction stage, the scheme constructs a closed smoothing coefficient based on the local fluctuation index, and performs weighted reconstruction on the original throughput rate sequence to obtain the smoothed data output. The closed-form analytical expression is used to directly calculate the optimal smoothing weight without iteration optimization or filter design. This design is beneficial to reduce the computational complexity and avoid filtering delay, making the smoothing process efficient and real-time. Traditional smoothing often relies on sliding window filtering or iterative least squares, and this stage completes the reconstruction at one time through the closed-form scheme, solving the problem of large time delay and heavy calculation burden of filtered data, while improving the timeliness and accuracy of smoothed data.

[0068] 7、In the adaptive sampling interval updating stage, the scheme dynamically adjusts the sampling interval and starting time of the next window based on the smooth reconstruction result and local fluctuation index. The smooth and fluctuation results are combined to make a dual decision, and the adjustment strategy combining global reference and local index is introduced. This design is beneficial to automatically prolong the sampling interval to save resources when the network is stable, and shorten the interval to capture key information in time when the network fluctuation increases. Traditional methods are fixed or single based on fluctuation judgment. This stage improves the adaptability and reliability of the regulation strategy through multi-element fusion decision, and solves the problem of delay or overreaction caused by single index decision.

[0069] 8、In the end condition judgment and result output stage, the scheme automatically judges the test termination according to the preset threshold of the total test time or the number of effective processing windows, and outputs the smooth throughput rate sequence and the corresponding sampling interval sequence. The test progress and processing quality are considered to make the termination decision meet the time constraint and ensure the integrity of the cumulative data. This design is beneficial to avoid insufficient data caused by premature termination, and avoid wasting resources caused by meaningless overtime operation. Unlike traditional methods that are often based on time or experience, this stage realizes intelligent termination by comprehensively considering progress and data effectiveness, and solves the problems of uneven data segments and invalid system burden caused by too short or too long test. BRIEF DESCRIPTION OF DRAWINGS

[0070] Figure 1 The flowchart of the present application is shown. DETAILED DESCRIPTION

[0071] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0072] Embodiment, refer to Figure 1 A test regulation method for a 5G terminal, comprising:

[0073] End-to-end test (base station and user) premise:

[0074] 1、Connection establishment: connect the test equipment with the 5G network.

[0075] 2、Data transmission: large-scale data transmission, test file upload and download speed.

[0076] 3. Start the test equipment and begin the throughput test. During the file upload and download process, the test tool will display the data transfer rate in real time. Pay attention to the test time during the test, which is usually a few minutes to several hours, and the network environment, that is, the 5G network signal coverage and signal quality may affect the throughput.

[0077] 4. Data monitoring: Use network monitoring tools to track throughput in real time and record transmission rates.

[0078] S1, according to the smooth reconstruction result and the local fluctuation, adaptively adjust the sampling interval and starting time of the next window;

[0079] S2, based on the sampling interval of the current window, calculate each sampling time point in turn, and synchronously obtain the throughput rate at the corresponding time;

[0080] S3, calculate the sample mean and sample standard deviation of the sampled sample, and adaptively determine the fractional derivative order according to the coefficient of variation;

[0081] S4, according to the adaptively determined fractional order, use the fractional binomial coefficient to perform derivative operation on the sampled data to obtain the fractional derivative sequence;

[0082] S5, statistically analyze the fractional derivative sequence, calculate its sample mean and sample standard deviation, and calculate the local throughput fluctuation according to the standard deviation;

[0083] S6, according to the predetermined closed smooth algorithm, construct the smooth coefficient, perform weighted reconstruction on the sampled throughput rate sequence, and output the smoothed throughput rate sequence;

[0084] S7, update the sampling interval based on the smooth reconstruction result and the local fluctuation, and set the starting time and sampling interval of the next test window;

[0085] S8, judge whether the preset ending condition is met, if yes, output the smoothed throughput rate sequence and the adaptive sampling interval sequence and terminate the test.

