A network optimization method and device based on speed measurement curve features

Abstract

The application discloses a network optimization method and device based on speed measurement curve characteristics, the method comprising: acquiring time sequence throughput curve data under a current running scene, determining network throughput fluctuation value, network speed drop ratio and network throughput change after eliminating the initial unstable data in the first quarter; and determining network stability factor according to the above parameters to estimate network transmission stability; meanwhile, determine the environmental quality factor according to the co-frequency interference intensity and the queue congestion intensity to estimate the network smoothness; multiply the integrity factor, the network stability factor and the environmental quality factor to obtain the network experience confidence; when the confidence is less than the preset threshold, determine the abnormal network performance index, and generate the matching network state adjustment strategy to adjust the network state of the current running scene. In this way, the network use experience can be accurately predicted, so that intelligent diagnosis and adaptive optimization of the network use experience are realized.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a network optimization method and device based on velocity curve characteristics. Background Technology

[0002] With the rapid development of mobile internet and fiber broadband access technologies, high-bandwidth, low-latency services such as high-definition video streaming, cloud gaming, and online conferencing have become important components of users' daily network activities. To measure network service quality, network experience evaluation methods typically rely on network speed testing, which measures the downlink or uplink throughput of terminal devices within a specific time period, using peak or average rates as the core indicators for evaluating network quality. In these network speed testing systems, the results usually output directly collected basic network performance indicators or the maximum / average value of a single / multiple sampling, and the results are assumed to be reliable. However, in real-world home networks, WiFi signal interference, channel contention, background applications consuming excessive bandwidth, sudden bandwidth allocation by operators, and the slow start mechanism of the TCP protocol can all cause drastic fluctuations in instantaneous throughput. This leads to the direct inclusion of process noise in the final estimated result during speed testing, making the estimated results unreliable and hindering the identification of network optimization methods.

[0003] In real-world applications, users often face the contradictory phenomenon of "speed test results meeting the standards, but actual service experience being sluggish," such as in home Wi-Fi... In a Fi environment, although speed test software can show instantaneous bandwidth of hundreds of megabits, due to wireless signal interference or network congestion, frequent buffering or image quality degradation still occurs when watching videos. However, the unreliability of speed test predictions makes it difficult to find network optimization methods that are suitable for the current application scenario, resulting in a poor user network experience. Summary of the Invention

[0004] This application provides a network optimization method and device based on speed curve characteristics to solve the following technical problem: how to accurately predict network user experience and achieve intelligent diagnosis and adaptive optimization of network user experience.

[0005] In a first aspect, embodiments of this application provide a network optimization method based on velocity curve features, the method comprising: Obtain the timing throughput curve data under the current running scenario, and remove the first quarter of the initial unstable data from the timing throughput curve data to obtain the timing throughput curve data after removal. Based on the removed time-series throughput curve data, determine the network throughput fluctuation value, network speed drop ratio, and network throughput change in the current operating scenario. Based on the network throughput fluctuation value, the network speed drop ratio, and the network throughput change, a network stability factor for the current operating scenario is determined to predict the network transmission stability of the current operating scenario. Based on the intensity of co-channel interference and queue congestion in the current operating scenario, the environmental quality factor of the current operating scenario is determined to predict the network smoothness of the current operating scenario. The network experience confidence score of the current operating scenario is obtained by multiplicatively fusing the integrity factor, the network stability factor, and the environmental quality factor corresponding to the current operating scenario. The integrity factor is used to characterize the network transmission integrity of the current operating scenario. In response to the network experience confidence level being less than a preset confidence threshold, abnormal network performance indicators in the current operating scenario are determined; Based on the abnormal network performance indicators, a network state adjustment strategy matching the current operating scenario is generated, so as to adjust the network state of the current operating scenario based on the network state adjustment strategy.

[0006] Secondly, embodiments of this application also provide a network optimization device based on speed curve features. The device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform a network optimization method based on speed curve features as described above.

[0007] Thirdly, embodiments of this application also provide a computer storage medium storing computer-executable instructions, which, when executed, implement a network optimization method based on speed curve features as described above.

[0008] The network optimization method and device based on velocity curve features provided in this application have the following beneficial effects: First, by acquiring the time-series throughput curve data under the current operating scenario and removing the first quarter of the initial unstable data, non-steady-state noise such as handshake negotiation and cache filling during the speed test startup phase can be effectively filtered out, retaining the stable segment that reflects the actual network carrying capacity, laying a high signal-to-noise ratio data foundation for subsequent analysis. Next, based on the removed data, the network throughput fluctuation value, speed drop ratio, and change amount are accurately extracted. This quantifies the degree of fluctuation and frequency of abrupt changes in the network transmission process from the perspective of curve shape, overcoming the deficiency that traditional single mean indicators cannot reflect transmission stability. Subsequently, a network stability factor is constructed by weighting the above parameters to achieve an objective prediction of the current network transmission stability, accurately characterizing the negative experience characteristics of sudden speed fluctuations. Simultaneously... By combining environmental quality factors determined by co-channel interference intensity and queue congestion intensity, wireless environmental interference and internal link congestion can be independently quantified, thus effectively distinguishing the causes of network instability. Then, the integrity factor, stability factor, and environmental quality factor are multiplicatively fused to generate a network experience confidence score, which can realistically map the user's actual experience using the product's bottleneck effect mechanism. Furthermore, when the confidence score falls below a preset threshold, the system can automatically lock abnormal network performance indicators, achieving a leap from vague perception to precise location. Finally, targeted network state adjustment strategies are generated based on abnormal indicators, realizing closed-loop control from "passive speed measurement and evaluation" to "proactive experience optimization," thereby significantly improving network reliability and user satisfaction. Attached Figure Description

[0009] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart of a network optimization method based on velocity curve features provided in this application embodiment; Figure 2 This is a schematic diagram of the internal structure of a network optimization device based on speed curve features, provided as an embodiment of this application. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0011] It is understood that in the embodiments of this application, data related to user information (such as user accounts) is involved. When the embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with relevant laws, regulations and standards.

