A telephone line switching device and method

Abstract

The application discloses a telephone line switching device and method, and relates to the technical field of telephone communication, and comprises the following steps: collecting the nominal maximum bandwidth and the actual used bandwidth of each line, obtaining line reference load data based on the nominal maximum bandwidth and the actual used bandwidth; obtaining the actual available bandwidth of the line, adjusting the line reference load data based on the actual available bandwidth, and obtaining line capacity data; obtaining individual sudden drop threshold values through a dynamic adaptive threshold model based on the line capacity data, and judging whether a multi-line synchronous bandwidth sudden drop event occurs based on the individual sudden drop threshold values; the application aims at the problem of bandwidth instantaneous sudden drop caused by the synchronous triggering of the heat protection mechanism of multiple lines in extreme environments such as high temperature, sets individual sudden drop threshold values, screens a bandwidth sudden drop line list, and quickly matches standby switching lines, so that fast response and switching in the multi-line synchronous sudden drop scene are realized, large-scale business interruption is avoided, and communication continuity is ensured.

Description

Technical Field

[0001] This invention relates to the field of telephone communication technology, and in particular to a telephone line switching device and method. Background Technology

[0002] In modern communication infrastructure such as data centers and core computer rooms, the deployment density of telephone lines has increased significantly with the popularization of technologies such as 5G and the Internet of Things. The services carried have also expanded to include critical services with extremely high continuity requirements, such as remote control and real-time video transmission. Existing technologies generally adopt load-rate-based line switching schemes, as revealed in comparative documents CN111147366B and CN104094564A. These schemes determine and switch lines by monitoring the real-time load rate of each line, i.e., the ratio of actual used bandwidth to nominal maximum bandwidth. Such schemes are effective in dealing with single-line failures or general overloads.

[0003] However, existing technologies have significant limitations, particularly in dealing with systemic concurrent failures caused by extreme environments. Under harsh conditions such as high temperatures and unstable voltage, densely deployed communication equipment hardware (such as optical modules and network interface cards) will simultaneously trigger thermal protection mechanisms. This mechanism does not cause a complete line outage, but rather forcibly limits its available bandwidth to a low percentage of its nominal value, such as below 30%, within milliseconds. At this time, the actual available bandwidth of multiple lines will experience a synchronous and correlated step-like drop, forming a collective bandwidth collapse phenomenon.

[0004] In this scenario, existing technologies reveal fundamental flaws:

[0005] The load assessment model failed. The assessment based on relative load rate could not distinguish the absolute capacity difference between high and low bandwidth lines, nor could it detect bandwidth synchronous degradation caused by external environment rather than by the service itself.

[0006] When the switching decision logic fails, traditional systems based on a single thread and isolated judgment (such as CN103024086A and CN103095571A) will incorrectly determine that all lines are unavailable due to the lack of identification and response mechanisms for collective performance degradation. They will be unable to find a qualified backup line, ultimately leading to large-scale and long-term service interruptions.

[0007] Therefore, the shortcomings of existing technologies in terms of the accuracy of dynamic resource assessment and the mechanism for responding to systemic risks make it difficult to guarantee the communication continuity of critical services under extreme environments. There is an urgent need for a switching device and method that can accurately assess the actual transmission capacity of lines, intelligently identify correlation performance degradation, and quickly match stable communication paths in the event of systemic risks. Summary of the Invention

[0008] The purpose of this invention is to solve the problem that when multiple communication lines are simultaneously triggered by thermal protection under extreme environments such as high temperature, the bandwidth drops suddenly, leading to a large-scale service interruption. Therefore, a new telephone line switching device and method are proposed.

[0009] To achieve the above objectives, the present invention employs the following telephone line switching method, comprising the following steps:

[0010] Collect the nominal maximum bandwidth and actual used bandwidth of each line, and obtain the line baseline load data based on the nominal maximum bandwidth and actual used bandwidth;

[0011] Obtain the actual available bandwidth of the line, and adjust the line baseline load data based on the actual available bandwidth to obtain the line capacity data;

[0012] Based on line capacity data, an individual bandwidth drop threshold is obtained through a dynamic adaptive threshold model. Based on the individual bandwidth drop threshold, it is determined whether a multi-line synchronous bandwidth drop event has occurred. If so:

[0013] Based on the individual bandwidth drop threshold and the absolute bandwidth data of the line, a list of lines with bandwidth drops greater than the individual bandwidth drop threshold is obtained.

[0014] Based on the list of lines with sudden bandwidth drops and the absolute bandwidth data of the lines, a backup switching line is obtained. Based on the backup switching line, the line is switched to obtain the switched line. Real-time data of the switched line is collected, and the accuracy of the switched line is verified. The accuracy verification includes real-time performance verification and time-series-based performance trend prediction verification. If the switched line passes the accuracy verification, the line switching is complete.

[0015] Furthermore, methods for filtering to obtain a list of bandwidth drop lines that exceed an individual drop threshold include:

[0016] From the line capacity data, the average available bandwidth of each line over the past three monitoring periods is extracted as the historical baseline bandwidth; based on the historical baseline bandwidth and the actual used bandwidth, the bandwidth drop of each line is calculated.

[0017] By comparing the bandwidth drop of each line with the individual drop threshold, a preliminary set of lines with bandwidth drops greater than the individual drop threshold is obtained.

[0018] Based on the initial set of lines, real-time bandwidth data within a preset time period is collected. Based on the real-time bandwidth data, the average bandwidth within the preset time period is calculated. Based on the average bandwidth, the average bandwidth drop magnitude is calculated, and the average bandwidth drop magnitude is compared with the individual drop threshold. If the average bandwidth drop magnitude is greater than the individual drop threshold, the line corresponding to the average bandwidth is retained, resulting in a list of lines with bandwidth drops.

[0019] Furthermore, methods for obtaining backup switching lines based on a list of lines experiencing sudden bandwidth drops and absolute bandwidth data include:

[0020] Extract the identifier, connection node pairs, and historical baseline bandwidth for each line from the list of lines experiencing a sudden bandwidth drop.

[0021] Based on the connection nodes of the bandwidth drop line list, and through node pair consistency constraints, all normal lines that are completely consistent with the node pairs of the drop lines are filtered out from the line capacity data.

[0022] Based on the current bandwidth of the normal line and the historical baseline bandwidth of the line with sudden drop, the bandwidth adaptation value between the normal line and the corresponding line with sudden drop is calculated; the bandwidth adaptation value is compared with the preset adaptation value threshold to select the suitable line.

[0023] The load fluctuation data of the adapted lines is collected. Based on the load fluctuation data, the load standard deviation is calculated through load stability constraints. The load standard deviation is compared with the standard deviation threshold to screen out lines with stable load.

[0024] Based on the bandwidth adaptation value of the load-stable line and the load fluctuation data, the reserve value of the load-stable line is calculated; based on the reserve value of the load-stable line, the backup switching line is obtained.

[0025] Furthermore, methods for obtaining individual sudden drop thresholds based on line capacity data through a dynamic adaptive threshold model include:

[0026] The historical available bandwidth sequence for each line over the past 30 working days is extracted from the line capacity data to construct the real-time available bandwidth sequence. The dynamic fluctuation baseline of the normal fluctuation range of the line is obtained through statistical modeling.

[0027] Based on the dynamic fluctuation baseline, its standard deviation is calculated, and the value of K times the standard deviation is subtracted from the dynamic fluctuation baseline to determine the lower limit of the healthy fluctuation of the line.