[0086] The test parameters are consistent by the unified initialization step of test environment and cycle variables, and the clear time boundary is provided for the subsequent window division. The high time sequence accuracy of network state is captured by collecting the throughput rate data in the set interval and recording synchronously. The average level and fluctuation amplitude of current window data are calculated in the statistical analysis stage, and the differential order is adaptively selected, which solves the problem of insufficient sensitivity or waste of computing resources caused by fixed parameters in traditional methods. In the fractional order differential operation stage, the method can consider the throughput rate information of multiple time periods in the past, pay attention to sudden fluctuations, and retain long-term trends, so that the test has higher sensitivity to network mutations. The fluctuation evaluation step converts the differential result into an intuitive and quantifiable index, providing an accurate input basis for smooth reconstruction. The closed smooth reconstruction stage uses an analytical expression for weighted processing, which responds to network jitter in real time and avoids time-consuming iterative filtering. Then, the adaptive sampling interval is updated to make full use of the smooth result and fluctuation index. When the network is stable, the sampling interval is moderately extended to save resources, and when the network fluctuates violently, the interval is shortened to improve the sampling density. Finally, the test process is automatically controlled by the termination judgment of the total test time or the number of windows, which avoids premature termination or meaningless continuation. Through the above continuous steps, the method solves the contradiction between precision and efficiency of fixed sampling and lagging smoothing, improves the dynamic adaptability and resource utilization of the test process, and ensures the integrity and reliability of the test results.

[0087] The adaptive adjustment of the sampling interval and the starting time of the next window according to the smooth reconstruction result and the local fluctuation is specifically:

[0088] The total test time is set as , which clearly defines the duration of the entire test process and provides a time boundary for subsequent window division. The first window sampling interval is , which determines the time resolution of the samples in the first window as the initial benchmark for adaptive adjustment.

[0089] The test starting time is recorded to mark the starting point of the test process, and all sampling times are referenced to this time. The test ending time is clearly defined to facilitate out-of-boundary judgment in the loop.

[0090] The throughput rate collection function is set as , ; wherein is a set of real numbers greater than 0; the encapsulation of the terminal throughput rate reading operation at any time facilitates subsequent consistent calls.

[0091] The number of sampling points per window Balance the sample size and computational complexity of each window to ensure sufficient data for subsequent analysis;

[0092] Initialize the loop variable:

[0093] , , ;in, Number the windows; For the first Window start time; For the first Window sampling interval; set the initial values ​​for window number, start time, and sampling interval to prepare for the first round of sampling.

[0094] By pre-setting the total test duration and the sampling interval for the first window, a strict time framework is provided for the entire test, ensuring that each window runs within this range and avoiding data collection exceeding limits or omissions. Recording the start and end times of the test provides an accurate reference timescale, ensuring a unified timing benchmark for all subsequent sampling points, thus resolving the data incomparability issue caused by inconsistent timing benchmarks in multiple runs. Configuring the throughput acquisition process as an encapsulated operation simplifies subsequent calls and ensures the uniformity of the acquisition function interface, reducing implementation complexity. By calculating the number of sampling points per window, the method achieves a balance between different test durations and sampling requirements, ensuring sufficient data to support statistical analysis within each window while controlling the computational workload within the window, avoiding resource burden caused by excessive samples. Finally, initializing loop variables such as window number, start time, and sampling interval ensures a smooth start to the first round of sampling and provides an accurate starting point for subsequent adaptive adjustments. This series of steps, through systematic pre-test configuration, solves the problems of scattered environmental parameters, susceptibility to human error, and low efficiency of repetitive configuration in traditional test schemes, thus improving test reproducibility and engineering deployment efficiency.

[0095] The process of calculating the throughput at each sampling time point based on the current window sampling interval and synchronously obtaining the throughput at the corresponding time point specifically includes:

[0096] when When established, execute this window. Level 1 collection; otherwise, proceed to step S8 to end; ensure all items in the current window are collected. Each sample can be collected within the test duration to avoid exceeding the limits;

[0097] right Perform the following steps:

[0098] S201, Calculate the... Next sampling time: ; obtained the first Window the accurate time point of the subsampling;

[0099] synchronously collecting the throughput rate; ; wherein, is the sampling point index within the window; is the first window the first subsampling time point; the first window the first subsampling to obtain the throughput rate; read and record the throughput rate at this time point for subsequent statistics and analysis.