[0012] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0013] In the following description, the terms “first, second, ...” are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that “first, second, ...” may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0014] With the rapid development of mobile internet and fiber broadband access technologies, high-bandwidth, low-latency services such as high-definition video streaming, cloud gaming, and online conferencing have become important components of users' daily network activities. To measure network service quality, network experience evaluation methods typically rely on network speed testing, which measures the downlink or uplink throughput of terminal devices within a specific time period, using peak or average rates as the core indicators for evaluating network quality. In these network speed testing systems, the results usually output directly collected basic network performance indicators or the maximum / average value of a single / multiple sampling, and the results are assumed to be reliable. However, in real-world home networks, WiFi signal interference, channel contention, background applications consuming excessive bandwidth, sudden bandwidth allocation by operators, and the slow start mechanism of the TCP protocol can all cause drastic fluctuations in instantaneous throughput. This leads to the direct inclusion of process noise in the final estimated result during speed testing, making the estimated results unreliable and hindering the identification of network optimization methods.

[0015] In real-world applications, users often face the contradictory phenomenon of "speed test results meeting the standards, but actual service experience being sluggish," such as in home Wi-Fi... In a Fi environment, although speed test software can show instantaneous bandwidth of hundreds of megabits, due to wireless signal interference or network congestion, frequent buffering or image quality degradation still occurs when watching videos. However, the unreliability of speed test predictions makes it difficult to find network optimization methods that are suitable for the current application scenario, resulting in a poor user network experience.

[0016] Based on this, embodiments of this application provide a network optimization method based on speed curve features, which can accurately predict network user experience and achieve intelligent diagnosis and adaptive optimization of network user experience.

[0017] The technical solutions proposed in the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0018] Figure 1 This application provides a flowchart of a network optimization method based on speed curve characteristics. This method can be applied to various network optimization scenarios, such as home broadband and WiFi network diagnostics, cloud gaming and real-time esports platforms, high-definition video streaming services, enterprise remote office and VPN access assessment, and operator intelligent operation and maintenance and installation acceptance. Certain input parameters or intermediate results in the process can be manually adjusted to help improve accuracy.

[0019] This application provides a network optimization method based on speed curve characteristics. It should be noted that the execution entity in these embodiments can be a server or any terminal device with data processing capabilities. For example, the server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. The terminal device can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, in-vehicle terminal, etc., but is not limited to these.

[0020] like Figure 1 As shown in the figure, the network optimization method based on velocity curve features provided in this application embodiment specifically includes the following steps: Step 101: Obtain the timing throughput curve data for the current running scenario.

[0021] It should be noted that the current operating scenario refers to the technical environment or carrier in which the software runs. In this application, the current operating scenario may include client / APP and web page. Time-series throughput curve data refers to the change data of network throughput (the amount of data transmitted per unit time) recorded in chronological order over a period of time. It is usually plotted as a curve with time on the horizontal axis and throughput on the vertical axis.

[0022] As an example, when a user initiates a network experience evaluation request, the system starts the background data collection module and calls the corresponding underlying interface according to the current operating scenario (e.g., PC client, mobile APP, or web browser) to begin periodic network probing. Subsequently, the system sends specific probe data packets (usually TCP or UDP streams) and records the data transmission volume in each sampling period. Assuming a preset sampling interval of 100ms, the system calculates the actual downlink / uplink throughput (unit: Mbps) within each 100ms time window, forming a discrete data point. Based on the duration of the speed measurement process, these continuous data points are arranged in chronological order to obtain the corresponding time-series throughput curve data. When the speed measurement duration reaches the preset end time, the system outputs the collected time-series throughput curve data, thus obtaining the time-series throughput curve data under the current operating scenario.

[0023] Step 102: Remove the first quarter of the initial unstable data from the time-series throughput curve data to obtain the time-series throughput curve data after removal.

[0024] It should be noted that, in the initial stage of the speed test, the TCP protocol goes through a slow start phase, and the congestion window has not yet opened. At this time, the measured throughput is much lower than the actual bandwidth capacity. In addition, at the beginning of the speed test, the system needs to establish a connection, perform a three-way handshake, and negotiate initial parameters. This part of the data will generate additional signaling overhead and latency, causing abnormal lows in the throughput curve. Therefore, in order to eliminate the impact of TCP slow start and filter the connection establishment overhead, the first quarter of the initial unstable data in the time-series throughput curve data is removed to ensure the accuracy of subsequent calculations.

[0025] As an example, suppose a user initiates a 10-second broadband speed test on a PC client with a sampling interval of 100ms, monitoring the network card's data reception in real time. Within 10 seconds, a total of 100 consecutive throughput data points (i.e., time-series throughput curve data) are collected, spanning from T0 (second 0) to T9.9 (second 9.9). To eliminate the impact of TCP slow start and filter connection establishment overhead, the first quarter of the initial unstable data in the time-series throughput curve is truncated, that is, the first 25 data points are discarded, and only the data from the 26th data point to the 100th data point is retained, that is, the remaining 75 data points constitute the time-series throughput curve data after truncation.

[0026] Step 103: Based on the removed time-series throughput curve data, determine the network throughput fluctuation value, network speed drop ratio, and network throughput change in the current operating scenario.

[0027] It should be noted that network throughput fluctuation value is a statistical measure used to quantify the dispersion of throughput during speed measurement, reflecting the stability of network transmission rate; network speed drop ratio refers to the frequency of sampling points where the throughput is significantly lower than the normal level during the speed measurement; network throughput change refers to the degree of drastic change in the speed measurement curve between adjacent time points, reflecting the frequency of throughput jitter.

[0028] In some embodiments, step 103 described above can be implemented as follows: obtaining the 10th percentile throughput value, the 50th percentile throughput value, and the 90th percentile throughput value from the removed time-series throughput curve data; determining that the 90th percentile throughput value is subtracted from the 10th percentile throughput value to obtain a first difference; using the ratio between the first difference and the 50th percentile throughput value as the network throughput fluctuation value; using a preset proportion of the 50th percentile throughput value as a rate drop threshold; counting the number of sampling points with throughput values ​​less than the rate drop threshold, and using the ratio between the number of sampling points with throughput values ​​less than the rate drop threshold and the total number of sampling points as the network rate drop ratio; determining the average throughput change between adjacent sampling points in the removed time-series throughput curve data, and using the ratio between the average throughput change and the 50th percentile throughput value as the network throughput change.