[0028] The search line calculates its historical maximum bandwidth reduction ratio over the past statistical period, and obtains historical extreme degradation characteristics based on the historical maximum bandwidth reduction ratio;

[0029] Based on the lower limit of health fluctuations, the individual sudden drop threshold is generated by integrating the historical extreme degradation characteristics of the line, the preset line importance level weight, and the real-time traffic load ratio through weighted calculation.

[0030] Furthermore, methods for determining whether a multi-line synchronous bandwidth drop event has occurred include:

[0031] The average real-time bandwidth of the line is collected. Based on the average real-time available bandwidth of the line capacity data and the average real-time bandwidth of the line, the bandwidth drop of the line is calculated.

[0032] The bandwidth drop magnitude of the line is compared with the individual drop threshold to filter out preliminary lines whose bandwidth drop magnitude is greater than the individual drop threshold.

[0033] The real-time average value of the preliminary line is collected and compared with the fluctuation threshold value in the line capacity data. If the real-time average value is less than the fluctuation threshold value in the line capacity data, the bandwidth drop of the preliminary line is greater than the individual drop threshold.

[0034] Furthermore, methods for collecting real-time data of the switched lines and verifying its accuracy include:

[0035] Collect the core performance parameters and auxiliary quality indicators of the line after the switchover;

[0036] Among them, the core performance parameters include real-time available bandwidth and actual used bandwidth, and the auxiliary quality indicators include transmission delay, jitter and bit error rate;

[0037] Based on core performance parameters, the bandwidth adequacy of the switched lines is verified; based on auxiliary quality indicators, the signal quality of the switched lines is verified.

[0038] If the switched line passes the bandwidth adequacy verification and the switched line passes the signal quality verification, the core performance parameters and auxiliary quality indicators of the switched line are collected during the preset verification period to obtain the line performance time series data.

[0039] Based on the time series data of line performance, a performance prediction curve is obtained; based on the performance prediction curve, bandwidth utilization prediction data and transmission delay prediction data are obtained.

[0040] If the bandwidth utilization prediction data is less than the utilization threshold and the transmission delay prediction data is less than the delay threshold, then the switched line passes the accuracy verification and the line switch is complete.

[0041] Furthermore, methods for verifying bandwidth sufficiency of the switched lines based on core performance parameters include:

[0042] Real-time available bandwidth and actual used bandwidth based on core performance parameters; calculate real-time bandwidth utilization.

[0043] The real-time bandwidth utilization rate is compared with the first-level verification threshold. If the real-time bandwidth utilization rate is less than the first-level verification threshold, the line passes the bandwidth sufficiency verification after the switch.

[0044] Furthermore, methods for verifying signal quality of the switched line based on auxiliary quality indicators include:

[0045] Based on auxiliary quality indicators, the transmission delay of the switched line is compared with the preset delay limit, the jitter of the switched line is compared with the preset jitter tolerance, and the bit error rate of the switched line is compared with the preset bit error threshold.

[0046] If the transmission delay of the switched line is less than the preset delay limit, the jitter of the switched line is less than the preset jitter tolerance, and the bit error rate of the switched line is less than the preset bit error threshold, then the switched line passes the signal quality verification.

[0047] Furthermore, methods for obtaining time series data of line performance include:

[0048] The system collects the real-time available bandwidth and actual used bandwidth of the line at a preset time interval after the handover, and auxiliary quality indicators including transmission delay, jitter and bit error rate to obtain a discrete performance data point set; based on the discrete performance data point set, it performs time stamping to obtain a primary time series data set;

[0049] Based on the initial time series dataset, and combined with the outlier data cleaning subroutine, a stable time series dataset is obtained;

[0050] Based on a stable time series dataset, normalization processing is performed to obtain time series data on line performance.

[0051] Furthermore, methods for obtaining performance prediction curves based on line performance time series data include:

[0052] Based on the time series data of line performance, the temporal characteristics of bandwidth utilization and transmission delay are obtained;

[0053] A weighted time series prediction model is constructed based on time series features and the first-level verification threshold.

[0054] Collect current line performance time series data, input the current line performance time series data into the weighted time series prediction model, obtain bandwidth utilization and transmission delay prediction sequences, and obtain performance prediction curves.

[0055] A telephone line switching device includes: a line absolute bandwidth data processing module, which collects the nominal maximum bandwidth and actual used bandwidth of each line, obtains line reference load data based on the nominal maximum bandwidth and actual used bandwidth, obtains the actual available bandwidth of the line, and adjusts the line reference load data based on the actual available bandwidth to obtain line capacity data.

[0056] The individual bandwidth drop threshold calculation module calculates individual bandwidth drop thresholds based on line capacity data and a dynamic adaptive threshold model. Based on the individual bandwidth drop thresholds, it determines whether a multi-line synchronous bandwidth drop event has occurred.

[0057] The bandwidth drop line filtering module is used to filter out a list of bandwidth drop lines whose bandwidth drop is greater than the individual drop threshold if there are multiple synchronous bandwidth drop events.

[0058] The backup switching line matching module obtains backup switching lines based on the list of lines with sudden bandwidth drops and the absolute bandwidth data of the lines through a multi-constraint matching method. Based on the backup switching lines, the line is switched to obtain the line after the switch.

[0059] The line accuracy verification module after switching collects real-time data of the line after switching and verifies the accuracy of the line after switching. The accuracy verification includes real-time performance verification and performance trend prediction verification based on time series. If the line after switching passes the accuracy verification, the line switching is completed.

[0060] Beneficial effects:

[0061] This invention addresses the problem of sudden bandwidth drop caused by multiple lines simultaneously triggering thermal protection mechanisms under extreme environments such as high temperatures. By setting individual drop thresholds, filtering the list of lines with sudden bandwidth drops, and quickly matching backup switching lines, it achieves rapid response and switching in scenarios where multiple lines experience simultaneous bandwidth drops, avoiding large-scale service interruptions and ensuring communication continuity.

[0062] After line switching, this invention collects the core performance parameters and auxiliary quality indicators of the switched line, performs bandwidth adequacy verification and signal quality verification, and generates performance prediction curves based on line performance time series data, further ensuring the long-term stability of the switched line in terms of bandwidth utilization and transmission delay, thereby comprehensively improving the reliability and robustness of the communication system.

[0063] This invention collects the nominal maximum bandwidth and actual used bandwidth of each line to form the line reference load data, and further dynamically adjusts the line capacity data based on the actual available bandwidth. This accurately distinguishes the actual load difference between high-bandwidth lines and low-bandwidth lines, avoiding the problem of high-bandwidth lines being overloaded and low-bandwidth lines being idle due to the same load rate but different actual bandwidth usage in traditional technologies, and realizing the rational allocation of line resources. Attached Figure Description

[0064] Figure 1 A flowchart of the method of the present invention is shown;

[0065] Figure 2 A flowchart illustrating the method for obtaining a list of lines with drastically reduced bandwidth according to the present invention is shown.

[0066] Figure 3 A flowchart for verifying the accuracy of the present invention is shown;

[0067] Figure 4 The flowchart of the weighted time series prediction model of the present invention is shown;

[0068] Figure 5 A module connection diagram of the device of the present invention is shown. Detailed Implementation

[0069] The following will describe in detail, with reference to the accompanying drawings of the embodiments of the present invention, a telephone line switching device and method according to the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0070] To more clearly and intuitively demonstrate the practical application effects and advantages of the telephone line switching device and method of the present invention, and to verify its feasibility and effectiveness, the present invention will be further described below in conjunction with embodiments. Through specific scenario simulations and data calculations, the method is explained in detail how it plays a role in actual telephone line switching, helping readers to better understand the technical details and practical value of the invention. The present invention will be further described below in conjunction with embodiments;

[0071] Example 1

[0072] See Figure 1 - Figure 2 A method for switching telephone lines, comprising:

[0073] The system collects the nominal maximum bandwidth and actual used bandwidth of each line in real time to establish line benchmark load data, breaking the limitations of existing fixed benchmarks and providing initial data support for subsequent accurate evaluation, thus solving the problem of missing adaptation for absolute bandwidth differences.