[0100] By judging whether the current window is still within the test duration range, the method avoids the out-of-bound problem of continuing sampling after the test is completed, ensuring the consistency of data collection and overall time constraints. When confirming that the window can be collected, data acquisition is performed in turn according to the preset sampling rhythm, so that the sampling timing strictly follows the time planning and has flexibility, which can be flexibly changed according to subsequent adaptive adjustment. The synchronous collection of the throughput rate binds data reading and time scale, eliminating the data misplacement risk caused by timing deviation and asynchronous triggering. This design is beneficial to improve the sampling accuracy of network testing in high load or network switching scenarios, while reducing the system burden caused by uncoordinated calling and repeated measurement. Compared with the existing timing triggering or external event driven method, the method pays more attention to the close connection with the test duration and window structure, so that the sampling process is stable and controllable, and can quickly respond to the dynamic adjustment requirements of the upper control strategy, thereby improving the data integrity and system running efficiency.

[0101] The sample average and sample standard deviation are calculated for the sampled sample, and the fractional derivative order is adaptively determined according to the coefficient of variation, specifically including:

[0102] Calculate the original average and standard deviation of the window:

[0103] obtain the overall throughput level reference of the window;

[0104] quantify the fluctuation amplitude of the throughput rate within the window;

[0105] wherein, is the sample average throughput rate of the first window; is the sample standard deviation of the first window;

[0106] Calculate the coefficient of variation and the fractional derivative order of the window:

[0107] , , ; wherein, The coefficient of variation of the window provides a relative fluctuation index for adaptive order generation. The coefficient of variation of the window provides a relative fluctuation index for adaptive order generation. The fractional derivative order of the window automatically determines the fractional derivative order according to the fluctuation degree, and improves the sensitivity of the model to transient changes. The fractional derivative order of the window automatically determines the fractional derivative order according to the fluctuation degree, and improves the sensitivity of the model to transient changes.

[0108] By calculating the average level and dispersion degree of the sampling data, the method first obtains the baseline value and fluctuation quantization description of the current network throughput, providing a reliable basis for subsequent fractional order selection. The ratio of fluctuation to average level is used as an adaptive parameter, so that the selection of fractional order can automatically adjust with the smooth or severe fluctuations of network state, solving the problem of insufficient sensitivity or excessive calculation of traditional fixed order differential under different network conditions. In the network state, a lower fractional order can reduce the model complexity and operation cost; in the network state, a higher fractional order can enhance the sensitivity of differential operation to transient changes. This adaptive mechanism beneficially improves the intelligent level and resource utilization efficiency of the overall test process through real-time feedback of data fluctuation information, avoids the performance instability and test result deviation caused by subjective experience parameter setting, and brings higher precision and reliability to the throughput capture in complex network environment.

[0109] The fractional order determined according to the adaptive determination is used to perform derivative operation on the sampling data by using the fractional order binomial coefficient, and a fractional derivative sequence is obtained, which specifically includes:

[0110] For each Calculation: ;

[0111] Wherein, is the fractional order binomial coefficient; is the fractional derivative of the first window sample; is the summation index;

[0112] Based on definition calculation fractional derivative, capture the long memory characteristics and subtle fluctuations of throughput.

[0113] By using the fractional binomial coefficient to weight and accumulate each piece of sampling data in the current window, the method has dual attention to recent changes and historical trends, which can capture short-term information caused by sudden fluctuations and reflect the cumulative trend of long-term throughput rate changes. Unlike ordinary integer-order differentiation, which only focuses on instantaneous changes, this method converges the contributions of multiple time periods through fractional weighting, enabling the operation result to show the "long memory" effect of network state, thereby improving the prediction ability of slow trend changes and the response speed to sudden events. This step integrates multi-period information into a single-point operation, achieving a more comprehensive perception of network dynamics, which is beneficial to providing more stable and reliable data support in scenarios with severe network fluctuations or sudden failures, while not relying on additional historical data window management, simplifying system implementation and improving operational efficiency.

[0114] The statistical analysis of the fractional derivative sequence is performed to calculate the sample mean and sample standard deviation, and the local throughput fluctuation is calculated according to the standard deviation, specifically including:

[0115] The fractional derivative mean of the first window is calculated ; the concentration trend of the fractional derivative in the current window is measured;

[0116] The fractional derivative standard deviation of the first window is calculated ; the dispersion degree of the fractional derivative is measured, reflecting the fluctuation intensity;

[0117] The local throughput fluctuation of the first window is set to ; the fractional fluctuation is converted back to the same dimension as the throughput rate, which is convenient for subsequent smoothing.