[0029] Thus, by utilizing the fluctuation data constructed from the 10th, 50th, and 90th percentile values, the relative dispersion of throughput can be effectively quantified to eliminate absolute numerical differences across different bandwidth levels, making the evaluation results comparable across network environments. Furthermore, by statistically analyzing the rate drop ratio using a preset proportional threshold, abnormal low frequencies in network transmission can be accurately captured, thereby effectively identifying persistent performance degradation caused by interference or congestion. The average change intensity based on the difference between adjacent sampling points keenly reflects the severity of jitter in the throughput curve, compensating for the deficiency that relying solely on throughput levels cannot reflect instantaneous fluctuations. This solution transforms the time-series curve into fluctuation, rate drop, and change intensity indicators with clear physical meaning, providing high-fidelity, interference-resistant feature inputs for the subsequent confidence model construction, ensuring the accuracy of subsequent confidence assessments.

[0030] It should be noted that percentile throughput values ​​refer to representative throughput values ​​extracted from time-series throughput curve data to describe the overall distribution of the speed measurement process. The 10th percentile throughput value is used to represent the representative throughput value of the lower throughput range during the speed measurement process, the 50th percentile throughput value is used to represent the median throughput value during the speed measurement process, and the 90th percentile throughput value is used to represent the representative throughput value of the higher throughput range during the speed measurement process.

[0031] As an example, suppose that during the speed test of a home gigabit broadband (1000Mbps), the system obtains 80 valid time-series sampling points after removing the first quarter of unstable data; then, the system sorts the 80 sampling points in ascending order and calculates the 10th percentile throughput value (P10) as 350Mbps, the 50th percentile throughput value (P50) as 920Mbps, and the 90th percentile throughput value (P90) as 980Mbps; then, the corresponding network throughput fluctuation value Spread, network speed drop ratio Drop_ratio, and network throughput change amount Slope_energy can be calculated according to formulas (1), (2), and (3), respectively.

[0032] (1) (2) (3) in, This is the 10th percentile throughput value. This is the 50th percentile throughput value. This is the 90th percentile throughput value. This represents the instantaneous throughput value at each measurement point during the speed measurement process. For the proportional parameter ( It can be equal to 0.6). Used to indicate the normal throughput level of the current network. Used to indicate the number of sampling points that meet the conditions. This represents the total number of sampling points. This represents the throughput change between the i-th sampling point and the (i-1)-th sampling point (i.e., adjacent sampling points). This represents the absolute value of the change in throughput. Used to represent the average intensity of change.

[0033] Step 104: Based on the network throughput fluctuation value, the network speed drop ratio, and the network throughput change, determine the network stability factor for the current operating scenario to predict the network transmission stability of the current operating scenario.

[0034] It should be noted that the network stability factor is a core indicator used to quantitatively evaluate the smoothness of the speed test process. The value ranges from 0 to 1, and the higher the value, the better the network stability.

[0035] In some embodiments, the determination of the network stability factor for the current operating scenario based on the network throughput fluctuation value, the network speed drop ratio, and the network throughput change in step 104 above can be achieved in the following way: normalize the network throughput fluctuation value, the network speed drop ratio, and the network throughput change to obtain network throughput fluctuation penalty, network speed drop frequency penalty, and network curve oscillation penalty; based on a preset first weight, perform a weighted summation of the network throughput fluctuation penalty, the network speed drop frequency penalty, and the network curve oscillation penalty to obtain a first weighted summation result; subtract the value 1 from the first weighted summation result to obtain the network stability factor.

[0036] In this way, by normalizing the fluctuations, decelerations, and oscillations with different dimensions and mapping them to a standard scale, the weight imbalance caused by differences in numerical ranges between different indicators can be eliminated, ensuring the fairness of subsequent calculations. Then, by using preset weights for weighted summation, the overall degree of network instability can be accurately quantified, making the evaluation results more in line with the real network experience. Finally, by mapping the accumulated penalty value inversely to a more intuitive stability factor, a natural transformation from multi-dimensional negative features to a single positive score can be achieved. Thus, at the algorithm level, a logically rigorous, hierarchical, and highly interpretable quantitative evaluation model for network stability can be constructed, significantly improving the accuracy of confidence determination.

[0037] It should be noted that the network throughput fluctuation penalty is the result of normalizing the network throughput fluctuation value, used to quantify the relative dispersion of network throughput; the network frame drop frequency penalty is the result of normalizing the network speed drop ratio, used to measure the frequency of abnormal lows in network transmission; and the network curve oscillation penalty is the result of normalizing the network throughput change, used to characterize the degree of drastic change in throughput between adjacent sampling points.

[0038] As an example, assuming that during the speed test of a home gigabit broadband (1000Mbps), after removing the first quarter of unstable data, the system obtains 80 valid time-series sampling points. Using the formulas (1)-(3) above, the network throughput fluctuation value Spread is calculated to be 0.68, the network speed drop ratio Drop_ratio is 0.15, and the network throughput change Slope_energy is 0.16. Subsequently, the network throughput fluctuation value, network speed drop ratio, and network throughput change are normalized respectively. For the network throughput fluctuation value, the system converts the ratio of 0.68 into a standardized penalty value (e.g., 0.7) based on the normal fluctuation range set by historical data, thus obtaining a network throughput fluctuation penalty of 0.7. For the network speed drop ratio, the system converts a 15% speed drop ratio into another standardized penalty value (e.g., 0.3) based on a preset severity standard, resulting in a network speed drop frequency penalty of 0.3. For network throughput changes, the system also normalizes the change intensity of 0.16 to obtain a standardized penalty value (e.g., 0.1), resulting in a network curve oscillation penalty of 0.1. Then, the system performs a weighted summation of the above three penalty values ​​based on preset first weights (e.g., network throughput fluctuation penalty of 0.5, network speed drop frequency penalty of 0.3, and network curve oscillation penalty of 0.2), resulting in a first weighted summation result of 0.42. Finally, subtracting the first weighted summation result from the value 1 yields a network stability factor of 0.58.