[0074] It should be noted that the acquisition of the nominal maximum bandwidth, actual used bandwidth, and actual available bandwidth of each line described in this invention can be achieved through various well-known technical means in the field, including but not limited to the following methods:

[0075] Data is collected based on the SNMP protocol. The system initiates query requests to network devices connected to the line, such as routers, switches, and gateways, via the Simple Network Management Protocol (SMMP) to retrieve relevant object identifiers from their specific management information bases. For example:

[0076] Nominal maximum bandwidth: can be obtained by querying the IF-MIB::ifSpeed ​​or IF-MIB::ifHighSpeed ​​object;

[0077] The actual bandwidth used can be obtained by periodically querying the byte counters IF-MIB::ifHCInOctets and IF-MIB::ifHCOutOctets, and calculating the change in the number of bytes per unit time. The calculation formula is: ((Number of bytes input this time - Number of bytes input last time) + (Number of bytes output this time - Number of bytes output last time)) * 8 / sampling time interval.

[0078] Traffic analysis and collection based on NetFlow / sFlow / IPFIX involves enabling traffic export protocols such as NetFlow, sFlow, or IPFIX on network devices and configuring them to send traffic statistics data to a designated collector. After aggregating and analyzing the data, the collector can obtain the actual bandwidth used for each line.

[0079] Based on device driver or operating system interface data collection, on the device running this invention or the agent program communicating with it, the actual bandwidth used is calculated by calling the network device driver interface of the operating system kernel (such as the / proc / net / dev file or ethtool tool under Linux) or using libraries such as libpcap to perform low-level traffic statistics.

[0080] Those skilled in the art can flexibly select and deploy any one or more of the above-mentioned data acquisition schemes based on the target network environment, equipment support, and management strategies. The nominal maximum bandwidth can also be directly read from the device configuration file or pre-configured in the system by the network administrator.

[0081] It should be noted that the methods for obtaining the line reference load data include:

[0082] For each line whose nominal maximum bandwidth is collected, cross-verify it with the line's factory specifications and operator filing data, and remove data with deviations exceeding 3% to ensure the authenticity of the basic data. This is different from the existing method of directly using collected values ​​to obtain verified lines.

[0083] For the verified line corresponding to the nominal maximum bandwidth, the actual used bandwidth of the verified line is collected for a preset time period, and the hourly average is calculated to obtain the hourly average sequence of the actual used bandwidth of the verified line.

[0084] For each verified line, the actual hourly average of the used bandwidth is subtracted from the verified nominal maximum bandwidth to obtain 24 real-time available bandwidth values, forming an hourly sequence of available bandwidth for that verified line.

[0085] The 24-hour period is divided into peak and off-peak periods. The average actual bandwidth used and the average real-time available bandwidth of each verified line are calculated in the two periods to supplement the time period characteristics. The peak period is from 8:00 to 20:00, and the off-peak period is from 20:00 to 8:00.

[0086] Based on the actual bandwidth used during the preset time period of the verified lines and the nominal maximum bandwidth of each line, the ratio of the total average actual bandwidth used (24-hour average) to the nominal maximum bandwidth of each verified line is calculated to obtain the bandwidth utilization rate, which serves as a key data indicator.

[0087] Data is integrated based on the unique identifier of each line, including the verified nominal maximum bandwidth, the average actual bandwidth used in each segment, the average real-time available bandwidth in each segment, and the bandwidth utilization rate, forming structured baseline load data for the lines, which differs from existing non-segmented recording methods. By cross-verifying the nominal maximum bandwidth and the actual bandwidth used in multiple time periods, preliminary structured data containing time period characteristics and bandwidth utilization rate is formed, accurately distinguishing the actual load differences between high-bandwidth lines and low-bandwidth lines. This avoids the problem in traditional technologies where only the load rate is considered, ignoring the actual bandwidth usage differences, which leads to overloaded high-bandwidth lines and idle low-bandwidth lines, thus laying an accurate data foundation for subsequent load assessment.

[0088] Obtain the actual available bandwidth of the line, and adjust the line baseline load data based on the actual available bandwidth to obtain the line capacity data.

[0089] It should be noted that methods for obtaining line capacity data by adjusting the line baseline load data based on the actual available bandwidth include:

[0090] Collect the actual available bandwidth of each line for a preset number of days, compare it with the average real-time available bandwidth of the corresponding line in the line baseline load data, calculate the hourly deviation value, and filter out abnormal periods where the absolute value of the hourly deviation value exceeds 5% to ensure data timeliness.

[0091] Abnormal periods are categorized into three levels based on the magnitude of deviation: minor (5%–10%), moderate (10%–20%), and severe (over 20%). Differentiated weights are then applied to each level to achieve tiered processing.

[0092] For periods with slight deviations, the preliminary data value and the actual value are updated using a 7:3 weighting; for periods with moderate deviations, the data is updated using a 5:5 weighting; for periods with severe deviations, the data is directly replaced with the actual value, and the real-time available bandwidth sequence is adjusted.

[0093] Based on the adjusted real-time available bandwidth sequence, the average actual available bandwidth during peak and off-peak periods is recalculated and replaced with the segmented average in the initial data.

[0094] The updated average actual available bandwidth (24 hours) is divided by the verified nominal maximum bandwidth to obtain the new bandwidth utilization rate, replacing the old bandwidth utilization rate in the preliminary data. A data update timestamp is added to the determined data for each line, and the updated real-time available bandwidth sequence, segment average, and new bandwidth utilization rate are integrated to form dynamic and traceable line capacity data. Through deviation analysis between the 24-hour actual available bandwidth and the preliminary data, differential weighting adjustments are made according to the deviation level to correct the timeliness error in the preliminary data. At the same time, timestamps are added to achieve data traceability, ensuring that the line capacity data can dynamically reflect the real load status of the line. This avoids the shortcomings of traditional fixed data in adapting to dynamic bandwidth lines such as fourth-generation or fifth-generation mobile communication public networks and enterprise shared virtual lines. It provides real-time and reliable data support for subsequent individual drop threshold calculation and line screening, further optimizing the accuracy of line resource allocation.

[0095] Based on line capacity data, an individual bandwidth drop threshold is obtained to determine whether a multi-line synchronous bandwidth drop event has occurred. If so:

[0096] Based on individual bandwidth drop thresholds and line capacity data, a list of lines with bandwidth drops greater than the individual bandwidth drop thresholds is obtained.

[0097] It should be noted that the methods for obtaining individual sudden drop thresholds based on line capacity data through a dynamic adaptive threshold model include:

[0098] A dynamic fluctuation baseline is established by extracting the historical available bandwidth sequence of each line over the past 30 working days from the line capacity data. The exponentially weighted moving average of the data is calculated using the sliding window statistical method and used as the dynamic fluctuation baseline of the line. This baseline can smooth short-term jitter and reflect the long-term trend of bandwidth changes.