[0118] After obtaining the fractional derivative sequence, the concentration trend and dispersion degree of the sequence are calculated, and the method forms a second quantitative processing of the advanced differential information, which converts the complex fractional fluctuation characteristics into a directly usable local fluctuation index. This index retains the network dynamic details revealed by fractional operation, and is converted to the same dimension as the throughput rate, which is convenient for subsequent smoothing and sampling adjustment module to directly call. Through this step, the method solves the problem of being difficult to intuitively evaluate the fluctuation degree by using differential results only, so that the overall algorithm can automatically extract and quantify the network fluctuation characteristics without manual threshold setting, providing a clearer decision basis for smoothing reconstruction and balanced resource allocation, thereby improving the robustness and automation level of the method.

[0119] The smoothing coefficient is constructed according to the predetermined closed smoothing algorithm, the sampling throughput rate sequence is weighted and reconstructed, and the smoothed throughput rate sequence is output, specifically including:

[0120] The first Closed form smoothing coefficient ; the weight ratio of the original value and the average value when determining the smoothing, closed form adaptive;

[0121] For the current window , the smoothing is constructed The smoothed throughput rate of the first sampling of the window :

[0122] Output the smoothed throughput rate sequence, reduce the influence of extreme jitter.

[0123] By constructing the smoothing coefficient of the analytical closed form, the method can determine the optimal weighting ratio in one operation, without the iterative calculation or the design of complex filter required by the traditional filtering algorithm. This design effectively reduces the calculation delay, so that the smoothing process can follow the real-time data input, thereby reducing the signal lag introduced by filtering. The weighting reconstruction step balances the original data and the local average value, which can suppress burst jitter and noise, and also retain the true throughput rate trend. This mechanism solves the problem of distortion or delay when processing burst fluctuations in conventional moving average or iterative least squares smoothing, and improves the timeliness and accuracy of obtaining stable and high-quality throughput rate curve, providing more reliable basic data for subsequent adaptive sampling.

[0124] The sampling interval is updated based on the smoothing reconstruction result and the local fluctuation amount, and the start time and sampling interval of the next test window are set, specifically including:

[0125] If , let the global reference average throughput rate ; the average of the first window is used as the standard for adjusting the sampling interval of all subsequent windows;

[0126] Let the sampling interval of the first window be ; according to the comparison between the current window and the global reference, the sampling frequency is automatically accelerated or slowed down;

[0127] Let the sampling time of the first window be , , ; calculate the start time of the next window and increment the window number, and enter the next round of collection;

[0128] Return to step S2.

[0129] By combining the smoothed data curve with the fluctuation index, the method automatically extends the sampling interval in the network stable stage to reduce invalid data collection and system resource consumption; when the network fluctuation is enhanced, it automatically shortens the interval and increases the sampling frequency to capture key information. This dual decision strategy takes into account test efficiency and information integrity, solving the problem of response lag or resource waste caused by traditional fixed sampling or single fluctuation judgment. At the same time, by cyclically increasing the window number and calculating the starting time of the next window, the method ensures the continuity and timing accuracy of the test process, which is beneficial to realize long-time and high-reliable network performance monitoring, and improves the intelligent scheduling ability and operation stability of the test system.

[0130] The method comprises the steps of:

[0131] When the sampling entry condition in step S2 is no longer met , that is, when , the sampling loop is terminated, the total number of valid processing windows is set to ; determine how many complete windows have been processed in actuality;

[0132] Output the smoothed throughput rate sequence , wherein the index mapping , ; the adaptive sampling interval sequence ; wherein, is the th smoothed throughput rate sample; is the global sample number of the linear sequence of smoothed throughput rates; is the sampling interval actually used by the th window; is the window index; provide the smoothed throughput rate and sampling interval data sequences required for subsequent analysis and visualization.