[0039] In some embodiments, the first weighted summation of the network throughput fluctuation penalty, the network speed drop frequency penalty, and the network curve oscillation penalty based on a preset first weight can be achieved as follows: Continuous throughput calculation is performed on the removed time-series throughput curve data to obtain a continuous throughput index for the current operating scenario; the ratio between the instantaneous download speed of the network in the current operating scenario and the continuous throughput index is used as the burst ratio; the burst ratio is truncated to a non-negative value of 1 to obtain a burst intensity index, and the burst intensity index is normalized to obtain a burst penalty; the first weighted summation of the network throughput fluctuation penalty, the network speed drop frequency penalty, the network curve oscillation penalty, and the burst penalty based on a preset second weight is obtained.

[0040] Thus, by introducing a sustained throughput metric as a benchmark for the network's true carrying capacity, the burst ratio is used to accurately quantify the deviation between instantaneous peak values ​​and sustainable capacity. Furthermore, by eliminating invalid negative values ​​through non-negative truncation, the calculation of burst intensity focuses more on the core issue of "unsustainable excess traffic." In addition, the normalized burst penalty is incorporated into the weighted summation system. Combined with the second weight, this allows for dynamic correction and improvement of the original stability assessment model. This effectively suppresses the interference of false high throughput caused by TCP slow start or link scheduling on the confidence level, thereby significantly improving the credibility and objectivity of the speed measurement results in scenarios where the Burst phenomenon exists.

[0041] It should be noted that the instantaneous download speed can be the average download speed or the peak download speed output by the speed test system.

[0042] As an example, assuming a home WiFi scenario, the time-series throughput curve data after removal is obtained. By performing continuous throughput calculation on the removed time-series throughput curve data, specifically referring to formula (1), the continuous throughput index sustained is obtained as 48Mbps. Then, the system measured the peak download speed (i.e., instantaneous download speed) as 52Mbps in this speed test, and the ratio between the instantaneous download speed and the continuous throughput index is 1.08, i.e., the burst ratio is 1.08. Subsequently, the burst ratio and the value 1 are truncated to non-negative values, and the burst intensity index is obtained as 0.08. The burst intensity index is normalized, and the burst penalty is obtained as 0.1. Then, the system calculates the weighted sum of the above four penalty values ​​according to the preset second weight (e.g., network throughput fluctuation penalty accounts for 0.4, network speed drop frequency penalty accounts for 0.25, network curve oscillation penalty accounts for 0.15, and burst penalty accounts for 0.2), and the first weighted sum is obtained as 0.39.

[0043] (4) (5) (6) in, It refers to the 50th percentile throughput value, that is, the median of the throughput value during the speed measurement process. For details, please refer to formula (5). It refers to the 90th percentile throughput value, which is the representative value of the higher throughput range during the speed measurement process. For details, please refer to formula (6). It refers to the i-th sample value after all sample values ​​are sorted in ascending order; This represents the number of sampled values.

[0044] For example, if a speed measurement yields 10 sampled values ​​S=[18,20,10,15,22,17,12,19,8,16], i.e., n=10, after sorting in ascending order, we can obtain S_sorted=[8,10,12,15,16,17,18,19,20,22]. Then, according to the above formulas (2) and (3), we know that... sort[5] = 17Mbps, The value of sort[9] is 22Mbps.

[0045] Step 105: Based on the co-channel interference intensity and queue congestion intensity of the current operating scenario, determine the environmental quality factor of the current operating scenario to estimate the network smoothness of the current operating scenario.

[0046] It should be noted that co-channel interference intensity refers to interference factors from outside the network transmission path that affect the local environment of the speed measurement terminal; queue congestion intensity refers to queuing and buffering phenomena caused by traffic overload within the network transmission path; the environmental quality factor is used to comprehensively reflect the quality of the current network environment, and can be calculated from multiple dimensions such as co-channel interference intensity (e.g., WiFi interference, signal fluctuation) and queue congestion intensity (e.g., queuing delay, load jitter), with values ​​ranging from 0 to 1, and higher values ​​indicating a better network environment.

[0047] In some embodiments, the determination of the environmental quality factor of the current operating scenario based on the co-channel interference intensity and queue congestion intensity in step 105 above can be achieved in the following way: based on a preset fifth weight, the co-channel interference intensity and the queue congestion intensity are weighted and summed to obtain a second weighted summation result; the value 1 is subtracted from the second weighted summation result to obtain an initial environmental quality factor; in response to detecting WiFi interference in the current operating environment, the interference penalty coefficient corresponding to the current operating environment is obtained; the interference penalty coefficient is multiplied by the initial environmental quality factor to obtain the environmental quality factor of the current operating scenario.

[0048] Thus, by introducing a fifth weight to weight and fuse the two heterogeneous dimensions of external network interference and network congestion, an environmental quality quantification model is constructed that can distinguish interference sources and make comprehensive judgments. A linear transformation converts the positively accumulated environmental severity index into a negatively decreasing environmental quality factor, maintaining mathematical consistency with the calculation paradigm of the stability factor to ensure the coherence and interpretability of subsequent confidence calculations. This effectively achieves accurate separation and comprehensive evaluation of local wireless environment interference and wide area network congestion. Furthermore, by introducing an interference penalty coefficient and dynamically multiplying it with the original environmental factor, accurate correction and degradation of environmental quality in WiFi interference scenarios can be effectively achieved. This allows the final environmental quality factor to more realistically reflect the severity of wireless signal interference, thereby improving the accuracy and reliability of the overall network stability assessment. Moreover, this solution not only enhances adaptability to complex network environments but also automatically reduces the confidence weight when significant interference exists to avoid misjudgment, thus significantly improving the system's robustness and practicality in diverse real-world scenarios.