[0099] To determine the lower limit of healthy fluctuations, the standard deviation of the dynamic fluctuation baseline is calculated, and the value obtained by subtracting K times the standard deviation (e.g., K=1.96, corresponding to a 95% statistical confidence level) from the dynamic fluctuation baseline is determined as the lower limit of healthy fluctuations for the line. This lower limit is used to distinguish the boundary between normal fluctuations and abnormal drops in the line.

[0100] To quantify historical extreme degradation characteristics, retrieve the line's performance data over a statistical period, such as 90 days, and calculate its historical maximum bandwidth reduction ratio. The formula is (historical peak available bandwidth - historical valley available bandwidth) / historical peak available bandwidth. This ratio quantifies the most severe performance degradation the line has experienced in history.

[0101] Introduce business strategy weights and assign preset importance level weights to the lines based on the criticality of the business they carry. For example, core control lines are assigned 1.2, ordinary data lines are assigned 1.0, and edge backup lines are assigned 0.8.

[0102] Calculate the individual drop threshold using the following fusion formula: Individual drop threshold = (lower limit of health fluctuation / historical peak available bandwidth) * historical maximum bandwidth drop ratio * preset line importance level weight * (1 + real-time traffic load ratio).

[0103] It should be noted that the real-time traffic load ratio is the ratio between the actual business traffic load in the current statistical period and the upper limit of available carrying capacity.

[0104] Within a preset statistical period, the real-time average bandwidth usage and the rated available bandwidth capacity of nodes or links are obtained. The real-time traffic load percentage is then calculated as follows:

[0105] ;

[0106] In the formula, This represents the real-time traffic load percentage. This represents the average real-time bandwidth usage for services within the statistical period. The rated available bandwidth capacity of a node or link.

[0107] Explanation of the design principles of the dynamic adaptive threshold model:

[0108] The design of the dynamic adaptive threshold model is based on the following: (lower limit of health fluctuation / historical peak) constitutes the basic ratio of individual sudden drop thresholds, which reflects the relative state of the line at the critical point of health and degradation; the historical maximum bandwidth reduction ratio serves as a correction factor to ensure that the individual sudden drop threshold can adapt to the line's own physical characteristics, i.e., the extent to which it was most likely to degrade historically; the business importance weight implements a differentiated protection strategy, using more sensitive thresholds for important lines to achieve early warning and protection; (1 + real-time traffic load ratio) is a dynamic adjustment factor that makes the individual sudden drop threshold slightly tighter during high load periods, thus making it more sensitive to performance degradation when the system is under great pressure.

[0109] Through the above calculations, the individual sudden drop threshold is no longer a static value, but an intelligent judgment benchmark that can adapt to the historical performance of the line, the real-time load situation, and the strategic intent of the business. This ensures that in extreme environments such as high temperature, the system can accurately identify multi-line synchronous bandwidth drop events that truly need attention, rather than being disturbed by ordinary random fluctuations.

[0110] Specifically, methods for obtaining the preset importance level weights of lines include:

[0111] Based on the service type and historical communication importance of the line, an evaluation standard for the importance level of the line is defined, including classification rules for core lines, ordinary lines and edge lines, to ensure that the evaluation standard covers the key role of the line in the communication network.

[0112] Based on the evaluation criteria, historical communication data for each line is collected. The historical communication data includes peak load frequency, fault recovery time and service priority indicators, which provide a quantitative basis for weight allocation.

[0113] Using collected historical communication data, a preset importance level weight is calculated for each line. The data is then converted into weight coefficients through normalization. For example, the preset importance level weight for core lines is assigned 1.2, for ordinary lines it is 1.0, and for edge lines it is 0.8. The preset importance level weight table is as follows:

[0114] Table 1. Preset Importance Level Weight Table for Railway Lines

[0115]

[0116] It should be noted that methods for determining whether a multi-line synchronous bandwidth drop event has occurred include:

[0117] The average real-time available bandwidth of each line over the past 24 hours is extracted from the line capacity data as the benchmark bandwidth. At the same time, the average real-time bandwidth of the current 5 minutes is collected to provide comparative data for the judgment of sudden drops, which is different from the existing method of using only single-point real-time data.

[0118] Calculate the bandwidth drop for each line using the formula (baseline bandwidth - current 5-minute real-time average) / baseline bandwidth to obtain a quantified drop value, connecting the baseline and real-time data from the previous step.

[0119] The individual descent threshold for each line is retrieved from the individual descent threshold system. The bandwidth descent magnitude of each line is compared with the individual descent threshold to initially screen out lines whose bandwidth descent magnitude is greater than the individual descent threshold, thus forming a candidate list.

[0120] Check whether the current 5-minute real-time average value of the preliminary lines in the candidate list is less than the fluctuation threshold value in the line capacity data; if the real-time average value is less than the fluctuation threshold value in the line capacity data, then the bandwidth drop of the preliminary lines is greater than the individual drop threshold.

[0121] It should be noted that the methods for obtaining a list of bandwidth drop lines that exceed the individual drop threshold include:

[0122] From the line capacity data, the average available bandwidth of each line over the past three monitoring periods is extracted as the historical benchmark bandwidth. At the same time, the real-time bandwidth of the line at the current moment is collected to provide an accurate data foundation for subsequent calculations, which is different from the existing method that only relies on nominal bandwidth.

[0123] Based on the extracted historical baseline bandwidth and actual used bandwidth, the bandwidth drop of each line is calculated using the formula (historical baseline bandwidth - actual used bandwidth) / historical baseline bandwidth, thus obtaining quantified drop data and avoiding existing fuzzy judgments.

[0124] Call the preset individual drop threshold, compare the calculated bandwidth drop of each line with the threshold one by one, and filter out the preliminary set of lines whose bandwidth drop value is greater than the individual drop threshold, directly connecting with the quantization results of the previous step;

[0125] For each line in the initial line set, real-time bandwidth data at a preset frequency is collected again, the average bandwidth within the preset frequency is calculated, and based on the average bandwidth, the average bandwidth drop is calculated. The average bandwidth drop is verified to see if it is still greater than the individual drop threshold, thus eliminating false screening caused by temporary fluctuations and supplementing the existing deficiency of no secondary verification.

[0126] Retrieve historical bandwidth fluctuation records from the preliminary qualified lines in the line capacity data. If the bandwidth drop of a certain line is within its historical maximum fluctuation range, it is retained; if it exceeds the historical fluctuation range, it is necessary to check whether there is a data collection error to further improve the accuracy of screening.

[0127] By integrating line information verified through secondary validation and historical fluctuation checks, a final list of lines experiencing bandwidth drops is generated, including line identifiers, bandwidth drop magnitudes, and current average bandwidth. Through a three-level filtering logic—calculating the initial bandwidth drop magnitude using historical baseline bandwidth and real-time bandwidth, verifying the secondary bandwidth drop magnitude using the average bandwidth over a preset time period, and checking historical fluctuation records—false bandwidth drop lines caused by temporary fluctuations or data errors are accurately eliminated. This results in a structured list containing line identifiers, bandwidth drop magnitudes, and current average bandwidth, avoiding the coarseness of traditional filtering methods. This ensures that subsequent backup line matching accurately points to the lines that truly need switching, providing a clear target for rapid response in scenarios where multiple lines experience simultaneous bandwidth drops and reducing invalid switching operations.

[0128] Based on the list of lines with sudden bandwidth drops and line capacity data, a backup switching line is obtained through a multi-constraint matching method. Based on the backup switching line, the line is switched to obtain the switched line. Real-time data of the switched line is collected and the accuracy is verified. If the accuracy verification is passed, the line switching is complete.