[0133] By considering both the total test duration and the number of valid processing windows, the method ensures that the test can cover a long enough monitoring time and process enough complete window data, thereby avoiding data insufficiency caused by premature termination or resource waste caused by excessive continuation. After meeting the termination condition, the method records the smoothed throughput rate sequence and the corresponding sampling interval, which provides a continuous and high-quality data stream for subsequent performance evaluation and visualization analysis. This step solves the problem of single time or single window number judgment that cannot balance test integrity and efficiency, making the test termination meet the engineering requirements and have flexibility, which beneficially improves the controllability of the test process and the usability of the results.

[0134] The embodiment also provides a system for testing and regulating method of 5G terminal, comprising:

[0135] An initialization module is configured to build a test environment and define variables;

[0136] A sampling module is configured to collect data of 5G terminal throughput according to a current sampling interval;

[0137] A statistical and adaptive module is configured to calculate original statistical quantities in a window and adaptively determine a fractional derivative order;

[0138] A fractional derivative calculation module is configured to accurately calculate a fractional derivative sequence based on the determined order;

[0139] A fluctuation quantity module is configured to calculate fractional derivative data and determine a local throughput fluctuation quantity;

[0140] A smoothing reconstruction module is configured to reconstruct the throughput based on a closed smoothing model;

[0141] An interval updating module is configured to adaptively update the sampling interval and set a next sampling window according to the reconstruction result and the fluctuation quantity;

[0142] A termination and output module is configured to judge whether the test is ended and output a smoothed throughput sequence and a sampling interval sequence.

[0143] By decomposing each function of the method into a modular system unit, the method improves the scalability and maintainability of system design. The environment initialization module uniformly configures test parameters and variables, laying a consistent running foundation for the system; the sampling module is responsible for high timing accuracy data acquisition, ensuring the real-time nature of information input; the statistical and adaptive module automatically processes window data and generates a differential order, avoiding errors in manual intervention and experience parameter settings; the fractional derivative calculation and fluctuation quantity evaluation module works collaboratively to convert data into advanced indicators for measuring network dynamics; the smoothing reconstruction and interval updating module continuously optimizes data quality and sampling strategies based on real-time feedback; and the termination and output module realizes closed-loop control of the entire test process and provides results. This modular design solves the problem of high coupling degree of traditional systems, which is difficult to upgrade or reuse flexibly, and provides convenience for rapid integration and customization of different application scenarios and hardware platforms, which is beneficial to efficient and reliable network testing and monitoring in industrial deployment.

[0144] It is to be noted that, as used in this document, the term "indicia" is intended to encompass any type of data, information, or other content, whether in the form of text, graphics, images, video, audio, or otherwise. It is to be further noted that, as used in this document, the terms "coupled" and "connected," along with derivatives thereof, can be used to mean one or more of the following: in electrical communication with; physically touching; in working communication with; and / or information can be shared between any two components. It is to be further noted that, as used in this document, the terms "include" and "comprise," along with derivatives thereof, can be used to indicate inclusion of one or more elements or steps; these terms are to be read expansively and do not preclude the addition of one or more additional elements or steps. Finally, the terms "first," "second," and the like can be used to describe a variety of elements, but do not imply a particular order or ranking of such elements.

[0145] The preferred embodiments herein disclosed merely by way of example can be varied in or der to accommodate specific situations by those skilled in the art. Thus, it is intended that the application cover modifications and variations of this application provided they come within the scope of the appended claims and their equivalents.

