A method and system for switching a portable WiFi network based on multi-parameter dynamic adjustment

Abstract

The present application relates to the technical field of wireless network optimization, and discloses a kind of based on multi-parameter dynamic adjustment's portable WiFi network switching method and system, a kind of based on multi-parameter dynamic adjustment's portable WiFi network switching method, comprising the following steps: step S1, the incidence factor change rate is calculated;Step S2, the thermal hysteresis compression factor is calculated;Step S3, the historical effective capacity and forward prediction capacity are calculated;Step S4, the service required service rate is calculated;Step S5, the arc segment bias is calculated;Step S6, the final income value is calculated;Step S7, the target network number, switching execution flag and stable maintenance duration are generated.The present application includes the switching decision system with the geometric penetration characteristics of vehicle window and equipment thermal inertia, through the quantitative identification of the asymmetry of serviceable arc segment, completes the comprehensive income evaluation of candidate network and target network selection, reduces meaningless frequent switching action, improves the continuity of cellular backhaul link.

Description

Technical Field

[0001] This invention relates to the field of wireless network optimization technology, and more specifically, to a method and system for switching portable WiFi networks based on dynamic adjustment of multiple parameters. Background Technology

[0002] With the increasing prevalence of high-speed rail travel, portable high-fidelity Wi-Fi devices have become a common means for multiple devices to access the internet within high-speed train carriages. These devices transmit data back through public terrestrial cellular networks while simultaneously providing local wireless access services to multiple mobile devices within the carriage. The switching performance of the cellular network directly determines the user experience of the terminal services. The high-speed mobility and enclosed environment of high-speed rail carriages place higher demands on the adaptability of cellular network switching methods.

[0003] Existing cellular network handover methods are mostly designed for ordinary single-hop terminals, primarily relying on wireless metrics such as instantaneous signal strength, signal-to-noise ratio (SNR), and throughput rate to determine network ranking and handover decisions. Some optimized solutions incorporate auxiliary parameters such as device temperature and movement speed for weighted adjustments to adapt to handover requirements in high-speed mobile scenarios. Existing solutions have been widely adopted in conventional civilian scenarios and can meet the basic handover needs of ordinary terminals.

[0004] The coated windows of high-speed trains generate penetration loss that is strongly correlated with the angle of electromagnetic wave incidence. The temperature rise from continuous equipment operation triggers RF performance degradation. The combined effect of these two factors results in an asymmetric characteristic of the serviceable range of candidate networks along the train's trajectory. Existing methods cannot identify this characteristic and rely solely on instantaneous indicators for switching decisions. This easily leads to switching to networks with insufficient remaining service time, causing frequent handovers and service interruptions, and is unsuitable for the operational requirements of portable high-fidelity Wi-Fi network devices in high-speed rail scenarios. Summary of the Invention

[0005] This invention provides a portable WiFi network switching method and system based on dynamic adjustment of multiple parameters, solving the technical problems mentioned in the background.

[0006] This invention provides a portable WiFi network switching method based on dynamic adjustment of multiple parameters, comprising the following steps: Step S1: Obtain the incident angle of the candidate network, calculate the oblique incident factor, and calculate the rate of change of the incident factor in combination with the time interval; Step S2: Collect the equipment temperature and calculate the equipment temperature rise rate by combining the time interval; merge the equipment temperature and the equipment temperature rise rate to calculate the equivalent hysteresis temperature; and calculate the thermal hysteresis compressibility factor accordingly. Step S3: Obtain the signal-to-noise ratio and network bandwidth of the candidate network, and calculate the historical effective capacity and forward predicted capacity by combining the oblique incidence factor, the rate of change of the incidence factor and the thermal hysteresis compression factor. Step S4: Extract the real-time service arrival rate and the backlog queue size, and calculate the service rate required for the service using the observation window duration; Step S5: Calculate the historical serviceable capacity and forward serviceable capacity using the historical effective capacity, forward predicted capacity and service rate required by the business, and calculate the arc offset accordingly. Step S6: Calculate the handover interruption cost using the real-time service arrival rate and handover interruption time, and combine the forward serviceable quantity, arc offset, and handover interruption cost to calculate the final revenue value. Step S7: Select the candidate network with the largest final benefit value to generate the target network number, combine it with the current service network to generate the switching execution flag, and divide the forward serviceable quantity by the service rate required by the business to obtain the stable maintenance time.