[0129] It should be noted that the methods for obtaining backup switching lines through multi-constraint matching include:

[0130] First, extract the identifier, connection node pairs, and historical baseline bandwidth of each line from the list of lines with sudden bandwidth drops. This clarifies the characteristics of the lines that need to be replaced, providing precise target parameters for matching backup lines, which differs from the existing method that only relies on line identifiers.

[0131] By constraining node pairs, that is, based on the extracted connection node pairs, all normal lines that are completely consistent with the node pairs of the suddenly dropped lines are selected from the line capacity data, ensuring that the backup lines can be directly replaced and connecting the node pair information of the previous step.

[0132] By using bandwidth adaptability constraints, which are based on the current bandwidth of normal lines and the historical baseline bandwidth of sag lines, the bandwidth adaptability value between these normal lines and the corresponding sag lines is calculated. The formula is the current bandwidth of the normal line divided by the historical baseline bandwidth of the sag line. Adaptable lines with a bandwidth adaptability value ≥ 90% of the preset adaptability value threshold are retained.

[0133] By constraining load stability, we analyze the load fluctuation data of the adapted lines over the past 72 hours, calculate the load standard deviation from the load fluctuation data, screen adapted lines with a load standard deviation ≤ 5% of the standard deviation threshold, exclude adapted lines with unstable load, supplement the existing defects of not considering load, and obtain lines with stable load.

[0134] The remaining lines are weighted by bandwidth adaptability and load fluctuation data to obtain the backup value. The line with the highest backup value for each line experiencing a sudden drop is selected as the backup switching line. This process generates results that differ from existing single-index selection methods. Through a multi-dimensional screening logic that considers node consistency, bandwidth adaptability, load stability, and backup value weighting, backup lines that can replace lines experiencing sudden drops while ensuring sufficient bandwidth and stable load are matched. This avoids the problem of line overload or insufficient bandwidth after switching caused by traditional matching based solely on bandwidth size. It enables rapid and effective switching of multiple lines in the same step in extreme environments such as high temperatures, avoiding large-scale service interruptions and ensuring communication continuity.

[0135] Explanation of the design principle of the multi-constraint matching method: The multi-constraint matching method of this invention is designed for the specific scenario of extremely scarce available resources under systemic failure; its non-obviousness lies in the selection and combination of constraints:

[0136] Node consistency is a hard prerequisite, ensuring the reachability of services after switching, which is different from general load balancing, which may change the path topology;

[0137] Bandwidth adaptability constraints require that the current capacity of the backup line must match the historical baseline bandwidth of the line that has experienced a sudden drop, rather than the current instantaneous value. The purpose of this design is to ensure that the service experience is not degraded after the switchover, and to avoid a situation where the service is connected but of poor quality.

[0138] Load stability constraints are key to avoiding secondary risks in resource-depleted environments; choosing a line with high current bandwidth but large fluctuations may immediately become a new point of failure after switching.

[0139] Example 2

[0140] See Figure 3 - Figure 4 Based on the backup switching line, the line is switched and the new line is obtained. Real-time data of the new line is collected and the accuracy of the new line is verified. If the new line passes the accuracy verification, the line switching is complete.

[0141] It should be noted that the methods for verifying the accuracy of the line after switching include:

[0142] Based on the completed line switching operation, the enhanced data acquisition system of the switched line is started. The system simultaneously collects real-time data at two levels: core performance parameters and auxiliary quality indicators. The core performance parameters are inherited from the absolute bandwidth data of the line, specifically referring to the real-time available bandwidth and the actual used bandwidth. The auxiliary quality indicators are newly added verification dimensions, including transmission delay, jitter and bit error rate.

[0143] The collected core performance parameters are input into the first verification module to verify the bandwidth adequacy. The module first calculates the real-time bandwidth utilization rate based on the real-time available bandwidth and the actual bandwidth used. Then, it compares this real-time bandwidth utilization rate with a first-level verification threshold. This first-level verification threshold is derived from the determined individual drop threshold and a preset safety factor. It is specifically used to preliminarily determine whether the bandwidth capacity of the line is sufficient after the switch.

[0144] The collected auxiliary quality indicators are input into the parallel second verification module for signal quality verification. This module compares the transmission delay, jitter, and bit error rate with preset delay upper limit, preset jitter tolerance, and preset bit error rate threshold, respectively. These three preset thresholds together constitute the signal quality standard, which is used to evaluate the transmission stability of the line after handover. If the transmission delay of the line after handover is less than the preset delay upper limit, the jitter of the line after handover is less than the preset jitter tolerance, and the bit error rate of the line after handover is less than the preset bit error rate threshold, then the line after handover passes the signal quality verification.

[0145] The outputs of the first and second verification modules are integrated to make the first round of comprehensive decision-making. Only when the line after the handover passes both the bandwidth adequacy verification and the signal quality verification can it proceed to the next stage of verification. Otherwise, the handover will be judged to have failed to achieve the expected performance and the handover rollback process will be triggered.

[0146] After the first round of comprehensive decision-making, a continuous trend analysis based on time series is initiated; that is, within a preset verification cycle, the system continuously records and stores the core performance parameters and auxiliary quality indicators of the line after the switch, forming a continuous time series data of line performance covering the entire verification cycle.

[0147] A trend prediction algorithm is applied to the time series data of line performance to generate a performance prediction curve. This algorithm is used to detect whether there is a potential statistical trend of performance indicators developing in the direction of performance degradation within the verification period. Its output is a prediction conclusion about whether the bandwidth utilization or transmission delay will exceed the acceptable range in the future. Based on the performance prediction curve, bandwidth utilization prediction data and transmission delay prediction data are obtained.

[0148] Based on bandwidth utilization prediction data and transmission delay prediction data from performance prediction curves, the final round of verification decisions is made. If the bandwidth utilization prediction data is less than the utilization threshold and the transmission delay prediction data is less than the delay threshold, the trend prediction algorithm determines that there is no risk of the various performance indicators of the post-switching line deteriorating to an unacceptable level during the prediction period. In this case, the post-switching line finally passes the accuracy verification, and the line switchover is officially confirmed to be complete. If there is a clear risk of deterioration in the prediction, even if the current indicators are normal, it is considered a failure to pass the verification, and the early warning and contingency plan handling process is initiated.

[0149] The utilization threshold is a performance benchmark determined by comprehensively considering service quality requirements, system capacity planning objectives, and historical bandwidth utilization statistics. This threshold is formed by integrating the minimum bandwidth guarantee requirements for the services carried by the line, empirical values ​​from long-term system operation optimization, and the acceptable performance fluctuation limit within a preset verification period. Its core function is to establish a clear qualification criterion for the bandwidth utilization prediction data provided by the performance prediction curve, ensuring that the long-term bandwidth carrying capacity of the line after switching always meets service continuity requirements. It is a key performance boundary indicator for accuracy verification. The delay threshold is a transmission performance boundary value jointly determined by service real-time requirements, network transmission characteristics, and historical delay distribution characteristics. This threshold is formed by comprehensively considering the maximum tolerable delay of key service interaction processes, the inherent transmission delay benchmark of the network topology, and the observable delay fluctuation tolerance within a preset verification period. Its core function is to provide a clear standard for judging the compliance of the transmission delay prediction data output by the performance prediction curve, ensuring that the long-term transmission timeliness of the line after switching always meets the quality requirements of real-time service interaction, constituting the core performance benchmark for evaluating transmission stability in accuracy verification.