Claims

1. A testing and control method for 5G terminals, characterized in that, include: S1. Based on the smooth reconstruction results and local fluctuations, adaptively adjust the sampling interval and start time of the next window; The adaptive adjustment of the sampling interval and start time of the next window based on the smooth reconstruction results and local fluctuations specifically includes: Set the total test duration to The first window sampling interval is ; Record the start time of the test Test end time ; Set the throughput acquisition function to , ;in, It is the set of real numbers greater than 0; Calculate the number of sampling points per window ; Initialize the loop variable: , , ;in, Number the windows; For the first Window start time; For the first Window sampling interval; S2. Based on the sampling interval of the current window, calculate each sampling time point in sequence, and synchronously obtain the throughput at the corresponding time. The process of calculating the throughput at each sampling time point based on the current window sampling interval and synchronously obtaining the throughput at the corresponding time point specifically includes: when This window is executed upon creation. Level 1 acquisition; otherwise, proceed to step S8 to end; right Perform the following steps: S201, Calculate the... Next sampling time: ; S202, Synchronous Acquisition Throughput: ;in, Index of sampling points within the window; For the first Window The next sampling time; No. Window Throughput obtained from the second sampling; S3. Calculate the sample mean and sample standard deviation for the sampled samples, and adaptively determine the order of the fractional derivative based on the coefficient of variation; The calculation of the sample mean and sample standard deviation of the sampled samples, and the adaptive determination of the order of the fractional derivative based on the coefficient of variation, specifically includes: Calculate the raw mean and standard deviation for this window: ； ； in, For the first Average throughput of the window's samples; For the first The sample standard deviation of the window; Calculate the coefficient of variation and fractional order of this window: , , ;in, For the first The coefficient of variation of the window; For the first The order of the fractional derivative of the window; S4. Based on the adaptively determined fractional order, perform derivative operations on the sampled data using the fractional binomial coefficients to obtain a fractional derivative sequence; The step of performing derivative operations on the sampled data using fractional binomial coefficients according to the adaptively determined fractional order to obtain a fractional derivative sequence specifically includes: For each calculate: ; in, , where is the coefficient of the fractional binomial; For the first Window Fractional derivative of each sample; For summation index; S5. Perform statistical analysis on the fractional derivative sequence, calculate its sample mean and sample standard deviation, and calculate the local throughput fluctuation based on the standard deviation. The statistical analysis of the fractional derivative sequence, calculating its sample mean and sample standard deviation, and calculating the local throughput fluctuation based on the standard deviation, specifically includes: Calculate the first Window fractional derivative mean ; Calculate the first Window fractional derivative standard deviation ; Set the first The local throughput fluctuation of the window is ; S6. Construct smoothing coefficients based on a predetermined closed-loop smoothing algorithm, perform weighted reconstruction on the sampled throughput sequence, and output the smoothed throughput sequence. The step of constructing smoothing coefficients based on a predetermined closed-loop smoothing algorithm, weighting and reconstructing the sampled throughput sequence, and outputting a smoothed throughput sequence specifically includes: Construct the first Window closing smoothness coefficient ; For this window Smooth construction of the first Window Smoothed throughput after subsampling : ； S7. Update the sampling interval based on the smooth reconstruction results and local fluctuations, and set the start time and sampling interval of the next test window; S8. Determine whether the preset termination condition is met. If it is met, output the smoothed throughput sequence and the adaptive sampling interval sequence and terminate the test.

2. The testing and control method for a 5G terminal according to claim 1, characterized in that, The step of updating the sampling interval based on the smooth reconstruction results and local fluctuations, and setting the start time and sampling interval of the next test window, specifically includes: like This makes the global baseline average throughput ; Let the first The sampling interval of the window is ; Let the first Window The sampling time is , ; Return to step S2.

3. The testing and control method for a 5G terminal according to claim 2, characterized in that, The determination of whether a preset termination condition is met, and if so, outputting the smoothed throughput sequence and the adaptive sampling interval sequence and terminating the test, specifically includes: When the sampling entry condition described in sampling step S2 is no longer met That is, when At that time, terminate the sampling cycle and set the total number of valid processing windows. ; Output smooth throughput sequence , where index mapping , Adaptive sampling interval sequence ;in, For the first One smoothed throughput sample; The global sample number for a smoothed linear sequence of throughput; For the first The actual sampling interval used by each window; For window indexing.

4. A system employing the testing and control method for 5G terminals as described in claim 3, characterized in that, include: Initialization module: Used to set up the test environment and define variables; Sampling module: Used to collect data on the throughput of 5G terminals according to the current sampling interval; The Statistics and Adaptive module is used to calculate the raw statistics within the window and adaptively determine the order of the fractional derivative. Fractional derivative calculation module: used to accurately calculate the fractional derivative sequence based on a given order; Volatility module: Used to statistically analyze fractional derivative data and determine local throughput volatility; Smoothing Reconstruction Module: Used to reconstruct throughput based on a closed smoothing model; Interval update module: Used to adaptively update the sampling interval and set the next sampling window based on the reconstruction results and fluctuations; Termination and Output Module: Used to determine whether the test has ended and output the smoothed throughput sequence and sampling interval sequence.

Images