[0049] It should be noted that if the current operating environment experiences significant throughput fluctuations (i.e., Spread > 0.8), frequent speed drops (i.e., Drop_ratio > 0.15), and a significant increase in load jitter (i.e., jitter_increase > 15ms), or if the current operating environment experiences a large increase in queuing latency under network load conditions (i.e., queue_latency > 40ms), a significant increase in load jitter (i.e., jitter_increase > 20ms), and a low base latency in idle state (i.e., idle_latency < 40ms), it indicates that there is WiFi interference in the current operating environment. Load jitter increment refers to the fluctuation range of network load (e.g., traffic rate, number of concurrent connections, number of requests, etc.) per unit time, used to measure the severity of load changes. Queuing latency refers to the waiting time consumed by the system (e.g., CPU, network interface) after a data packet, task, or request enters the processing queue and waits to be processed. Idle latency refers to the base response latency of network nodes (e.g., routers, servers) or the system when there is no load (no data packet processing, no task execution).

[0050] As an example, assuming a home gigabit broadband speed test scenario, the calculated co-channel interference intensity is 0.6 and the queue congestion intensity is 0.4. The system then uses a preset fifth weight (e.g., co-channel interference intensity 0.6, queue congestion intensity 0.4) to perform a weighted summation of the co-channel interference intensity and the queue congestion intensity, resulting in a second weighted summation of 0.52. Subtracting this second weighted summation from the value 1 yields an initial environmental quality factor of 0.48. Subsequently, by detecting that Spread > 0.8, Drop_ratio > 0.15, and jitter_increase > 15ms in the current operating scenario, it can be determined that WiFi interference exists in the current operating environment. The system then obtains the interference penalty coefficient k (e.g., 0.85) corresponding to the current operating environment and multiplies it by the environmental quality factor C_environment, obtaining the multiplication result k × C_environment, which is the environmental quality factor for the current operating scenario.

[0051] In some embodiments, before performing step 105 above, the following processing may also be performed: based on a preset third weight, the network throughput fluctuation value and the network speed drop ratio are weighted and summed to obtain the co-channel interference intensity, so as to estimate the network external interference intensity of the current operating scenario; based on the basic network performance indicators of the current operating scenario, the queue congestion intensity is determined to estimate the network congestion intensity of the current operating scenario.

[0052] Thus, by introducing preset weights to weighted summation of network throughput fluctuations and speed drop ratios, the intensity of co-channel interference can be quantified more accurately. This effectively distinguishes between fluctuations in device performance and the impact of the external environment, improving the objectivity of the assessment. Simultaneously, by combining basic network performance indicators of the current operating scenario to determine queue congestion intensity, the congestion assessment becomes more closely aligned with the actual network state, avoiding the bias of a single indicator. Furthermore, this solution achieves independent and accurate quantification of external network interference and internal congestion, providing a reliable data foundation for subsequent calculations of environmental quality factors. This significantly enhances the accuracy and practicality of network performance assessment, helping to more scientifically reflect the true quality of the network operating environment.

[0053] It should be noted that basic network performance indicators refer to raw network data directly detected and collected by speed test software without complex processing, such as download speed, upload speed, idle latency, download load latency, upload load latency, idle jitter, upload jitter, download jitter, packet loss rate, etc.; upload load latency refers to the network round-trip latency measured during data upload; download load latency refers to the network round-trip latency measured during data download; idle latency refers to the round-trip latency measured when sending probe packets to the server in the absence of any background traffic (i.e., idle state); idle jitter refers to the variation (variance or standard deviation) of the round-trip latency of data packets continuously measured in an idle network state (no large data transmission); upload jitter refers to the amount of change in the round-trip latency of data packets measured during data upload; download jitter refers to the amount of change in the round-trip latency of data packets measured during data download; packet loss rate refers to the proportion of data packets lost during network transmission, usually expressed as a percentage (%).

[0054] As an example, assuming that during the speed test of a home gigabit broadband (1000Mbps), after removing the first quarter of unstable data, the system obtains 80 valid time-series sampling points. Through the above formulas (1)-(2), the network throughput fluctuation value Spread is calculated to be 0.7 and the network drop ratio is 0.2. Then, according to the preset third weight (e.g., network throughput fluctuation value accounts for 0.4 and network drop ratio accounts for 0.6), the system performs a weighted summation of the network throughput fluctuation value and the network drop ratio, and the weighted summation result is 0.4, that is, the co-channel interference intensity of the current operating environment is 0.4. After that, according to the basic network performance indicators of the current operating scenario, the queue congestion intensity of the current operating environment is 0.3.

[0055] In some embodiments, determining the queue congestion intensity based on the basic network performance indicators of the current operating scenario can be achieved in the following ways: subtracting the load latency from the idle latency in the basic network performance indicators to obtain the network queuing latency, where the load latency is the maximum value of the upload load latency and download load latency in the basic network performance indicators; subtracting the load jitter from the idle jitter in the basic network performance indicators to obtain the network load jitter increment, where the load jitter is the maximum value of the upload jitter and download jitter in the basic network performance indicators; and weighting and summing the network queuing latency and the network load jitter increment based on a preset fourth weight to obtain the queue congestion intensity.

[0056] In this way, by subtracting idle latency from load latency, we can obtain the pure network queuing latency. By subtracting idle jitter from load jitter, we can obtain the network load jitter increment. This can effectively eliminate the interference of device hardware performance or physical distance, and only retain the real impact of network congestion. Furthermore, by using preset weights to perform a weighted summation of the two core dimensions, the final queue congestion intensity assessment result is both scientific and dynamically adaptable. It can keenly reflect whether the current network is slightly lagging or severely congested, providing solid and unbiased data support for subsequent environmental quality factor calculations.

[0057] It should be noted that network queuing latency refers to the time it takes for a data packet or request to be processed in a network device. It can reflect the current level of network congestion; the longer the queuing latency, the more congested the network. Network load jitter increment refers to the fluctuation range of network jitter under load conditions. It can reflect the impact of network load changes on stability; the larger the network load jitter increment, the more severe the impact of load changes on the network.