[0150] The two-level verification logic, which combines real-time performance verification with long-term trend prediction based on performance prediction curves, not only ensures that the current performance of the line after the switch meets the standards, but also predicts its long-term stability. This avoids the problem of short-term compliance but long-term degradation caused by traditional methods that only verify the current state. It comprehensively guarantees the long-term stability of the line in terms of bandwidth utilization and transmission delay after the switch, further improving the reliability and robustness of the communication system and ensuring uninterrupted communication.

[0151] Specifically, methods for obtaining the first-level verification threshold include:

[0152] Obtain the determined individual droop threshold, which is a bandwidth droop threshold calculated based on the absolute bandwidth data of the line, and is used to define the starting point of line performance degradation.

[0153] Based on system reliability requirements and historical data fluctuation analysis, a preset safety factor is selected. This preset safety factor is a multiplier less than 1, which is intended to provide a buffer for the threshold to cope with the uncertainty of real-time data.

[0154] The individual drop threshold is a specific value, such as the percentage decrease in bandwidth utilization, like 20%. It is calculated based on line capacity data to determine the bandwidth drop threshold. The preset safety factor is a preset multiplier, such as 0.8, used to adjust the individual drop threshold to increase the conservatism of the verification. The reason for choosing the individual drop threshold is that it directly reflects the historical or theoretical boundary of line performance degradation, ensuring consistency between the verification benchmark and the original fault judgment. The reason for choosing the preset safety factor is that by introducing redundant buffers, it compensates for uncertainties in real-time data acquisition, preventing misjudgments due to short-term fluctuations, thereby improving the reliability of the verification. The reason for jointly deriving the individual drop threshold and the preset safety factor is that by combining the performance critical point and the conservative adjustment factor, a new, more stringent and adaptable threshold can be generated, ensuring that the bandwidth utilization of the line after switching not only avoids a sudden drop but also remains within a safety margin, enhancing the overall system stability and robustness.

[0155] The individual drop threshold and the preset safety factor are input into the multiplication unit to calculate the candidate value of the first-level verification threshold. Finally, the verified and adjusted first-level verification threshold is output as a system parameter for subsequent bandwidth adequacy verification to ensure its rigor and applicability.

[0156] Specifically, the methods for setting the preset safety factor include:

[0157] The system's overall reliability level requirements, historical data fluctuation characteristics analysis results, and business continuity assurance needs are comprehensively determined. The setting process first assesses the system's tolerance for communication interruption risks, configuring smaller safety coefficient values ​​for critical business scenarios with lower tolerance. Then, it is fine-tuned by combining the long-term statistical fluctuation amplitude and frequency of sudden anomalies in line capacity data. Finally, a conservative scaling factor is formed to dynamically adjust the individual sudden drop threshold. This conservative scaling factor improves the reliability of switching decisions by strengthening the rigor of the verification process.

[0158] Specifically, methods for obtaining time series data of line performance include:

[0159] The high-frequency sampling engine is activated to synchronously collect real-time available bandwidth, actual used bandwidth, transmission delay and bit error rate data of the line after switching at preset time intervals, forming a set of discrete performance data points.

[0160] The discrete performance data points of the discrete performance data point set are input into the timestamp alignment module, and a precise absolute time mark is attached to each data point to generate a primary time series data set with a unified time base.

[0161] The time series data of the primary time series dataset is input into the outlier cleaning subroutine. This subroutine uses the sliding window statistical analysis method to identify and remove data outliers caused by transient network congestion, thereby cleaning the outlier data and generating a cleaned and stable time series dataset.

[0162] Stable time series data is fed into a multi-source data fusion processor, which normalizes indicators of different dimensions such as bandwidth, latency, and bit error rate, and finally outputs line performance time series data with complete structure and uniform dimensions.

[0163] Specifically, methods for obtaining performance prediction curves based on line performance time series data include:

[0164] From the time series data of line performance, the temporal features of bandwidth utilization and transmission delay are extracted. The temporal features include trend, periodic and random terms, which provide accurate support for the subsequent modeling of weighted time series prediction models.

[0165] Based on these temporal characteristics, a weighted temporal prediction model is constructed that integrates individual sudden drop thresholds and preset safety coefficients. The individual sudden drop thresholds are set as model constraints and the preset safety coefficients are used as feature weight adjustment factors.

[0166] The model was trained using historical stable period data from the time series data of line performance, and the parameters were iteratively adjusted until the error between the historical predicted value and the actual value was less than the preset accuracy threshold.

[0167] Select data from the fluctuating periods in the time series data of line performance and input them into the trained weighted time series prediction model to verify the prediction stability under the fluctuating scenario. If the deviation is consistently less than the error threshold, the model is confirmed to be effective.

[0168] The current line performance time series data is input into the weighted time series prediction model to obtain the bandwidth utilization and transmission delay prediction sequences, resulting in a performance prediction curve. Based on the time series characteristics of the line performance time series data, the weighted time series prediction model generates the performance prediction curve, enabling a forward-looking judgment on bandwidth utilization and transmission delay. This avoids the shortcomings of traditional methods that only focus on current performance while ignoring long-term degradation risks, providing a scientific basis for assessing the long-term stability of the line and proactively avoiding service interruptions caused by future performance failures.

[0169] Methods for obtaining bandwidth utilization prediction data and transmission delay prediction data based on performance prediction curves include:

[0170] From the performance prediction curve, the bandwidth utilization prediction trend segment and the transmission delay prediction trend segment are extracted, and the time nodes corresponding to the two trends are recorded synchronously to clarify the original trend source of the two types of prediction data.

[0171] The system calls the established individual drop threshold and preset safety factor to constrain and verify the extracted bandwidth utilization and transmission delay prediction trend segments, and removes trend segments that exceed the safety range formed by the individual drop threshold and preset safety factor.

[0172] According to the preset verification cycle time interval, the bandwidth utilization and transmission delay prediction trend segments after verification are aligned with the time granularity, and the continuous trend is split into discrete prediction points with equal time intervals, thereby ensuring the uniformity of the time dimension.

[0173] Based on the preset accuracy threshold during the training of the weighted time series prediction model, the confidence interval of each discrete prediction point is calculated. The discrete points of bandwidth utilization and transmission delay correspond to their respective confidence intervals, thereby realizing reliability quantification.

[0174] Filter discrete prediction points whose confidence intervals fall entirely within a preset accuracy threshold, and remove prediction points whose interval boundaries exceed the threshold, thereby retaining valid prediction points. The confidence intervals and discrete points are used as the basis to ensure data validity.

[0175] According to the time node order, the effective bandwidth utilization prediction points are integrated into bandwidth utilization prediction data, and the effective transmission delay prediction points are integrated into transmission delay prediction data, thus forming structured time series data.

[0176] Specifically, weighted time series prediction models include:

[0177] This weighted time-series prediction model takes line performance time-series data as input, extracts the time-series features of bandwidth utilization and transmission delay, and constructs a feature weight adjustment strategy by combining the first-level verification threshold and the preset security coefficient.

[0178] Its core mechanism lies in embedding the individual sudden drop threshold as a model constraint into the learning process, so that the prediction curve not only reflects the historical change pattern, but is also constrained by the system safety boundary.

[0179] The model, trained on historical stable data and validated in fluctuating scenarios, can output bandwidth utilization prediction data and transmission delay prediction data for specific future periods, thereby enabling quantitative judgment of line performance trends and providing decision-making basis based on time-series evolution patterns for accuracy verification.