[0007] This invention provides a portable WiFi network switching system based on multi-parameter dynamic adjustment, comprising: The first module obtains the incident angle of the candidate network, calculates the oblique incident factor, and calculates the rate of change of the incident factor in combination with the time interval. The second module collects equipment temperature and calculates the equipment temperature rise rate by combining the time interval, integrates the equipment temperature and the equipment temperature rise rate to calculate the equivalent hysteresis temperature, and calculates the thermal hysteresis compressibility factor accordingly. The third module obtains the signal-to-noise ratio and bandwidth of the candidate network, and calculates the historical effective capacity and forward predicted capacity by combining the oblique incidence factor, the rate of change of the incidence factor and the thermal hysteresis compression factor. The fourth module extracts the real-time service arrival rate and the backlog queue size, and calculates the service rate required for the service using the observation window duration. The fifth module uses historical effective capacity, forward predicted capacity, and service rate required by the business to calculate historical serviceable capacity and forward serviceable capacity, and calculates arc offset accordingly. The sixth module calculates the handover interruption cost using real-time service arrival rate and handover interruption time, and combines the forward serviceability, arc offset, and handover interruption cost to calculate the final revenue value. The seventh module selects the candidate network with the highest final benefit value to generate the target network number, combines it with the current service network to generate a switching execution flag, and divides the forward serviceable quantity by the service rate required by the business to obtain the stable maintenance duration.

[0008] The beneficial effects of this invention are as follows: This invention addresses the working characteristics of portable high-fidelity WiFi network devices in high-speed rail scenarios by incorporating the geometric penetration characteristics of train windows and the thermal inertia of the devices into the handover decision-making system. Through quantitative identification of the asymmetry before and after the serviceable arc segment, it completes the comprehensive benefit evaluation of candidate networks and the selection of target networks. This invention can reduce meaningless frequent handover actions, improve the continuity of cellular backhaul links, and adapt to the network change characteristics in high-speed rail scenarios. The computational complexity of this invention is low, and it can be directly deployed in the embedded systems of existing portable high-fidelity WiFi network devices without additional hardware costs, demonstrating strong engineering feasibility. Attached Figure Description

[0009] Figure 1 This is a flowchart of a portable WiFi network switching method based on multi-parameter dynamic adjustment according to the present invention. Detailed Implementation

[0010] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0011] It should be noted that, unless otherwise defined, the technical or scientific terms used in one or more embodiments of the present invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in one or more embodiments of the present invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" indicate that the element or object preceding the term encompasses the elements or objects listed following the term and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0012] like Figure 1 As shown, a method for switching portable WiFi networks based on dynamic adjustment of multiple parameters includes the following steps: Step S1: Obtain the incident angle of the candidate network, calculate the oblique incident factor, and calculate the rate of change of the incident factor in combination with the time interval; Step S2: Collect the equipment temperature and calculate the equipment temperature rise rate by combining the time interval; merge the equipment temperature and the equipment temperature rise rate to calculate the equivalent hysteresis temperature; and calculate the thermal hysteresis compressibility factor accordingly. Step S3: Obtain the signal-to-noise ratio and network bandwidth of the candidate network, and calculate the historical effective capacity and forward predicted capacity by combining the oblique incidence factor, the rate of change of the incidence factor and the thermal hysteresis compression factor. Step S4: Extract the real-time service arrival rate and the backlog queue size, and calculate the service rate required for the service using the observation window duration; Step S5: Calculate the historical serviceable capacity and forward serviceable capacity using the historical effective capacity, forward predicted capacity and service rate required by the business, and calculate the arc offset accordingly. Step S6: Calculate the handover interruption cost using the real-time service arrival rate and handover interruption time, and combine the forward serviceable quantity, arc offset, and handover interruption cost to calculate the final revenue value. Step S7: Select the candidate network with the largest final benefit value to generate the target network number, combine it with the current service network to generate the switching execution flag, and divide the forward serviceable quantity by the service rate required by the business to obtain the stable maintenance time.