[0058] As an example, assuming a home gigabit broadband speed test scenario, by testing the current network, we can obtain the basic network performance indicators under the current operating scenario, including download load latency (150ms), upload load latency (80ms), idle latency (50ms), upload jitter (20ms), download jitter (30ms), and idle jitter (5ms). Then, by comparison, we know that the current network load latency equals the download load latency, that is, the load latency is 150ms. By subtracting the load latency from the idle latency, we can obtain the network queuing latency. The value is 100ms. Similarly, by comparison, it can be seen that the current network load jitter is equal to the download jitter, that is, the load jitter is 30ms. By subtracting the load jitter from the idle jitter, the network load jitter increment can be obtained as 25ms. Subsequently, the system calculates the weighted sum of the network queuing delay and the network load jitter according to the preset fourth weight (e.g., network queuing delay accounts for 0.7, and network load jitter increment accounts for 0.3), and the weighted sum result is 28.5. Then, the weighted sum result can be normalized to map the final result to the interval [0, 1], and the normalized result is used as the current queue congestion intensity.

[0059] In some embodiments, if some network performance metrics are missing during the speed test, they can be replaced with similar metrics (e.g., missing download metrics can be replaced with upload metrics, and missing load metrics can be replaced with idle metrics), the missing metrics can be removed and their weights re-normalized, or the confidence level can be reduced according to preset rules. This ensures that the confidence level of the current network speed test results can still be assessed even when network performance metrics are missing.

[0060] Step 106: Multiplicatively fuse the integrity factor, the network stability factor, and the environmental quality factor corresponding to the current operating scenario to obtain the network experience confidence of the current operating scenario. The integrity factor is used to characterize the network transmission integrity of the current operating scenario.

[0061] It should be noted that the integrity factor is an indicator that measures the integrity of data transmission in the current network operating environment. It can reflect the degree to which data is complete and undamaged during transmission. The higher the integrity factor, the more complete the data transmission. For example, the integrity factor for client / APP scenarios is 1.0, and the integrity factor for web page scenarios is 0.8. The network experience confidence score is a comprehensive index obtained by multiplying the integrity factor, network stability factor, and environmental quality factor. It is used to quantify the credibility of the current network speed test results. The higher the value, the higher the credibility of the current network speed test results.

[0062] As an example, assuming the current operating scenario is a web-based scenario, the corresponding integrity factor is 0.8. Using the above calculation method, the network stability factor and environmental quality factor corresponding to the current operating scenario are 0.7 and 0.6, respectively. Then, by multiplying the integrity factor, network stability factor, and environmental quality factor (i.e., multiplicative fusion), the network experience confidence score of the current operating scenario is 0.336.

[0063] In some embodiments, after performing step 106, the following processing may also be performed: generating a confidence assessment explanation text corresponding to the current operating scenario based on preset prompt words and the network experience confidence level of the current operating scenario; performing a verification process on the confidence assessment explanation text based on preset verification rules to obtain a verification result; in response to the verification result being passed, using the confidence assessment explanation text as the network experience confidence assessment result of the current operating scenario; in response to the verification result being failed, regenerating the confidence assessment explanation text of the current operating scenario, and verifying the regenerated confidence assessment explanation text until the verification is passed, and using the regenerated confidence assessment explanation text as the network experience confidence assessment result of the current operating scenario.

[0064] Thus, by introducing an explanatory text generation mechanism driven by "preset prompt words + confidence level," abstract numerical results can be transformed into semantic descriptions that users can understand, achieving a closed loop from data to decision-making. At the same time, by using verification rules to automatically review the compliance, accuracy, and logical consistency of the generated content, AI illusions or expression biases can be effectively avoided, ensuring the seriousness and credibility of the evaluation conclusions. Furthermore, through the loop logic of regeneration and re-verification, the system can have the ability to self-correct while ensuring the quality of the output text, thereby significantly improving the interpretability and user acceptance of the network experience evaluation results.

[0065] As an example, assuming that the confidence level of the network experience in the current operating scenario is 0.336 after the aforementioned confidence level assessment method; then, based on preset prompts (e.g., "Please generate a Chinese assessment of no more than 100 characters based on the following network experience confidence level score, and explain the main reasons for the current result"), the network experience confidence level and prompts are input into the large model. The large model will generate corresponding confidence level assessment explanation text based on the network experience confidence level, such as "The current network speed test assessment result has low credibility, mainly because the current network environment has obvious WiFi signal interference and multiple devices downloading, leading to network congestion." After "blocking," the current confidence assessment explanation text is validated according to preset validation rules (such as detecting the presence of prohibited words). If the validation result is passed, the current confidence assessment explanation text can be used as the network experience confidence assessment result for the current running scenario. If the validation fails, the prompt words are automatically adjusted (such as adding constraints), and the confidence assessment explanation text is regenerated based on the new prompt words. The regenerated confidence assessment explanation text is then validated according to preset validation rules until the validation passes. The validated confidence assessment explanation text is then used as the network experience confidence assessment result for the current running scenario.

[0066] Step 107: In response to the network experience confidence level being less than a preset confidence threshold, determine the abnormal network performance indicators in the current operating scenario.

[0067] As an example, taking a user watching a high-definition live stream on their mobile phone in their bedroom as an example, the system calculates the current network experience confidence level as 0.378 using the above method, which is lower than the preset threshold of 0.6, thus determining that the current network is abnormal. Subsequently, the system traces back the underlying indicators that constitute the current confidence level and finds that the "network speed drop ratio" is 30% (greater than the preset ratio of 10%), and the "co-channel interference intensity" is 0.8, which is significantly higher than the normal value of 0.4, while the integrity factor is normal. Therefore, the system marks the "high speed drop ratio" and "high co-channel interference" as abnormal network performance indicators that cause the poor network experience, providing a basis for generating adjustment strategies in the future.

[0068] Step 108: Based on the abnormal network performance indicators, generate a network state adjustment strategy that matches the current operating scenario, and adjust the network state of the current operating scenario based on the network state adjustment strategy.