[0180] The post-switching line accuracy verification module, when initiating post-switching line accuracy verification, collects the real-time available bandwidth, actual used bandwidth, transmission delay, jitter, and bit error rate of the post-switching line. Based on the real-time available bandwidth and actual used bandwidth, it obtains the real-time bandwidth utilization rate of the post-switching line; based on the real-time bandwidth utilization rate and the first-level verification threshold, it obtains the bandwidth sufficiency verification result; based on the transmission delay and preset delay limit, jitter and preset jitter tolerance, and bit error rate and preset bit error threshold of the post-switching line, it obtains the signal quality verification result; when both the bandwidth sufficiency verification result and the signal quality verification result are passed, it collects the post-switching line data. Based on the real-time available bandwidth, actual used bandwidth, transmission delay, and bit error rate of the line within a preset verification period after handover, time-series data of line performance is obtained. Based on this time-series data, time-series characteristics of bandwidth utilization and transmission delay are obtained. Based on these time-series characteristics and the first-level verification threshold, combined with a time-series prediction algorithm, a performance prediction curve is obtained. Based on this performance prediction curve, bandwidth utilization prediction data and transmission delay prediction data are obtained. Based on the bandwidth utilization prediction data and utilization threshold, and the transmission delay prediction data and delay threshold, accuracy verification results are obtained.

[0181] Among them, the time-series prediction algorithm is a machine learning method for long-term stability analysis based on line performance time-series data. This algorithm receives real-time available bandwidth, actual used bandwidth, transmission delay, and bit error rate data from the switched line. First, it performs anomaly cleaning and normalization processing. Then, it employs a dynamic weight allocation strategy to extract multi-cycle time-series features of bandwidth utilization and transmission delay, with feature weights dynamically adjusted based on individual drop thresholds and preset safety factors. By constructing a prediction model that integrates time-dimensional dependencies, the algorithm generates performance prediction curves that reflect future performance trends. Finally, it outputs bandwidth utilization prediction data and transmission delay prediction data, providing quantitative evidence for the long-term stability assessment of the switched line and supporting the final decision-making for accuracy verification.

[0182] Example 3

[0183] See Figure 5 A telephone line switching device, comprising:

[0184] The absolute bandwidth data processing module collects the nominal maximum bandwidth and actual used bandwidth of each line. Based on the nominal maximum bandwidth and actual used bandwidth, it obtains the line reference load data, acquires the actual available bandwidth of the line, and adjusts the line reference load data based on the actual available bandwidth to obtain the line capacity data.

[0185] The individual bandwidth drop threshold calculation module calculates individual bandwidth drop thresholds based on line capacity data and a dynamic adaptive threshold model. Based on the individual bandwidth drop thresholds, it determines whether a multi-line synchronous bandwidth drop event has occurred.

[0186] The bandwidth drop line filtering module is used to filter out a list of bandwidth drop lines whose bandwidth drop is greater than the individual drop threshold if there are multiple synchronous bandwidth drop events.

[0187] The backup switching line matching module obtains backup switching lines based on the list of lines with sudden bandwidth drops and the absolute bandwidth data of the lines through a multi-constraint matching method. Based on the backup switching lines, the line is switched to obtain the line after the switch.

[0188] Specifically:

[0189] The absolute bandwidth data processing module collects the nominal maximum bandwidth and actual used bandwidth of each line when it starts processing absolute bandwidth data. Based on the nominal maximum bandwidth and actual used bandwidth of each line, it obtains the line reference load data. It also collects the actual available bandwidth of each line and obtains the line capacity data based on the actual available bandwidth and the line reference load data.

[0190] The individual sudden drop threshold calculation module, when initiated, collects the real-time available bandwidth sequence from the line capacity data. Based on this sequence, it obtains the normal fluctuation benchmark for the line; based on this benchmark, it obtains the line's fluctuation threshold; it collects the line's historical peak and valley available bandwidth, and based on these, it obtains the line's historical maximum bandwidth reduction ratio; based on the line's fluctuation threshold, historical peak available bandwidth, and historical maximum bandwidth reduction ratio, it obtains the preliminary threshold base; and based on the preliminary threshold base, preset line importance level weights, and line traffic proportions, it obtains the individual sudden drop threshold.

[0191] The bandwidth drop line filtering module, when initiated, collects historical bandwidth data from the line capacity data. Based on this historical bandwidth data, it obtains the historical baseline bandwidth of the line. It then collects the actual used bandwidth of the line and, based on both the historical baseline bandwidth and the actual used bandwidth, obtains the bandwidth drop magnitude. Based on the bandwidth drop magnitude and individual drop thresholds, it obtains a preliminary line set. Next, it collects real-time bandwidth data for each line in the preliminary line set over a preset time period. Based on this real-time bandwidth data, it obtains the average bandwidth of each line in the preliminary line set. Based on the average bandwidth and historical baseline bandwidth, it obtains the average bandwidth drop magnitude of each line in the preliminary line set. Finally, based on the average bandwidth drop magnitude and individual drop thresholds, it obtains a list of lines experiencing bandwidth drops.

[0192] The backup switching line matching module, when initiated, collects the connection nodes of each line in the bandwidth drop line list. Based on the connection nodes and line capacity data of each line in the bandwidth drop line list, it obtains the normal lines with the same node pair; it collects the current bandwidth of the normal lines with the same node pair and the historical baseline bandwidth of the bandwidth drop lines, and obtains the bandwidth adaptation value between the normal lines and the bandwidth drop lines with the same node pair; based on the bandwidth adaptation value and the preset adaptation value threshold, it obtains the adapted lines; it collects the load fluctuation data of the adapted lines, and obtains the load standard deviation of the adapted lines; based on the load standard deviation and the standard deviation threshold, it obtains the load stable lines; based on the bandwidth adaptation value and load fluctuation data of the load stable lines, it obtains the backup value of the load stable lines; and based on the backup value of the load stable lines, it obtains the backup switching lines.

[0193] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the present invention's technical concept, should be covered within the scope of protection of the present invention.

Claims

1. A method for switching telephone lines, characterized in that, Includes the following steps: Collect the nominal maximum bandwidth and actual used bandwidth of each line, and obtain the line baseline load data based on the nominal maximum bandwidth and actual used bandwidth; Obtain the actual available bandwidth of the line, and adjust the line baseline load data based on the actual available bandwidth to obtain the line capacity data; Based on line capacity data, an individual bandwidth drop threshold is obtained through a dynamic adaptive threshold model. Based on the individual bandwidth drop threshold, it is determined whether a multi-line synchronous bandwidth drop event has occurred. If so: Based on the individual bandwidth drop threshold and the absolute bandwidth data of the line, a list of lines with bandwidth drops greater than the individual bandwidth drop threshold is obtained. Based on the list of lines with sudden bandwidth drops and the absolute bandwidth data of the lines, a backup switching line is obtained through a multi-constraint matching method. Based on the backup switching line, the line is switched to obtain the line after the switch. Collect real-time data of the switched line and verify the accuracy of the switched line. The accuracy verification includes real-time performance verification and performance trend prediction verification based on time series. If the switched line passes the accuracy verification, the line switching is complete. Methods for obtaining backup switching lines through multi-constraint matching include: Extract the identifier, connection node pairs, and historical baseline bandwidth for each line from the list of lines experiencing a sudden bandwidth drop. Based on the connection nodes of the bandwidth drop line list, and through node pair consistency constraints, all normal lines that are completely consistent with the node pairs of the drop lines are filtered out from the line capacity data. Based on the current bandwidth of the normal line and the historical baseline bandwidth of the line with sudden drop, the bandwidth adaptation value between the normal line and the corresponding line with sudden drop is calculated; the bandwidth adaptation value is compared with the preset adaptation value threshold to select the suitable line. The load fluctuation data of the adapted lines is collected. Based on the load fluctuation data, the load standard deviation is calculated through load stability constraints. The load standard deviation is compared with the standard deviation threshold to screen out lines with stable load. Based on the bandwidth adaptation value of the load-stable line and the load fluctuation data, the reserve value of the load-stable line is calculated; based on the reserve value of the load-stable line, the backup switching line is obtained.