[0013] In one embodiment of the present invention, obtaining the incident angle of the candidate network to calculate the oblique incident factor, and combining the time interval to calculate the rate of change of the incident factor, specifically includes: The formula for calculating the oblique incidence factor based on the incident angle of the candidate network is as follows: in, The oblique incidence factor, For the candidate network incident angle, For the current moment, Assign candidate network numbers; The formula for calculating the rate of change of the incident factor, based on the oblique incident factor, historical oblique incident factor, and time interval, is as follows: in, The rate of change of the incident factor. For historical oblique incidence factor, For time intervals.

[0014] It should be noted that the oblique incidence factor is a quantitative indicator used to characterize the geometric penetration conditions of train windows, ranging from 0 to 1. A higher value indicates that the incoming wave is closer to grazing the window surface, resulting in poorer penetration conditions. The candidate network incidence angle is the angle between the direction of the incoming wave from the candidate network and the normal to the window. The candidate network number is a unique identifier assigned to each accessible cellular network, i.e., a consecutive positive integer starting from 1. The incidence factor change rate is a quantitative indicator used to characterize the trend of the window's oblique incidence factor over time, ranging from -0.5 to 0.5 per second. A positive value indicates a continuous deterioration in geometric penetration conditions, while a negative value indicates a continuous improvement. The historical oblique incidence factor is the oblique incidence factor value calculated at the previous sampling time. The preferred time interval is 100 milliseconds.

[0015] Specifically, the oblique incidence factor transforms the relationship between the electromagnetic wave incident angle and the penetration loss of the vehicle window into a dimensionless value that can be directly used in calculations. Existing wireless communication solutions typically only use large-scale fading models to describe the vehicle body penetration loss, without separately isolating the influence of the incident angle. This invention defines the oblique incidence factor by the absolute value of the sine of the incident angle, quantifying the impact of the incoming wave direction on the penetration conditions. When the incoming wave direction is perpendicular to the vehicle window, the incident angle is 0 degrees, the oblique incidence factor is 0, and the penetration loss is minimal. When the incoming wave direction is parallel to the vehicle window, the incident angle is 90 degrees, the oblique incidence factor is 1, and the penetration loss is maximal.

[0016] It should be noted that the acquisition of candidate network incident angles is achieved through the hardware natively integrated into the portable WiFi device, mainly using two feasible technical methods. The first method is based on the device's native multi-antenna array for wave direction estimation. The portable WiFi cellular communication module natively integrates at least four receiving antennas. The candidate network signals received by each antenna are acquired through this antenna array. The arrival angle of the wave is estimated using the multi-signal classification MUSIC algorithm or the rotation-invariant subspace ESPRIT algorithm. Combined with the pre-calibrated normal direction of the vehicle window during device installation, the incident angle of the candidate network is calculated. This method can acquire the incident angle of each candidate network in real time without relying on external location information, making it suitable for tunnel scenarios without satellite positioning signals. The second method is a geometric calculation method based on the device's native GNSS satellite positioning module. This method uses the satellite positioning module integrated into the portable WiFi device to obtain the train's real-time location and direction of travel. Combined with the pre-stored base station location database and the pre-calibrated window orientation, the angle between the candidate network's incoming wave direction and the window's normal is calculated using spatial geometric relationships. This method has low computational complexity, requires no additional device hardware, and is suitable for portable WiFi devices with a limited number of antenna arrays. Both methods are implemented through the device's native integrated hardware, requiring no additional external equipment, and will not be elaborated upon here.

[0017] It should be noted that the time interval ranges from 50 milliseconds to 500 milliseconds. Firstly, high-speed trains can travel up to 350 kilometers per hour. Within a 100-millisecond interval, the train travels approximately 9.7 meters, ensuring the sampling accuracy of geometric relationships and wireless status, preventing the omission of state changes due to excessively long sampling intervals. Secondly, shorter time intervals result in more decision calculations per unit time, placing greater computational pressure on the portable WiFi embedded device. A 100-millisecond interval keeps the device's computational load within an acceptable range while maintaining sampling accuracy. Finally, the handover decision and execution cycle of cellular networks is typically between 100 milliseconds and 1 second. A 100-millisecond time interval matches the network's handover characteristics, ensuring real-time decision-making. When train speeds are low, the time interval can be appropriately increased to 500 milliseconds; when train speeds are high, the time interval can be appropriately reduced to 50 milliseconds to adapt to different operating scenarios.