[0069] Continuing with the example above, in the scenario where a user is watching a live stream in their bedroom, after the system identifies the abnormal network performance indicators as "high speed drop rate" and "high co-channel interference," it also recognizes the current operating scenario as "the mobile terminal is on 2.4G Wi-Fi." The system first performs video streaming under Fi; then, based on the current operating scenario and abnormal network performance indicators, it generates a matching network status adjustment strategy: "On the one hand, it suggests that the router switch the working channel to a channel with less interference, and on the other hand, it instructs the live streaming application to temporarily reduce the playback resolution and increase the buffer window"; then, according to the generated network status adjustment strategy, the system executes the corresponding adjustment instructions, that is, it sends a channel adjustment request to the router and restricts the download tasks of the background app store on the mobile phone, thereby effectively suppressing network congestion and interference and significantly improving the smoothness of the live streaming picture.

[0070] In some embodiments, the generation of a network state adjustment strategy matching the current operating scenario based on the abnormal network performance indicators in step 108 above can be implemented in the following way: determining the indicator type of the abnormal network performance indicators, wherein the indicator type includes at least one of stability anomaly, environmental interference anomaly, and transmission integrity anomaly; in response to the indicator type of the abnormal network performance indicators being stability anomaly, generating a first adjustment strategy, wherein the first adjustment strategy is used to indicate adjusting the service bit rate, increasing the buffer duration, or limiting background traffic; in response to the indicator type of the abnormal network performance indicators being environmental interference anomaly, generating a second adjustment strategy, wherein the second adjustment strategy is used to indicate adjusting the wireless channel, switching the operating frequency band, or reducing the intensity of external interference; in response to the indicator type of the abnormal network performance indicators being transmission integrity anomaly, generating a third adjustment strategy, wherein the third adjustment strategy is used to indicate enabling the error correction and retransmission mechanism or switching the transmission link; and using at least one of the first adjustment strategy, the second adjustment strategy, and the third adjustment strategy as the network state adjustment strategy matching the current operating environment.

[0071] In this way, by subdividing abnormal network performance indicators into three categories—stability, environmental interference, and transmission integrity—and mapping specific adjustment strategies to different types, it is possible to accurately locate the root cause of the fault and differentiate the matching of countermeasures, thereby avoiding blind adjustments. Furthermore, this classification response mechanism enables the system to prioritize optimizing service bitrate and buffering when stability is poor, switch channels or frequency bands when interference is severe, and activate error correction or link switching when integrity is compromised. This ensures that adjustment actions are targeted while maximizing the balance between user experience and network resources, thereby significantly improving the intelligence level and actual optimization effect of network status adjustment.

[0072] As an example, consider a user in their bedroom using their phone on a 2.4G Wi-Fi connection. Taking watching a high-definition live stream under Fi as an example, the system detected an abnormal indicator mainly manifested as an excessively high "network speed drop ratio," which was classified as a stability anomaly. The system then generated a first adjustment strategy, instructing the live streaming application to reduce the playback resolution from 1080P to 720P and increase the buffer window to smooth out speed fluctuations. Simultaneously, background downloads from the app store were restricted on the mobile device, significantly reducing screen stuttering. Taking a user conducting a video conference on a tablet in the kitchen as another example, the system identified the anomaly as stemming from a surge in "co-channel interference intensity," determining it to be an environmental interference anomaly. The system then generated a corresponding second adjustment strategy, which involved sending a command to the router to adjust the operating channel... On the one hand, the system switches to an idle channel, and on the other hand, it guides the tablet to switch from the heavily interfered 2.4G frequency band to the cleaner 5G frequency band to effectively reduce the impact of external wireless interference on real-time calls. Taking the example of a user downloading a large file through a mobile phone in a weak signal area, the system finds that the "integrity factor" is extremely low and accompanied by a large number of packet losses, which is defined as a transmission integrity anomaly. Then, a third adjustment strategy is generated, which is to enable a stronger forward error correction and retransmission mechanism at the transport layer to improve the data arrival rate. At the same time, a link switching prompt is triggered when necessary, suggesting that the user move to a location with better signal or switch to a cellular network, thereby ensuring the integrity and success rate of file transmission.

[0073] The above are embodiments of the method proposed in this application. Based on the same inventive concept, embodiments of this application also provide a network optimization device based on speed curve characteristics, the structure of which is as follows: Figure 2 As shown.

[0074] Figure 2 This is a schematic diagram of the internal structure of a network optimization device based on velocity curve features, provided as an embodiment of this application. Figure 2 As shown, the device includes: At least one processor 201; And a memory 202 that is communicatively connected to at least one processor; The memory 202 stores instructions that can be executed by at least one processor. The instructions are executed by at least one processor 201 to enable at least one processor 201 to perform the steps of the method corresponding to any of the above embodiments.

[0075] Some embodiments of this application provide corresponding to Figure 1 A non-volatile computer storage medium stores computer-executable instructions, which are configured to perform the steps of the method corresponding to any of the above embodiments.

[0076] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for IoT devices and media are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0077] The systems, media, and methods provided in this application are one-to-one correspondences. Therefore, the systems and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the systems and media will not be repeated here.

[0078] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0079] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0080] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0081] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0082] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0083] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0084] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0085] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0086] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A network optimization method based on velocity curve features, characterized in that, The method includes: Obtain the timing throughput curve data under the current running scenario, and remove the first quarter of the initial unstable data from the timing throughput curve data to obtain the timing throughput curve data after removal. Based on the removed time-series throughput curve data, determine the network throughput fluctuation value, network speed drop ratio, and network throughput change in the current operating scenario. Based on the network throughput fluctuation value, the network speed drop ratio, and the network throughput change, a network stability factor for the current operating scenario is determined to predict the network transmission stability of the current operating scenario. Based on the intensity of co-channel interference and queue congestion in the current operating scenario, the environmental quality factor of the current operating scenario is determined to predict the network smoothness of the current operating scenario. The network experience confidence score of the current operating scenario is obtained by multiplicatively fusing the integrity factor, the network stability factor, and the environmental quality factor corresponding to the current operating scenario. The integrity factor is used to characterize the network transmission integrity of the current operating scenario. In response to the network experience confidence level being less than a preset confidence threshold, abnormal network performance indicators in the current operating scenario are determined; Based on the abnormal network performance indicators, a network state adjustment strategy matching the current operating scenario is generated, so as to adjust the network state of the current operating scenario based on the network state adjustment strategy.