2. The telephone line switching method according to claim 1, characterized in that, Methods for obtaining a list of bandwidth drop lines that exceed individual drop thresholds include: From the line capacity data, the average available bandwidth of each line over the past three monitoring periods is extracted as the historical baseline bandwidth; based on the historical baseline bandwidth and the actual used bandwidth, the bandwidth drop of each line is calculated. By comparing the bandwidth drop of each line with the individual drop threshold, a preliminary set of lines with bandwidth drops greater than the individual drop threshold is obtained. Based on the initial set of lines, real-time bandwidth data within a preset time period is collected. Based on the real-time bandwidth data, the average bandwidth within the preset time period is calculated. Based on the average bandwidth, the average bandwidth drop magnitude is calculated, and the average bandwidth drop magnitude is compared with the individual drop threshold. If the average bandwidth drop magnitude is greater than the individual drop threshold, the line corresponding to the average bandwidth is retained, resulting in a list of lines with bandwidth drops.

3. The telephone line switching method according to claim 1, characterized in that, Methods for obtaining individual sudden drop thresholds based on line capacity data and through a dynamic adaptive threshold model include: The historical available bandwidth sequence for each line over the past 30 working days is extracted from the line capacity data to construct the real-time available bandwidth sequence. The dynamic fluctuation baseline of the normal fluctuation range of the line is obtained through statistical modeling. Based on the dynamic fluctuation baseline, its standard deviation is calculated, and the value of K times the standard deviation is subtracted from the dynamic fluctuation baseline to determine the lower limit of the healthy fluctuation of the line. The search line calculates its historical maximum bandwidth reduction ratio over the past statistical period, and obtains historical extreme degradation characteristics based on the historical maximum bandwidth reduction ratio; Based on the lower limit of health fluctuations, the individual sudden drop threshold is generated by integrating the historical extreme degradation characteristics of the line, the preset line importance level weight, and the real-time traffic load ratio through weighted calculation.

4. The telephone line switching method according to claim 3, characterized in that, Methods for determining whether a multi-line synchronous bandwidth drop event has occurred include: The average real-time bandwidth of the line is collected. Based on the average real-time available bandwidth of the line capacity data and the average real-time bandwidth of the line, the bandwidth drop of the line is calculated. The bandwidth drop magnitude of the line is compared with the individual drop threshold to filter out preliminary lines whose bandwidth drop magnitude is greater than the individual drop threshold. The real-time average value of the preliminary line is collected and compared with the fluctuation threshold value in the line capacity data. If the real-time average value is less than the fluctuation threshold value in the line capacity data, the bandwidth drop of the preliminary line is greater than the individual drop threshold.

5. A telephone line switching method according to claim 1, characterized in that, Methods for collecting real-time data from the switched lines and verifying its accuracy include: Collect the core performance parameters and auxiliary quality indicators of the line after the switchover; Among them, the core performance parameters include real-time available bandwidth and actual used bandwidth, and the auxiliary quality indicators include transmission delay, jitter and bit error rate; Based on core performance parameters, the bandwidth adequacy of the switched lines is verified; based on auxiliary quality indicators, the signal quality of the switched lines is verified. If the switched line passes the bandwidth adequacy verification and the switched line passes the signal quality verification, the core performance parameters and auxiliary quality indicators of the switched line are collected during the preset verification period to obtain the line performance time series data. Based on the time series data of line performance, a performance prediction curve is obtained; based on the performance prediction curve, bandwidth utilization prediction data and transmission delay prediction data are obtained. If the bandwidth utilization prediction data is less than the utilization threshold and the transmission delay prediction data is less than the delay threshold, then the switched line passes the accuracy verification and the line switch is complete.

6. A telephone line switching method according to claim 5, characterized in that, Methods for verifying bandwidth adequacy of lines after handover based on core performance parameters include: Real-time available bandwidth and actual used bandwidth based on core performance parameters; calculate real-time bandwidth utilization. The real-time bandwidth utilization rate is compared with the first-level verification threshold. If the real-time bandwidth utilization rate is less than the first-level verification threshold, the line passes the bandwidth sufficiency verification after the switch.

7. A telephone line switching method according to claim 5, characterized in that, Methods for verifying signal quality of lines after handover based on auxiliary quality indicators include: Based on auxiliary quality indicators, the transmission delay of the switched line is compared with the preset delay limit, the jitter of the switched line is compared with the preset jitter tolerance, and the bit error rate of the switched line is compared with the preset bit error threshold. If the transmission delay of the switched line is less than the preset delay limit, the jitter of the switched line is less than the preset jitter tolerance, and the bit error rate of the switched line is less than the preset bit error threshold, then the switched line passes the signal quality verification.

8. A telephone line switching method according to claim 5, characterized in that, Methods for obtaining time series data of line performance include: The system collects the real-time available bandwidth and actual used bandwidth of the line at a preset time interval after the handover, and auxiliary quality indicators including transmission delay, jitter and bit error rate to obtain a discrete performance data point set; based on the discrete performance data point set, it performs time stamping to obtain a primary time series data set; Based on the initial time series dataset, and combined with the outlier data cleaning subroutine, a stable time series dataset is obtained; Based on a stable time series dataset, normalization processing is performed to obtain time series data of line performance.

9. A telephone line switching method according to claim 6, characterized in that, Methods for obtaining performance prediction curves based on line performance time series data include: Based on the time series data of line performance, the temporal characteristics of bandwidth utilization and transmission delay are obtained; A weighted time series prediction model is constructed based on time series features and the first-level verification threshold. Collect current line performance time series data, input the current line performance time series data into the weighted time series prediction model, obtain bandwidth utilization and transmission delay prediction sequences, and obtain performance prediction curves.

10. A telephone line switching device, employing the telephone line switching method according to any one of claims 1-9, characterized in that, include: The absolute bandwidth data processing module collects the nominal maximum bandwidth and actual used bandwidth of each line. Based on the nominal maximum bandwidth and actual used bandwidth, it obtains the line reference load data, acquires the actual available bandwidth of the line, and adjusts the line reference load data based on the actual available bandwidth to obtain the line capacity data. The individual bandwidth drop threshold calculation module calculates individual bandwidth drop thresholds based on line capacity data and a dynamic adaptive threshold model. Based on the individual bandwidth drop thresholds, it determines whether a multi-line synchronous bandwidth drop event has occurred. The bandwidth drop line filtering module is used to filter out a list of bandwidth drop lines whose bandwidth drop is greater than the individual drop threshold if there are multiple synchronous bandwidth drop events. The backup switching line matching module obtains backup switching lines based on the list of lines with sudden bandwidth drops and the absolute bandwidth data of the lines through a multi-constraint matching method. Based on the backup switching lines, the line is switched to obtain the line after the switch. The line accuracy verification module after switching collects real-time data of the line after switching and verifies the accuracy of the line after switching. The accuracy verification includes real-time performance verification and performance trend prediction verification based on time series. If the line after switching passes the accuracy verification, the line switching is completed.

Images