2. The method according to claim 1, characterized in that, The step of generating a network state adjustment strategy that matches the current operating scenario based on the abnormal network performance indicators includes: The type of the abnormal network performance metric is determined, and the metric type includes at least one of stability anomaly, environmental interference anomaly, and transmission integrity anomaly. In response to the abnormal network performance indicator being of the type of stability anomaly, a first adjustment strategy is generated, which is used to indicate adjusting the service bitrate, increasing the buffer duration, or limiting the background traffic. In response to the abnormal network performance indicator being of the type of environmental interference anomaly, a second adjustment strategy is generated. The second adjustment strategy is used to indicate adjusting the wireless channel, switching the operating frequency band, or reducing the intensity of external interference. In response to the abnormal network performance indicator being of the type of transmission integrity abnormality, a third adjustment strategy is generated, which is used to indicate whether to enable the error correction and retransmission mechanism or switch the transmission link. At least one of the first adjustment strategy, the second adjustment strategy, and the third adjustment strategy shall be used as the network state adjustment strategy that matches the current operating environment.

3. The method according to claim 1, characterized in that, After multiplicatively fusing the integrity factor, network stability factor, and environmental quality factor corresponding to the current operating scenario to obtain the network experience confidence level of the current operating scenario, the method further includes: Based on preset prompt words and the network experience confidence level of the current operating scenario, generate a confidence level evaluation explanation text corresponding to the current operating scenario; Based on preset verification rules, the confidence assessment explanation text is verified to obtain the verification result; In response to the verification result being passed, the confidence assessment explanation text is used as the network experience confidence assessment result for the current operating scenario; In response to the verification result being unsuccessful, the confidence assessment explanation text for the current operating scenario is regenerated, and the regenerated confidence assessment explanation text is verified until the verification passes. The regenerated confidence assessment explanation text is then used as the network experience confidence assessment result for the current operating scenario.

4. The method according to claim 1, characterized in that, The process of determining the network throughput fluctuation value, network speed drop ratio, and network throughput change under the current operating scenario based on the removed time-series throughput curve data includes: Obtain the 10th percentile throughput value, 50th percentile throughput value, and 90th percentile throughput value from the removed time-series throughput curve data; Subtracting the 90th percentile throughput value from the 10th percentile throughput value yields the first difference; The ratio between the first difference and the 50th percentile throughput value is taken as the network throughput fluctuation value; The 50th percentile throughput value, which is a preset ratio, is used as the rate drop threshold. The number of sampling points with throughput values ​​less than the rate drop threshold is counted, and the ratio between the number of sampling points with throughput values ​​less than the rate drop threshold and the total number of sampling points is taken as the network rate drop ratio. The average throughput change between adjacent sampling points in the removed time-series throughput curve data is determined, and the ratio between the average throughput change and the 50th percentile throughput value is taken as the network throughput change.

5. The method according to claim 1, characterized in that, The process of determining the network stability factor for the current operating scenario based on the network throughput fluctuation value, the network speed drop ratio, and the network throughput change includes: The network throughput fluctuation value, the network speed drop ratio, and the network throughput change are normalized respectively to obtain the network throughput fluctuation penalty, the network speed drop frequency penalty, and the network curve oscillation penalty. Based on a preset first weight, the network throughput fluctuation penalty, the network speed drop frequency penalty, and the network curve oscillation penalty are weighted and summed to obtain the first weighted summation result; Subtracting the value 1 from the first weighted summation result yields the network stability factor.

6. The method according to claim 5, characterized in that, The first weighted summation, based on a preset first weight, is performed on the network throughput fluctuation penalty, the network speed drop frequency penalty, and the network curve oscillation penalty to obtain a first weighted summation result, including: The continuous throughput calculation is performed on the removed time-series throughput curve data to obtain the continuous throughput index under the current operating scenario; The ratio between the instantaneous network download speed and the sustained throughput metric in the current operating scenario is used as the burst ratio; The burst ratio and the value 1 are truncated to obtain the burst intensity index, and the burst intensity index is normalized to obtain the burst penalty. Based on the preset second weight, the network throughput fluctuation penalty, the network speed drop frequency penalty, the network curve oscillation penalty, and the burst penalty are weighted and summed to obtain the first weighted summation result.

7. The method according to claim 1, characterized in that, Before determining the environmental quality factor of the current operating scenario based on the co-frequency interference intensity and queue congestion intensity of the current operating scenario, the method further includes: Based on a preset third weight, the network throughput fluctuation value and the network speed drop ratio are weighted and summed to obtain the co-frequency interference intensity, so as to estimate the network external interference intensity of the current operating scenario. Based on the basic network performance indicators of the current operating scenario, the queue congestion intensity is determined in order to estimate the network congestion intensity of the current operating scenario.

8. The method according to claim 7, characterized in that, The determination of queue congestion intensity based on the basic network performance indicators of the current operating scenario includes: Subtracting the load latency from the idle latency in the basic network performance metrics yields the network queuing latency, where the load latency is the maximum value of the upload load latency and the download load latency in the basic network performance metrics. The load jitter is subtracted from the idle jitter in the basic network performance indicators to obtain the network load jitter increment, where the load jitter is the maximum value of the upload jitter and download jitter in the basic network performance indicators. Based on a preset fourth weight, the network queuing delay and the network load jitter increment are weighted and summed to obtain the queue congestion intensity.

9. The method according to claim 1, characterized in that, The determination of the environmental quality factor for the current operating scenario based on the intensity of co-frequency interference and queue congestion includes: Based on the preset fifth weight, the co-frequency interference intensity and the queue congestion intensity are weighted and summed to obtain the second weighted summation result; Subtract the value 1 from the second weighted summation result to obtain the initial environmental quality factor; In response to the detection of WiFi interference in the current operating environment, the interference penalty coefficient corresponding to the current operating environment is obtained; The interference penalty coefficient is multiplied by the initial environmental quality factor to obtain the environmental quality factor of the current operating scenario.

10. A network optimization device based on velocity curve characteristics, characterized in that, The device includes: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method as described in any one of claims 1-9.

Images