[0018] In one embodiment of the present invention, the device temperature is collected and time intervals are combined to calculate the device temperature rise rate, the device temperature and the device temperature rise rate are fused to calculate the equivalent hysteresis temperature, and the thermal hysteresis compressibility factor is calculated accordingly, specifically including: Based on the equipment temperature, historical equipment temperature, and time interval, the formula for calculating the equipment temperature rise rate is as follows: in, For the equipment temperature rise rate, For equipment temperature, Historical equipment temperature; Based on the equipment temperature, thermal hysteresis constant, and equipment temperature rise rate, the formula for calculating the equivalent hysteresis temperature is as follows: in, For equivalent hysteresis temperature, It is the thermal hysteresis constant; The formula for calculating the thermal hysteresis compressibility factor is as follows, based on the equivalent hysteresis temperature, reference temperature, and thermal depreciation intensity coefficient: in, It is the thermal hysteresis compressibility factor. This is the heat depreciation intensity coefficient. For reference temperature, To obtain positive values.

[0019] It should be noted that the device temperature rise rate is a quantitative indicator used to characterize how quickly the temperature of the portable WiFi device changes, with a value ranging from -2 to 5 degrees Celsius per second. A positive value indicates that the device temperature is continuously rising, while a negative value indicates that the device temperature is continuously falling. The device temperature is the real-time operating temperature of the core chip and radio frequency module of the portable WiFi device. The historical device temperature is the device temperature value collected at the previous sampling time. The equivalent hysteresis temperature is a near-future thermal stress estimate obtained by combining the current device temperature and the temperature rise trend, with a value ranging from -40 to 125 degrees Celsius, used to predict the risk of thermal derating of the device in the short term. The thermal hysteresis constant is a fixed parameter used to characterize the time scale of the impact of the device's thermal inertia on the near future, with a preferred value of 2 seconds. The thermal hysteresis compression factor is a quantitative indicator used to characterize the degree to which thermal risk compresses the device's effective service capability, with a value ranging from 0 to 1. The closer the value is to 0, the higher the degree of compression of the device's capability.

[0020] The reference temperature is the temperature threshold that triggers thermal derating compression, preferably 85 degrees Celsius. The thermal derating intensity coefficient is a fixed parameter used to characterize the effect of temperature rise on the available throughput compression intensity of the equipment, preferably 0.05 per degree Celsius. A positive value operator is used to extract the part of the value greater than 0. When the input value is greater than 0, the output is the input value itself; when the input value is less than or equal to 0, the output is 0.

[0021] Specifically, the equivalent hysteresis temperature (EMT) is used to predict the thermal derating risk of equipment in the short term by combining the current temperature with the heating trend. This invention adds the product of the temperature rise rate and the thermal hysteresis constant to the current temperature to proactively predict future thermal conditions. This definition incorporates thermal inertia into the handover decision-making system, enabling proactive prediction of thermal risks and avoiding subsequent erroneous handovers. Furthermore, the thermal hysteresis compression factor converts the thermal risk corresponding to the EMT into a proportional compression factor that can be directly multiplied by the channel capacity. This invention uses an exponential function to construct the thermal hysteresis compression factor, achieving smooth and continuous quantification of the thermal derating impact. When the EMT does not exceed the reference temperature, the compression factor is 1, and the equipment capacity is not compressed. After the EMT exceeds the reference temperature, the compression factor decreases exponentially with increasing temperature, accurately fitting the actual characteristics of equipment thermal derating. This factor directly integrates the thermal state into the effective capacity calculation process, allowing handover decisions to respond to the impact of thermal derating in real time.

[0022] It should be noted that the thermal hysteresis constant ranges from 1 to 5 seconds. This means that there is a thermal conduction delay between the main chip, RF module, and casing of the device, resulting in a significant hysteresis effect in temperature changes. Temperature rise tests on mainstream portable WiFi devices on the market show that the thermal inertia time constant is mainly concentrated between 1 and 5 seconds, with a value of 2 seconds being suitable for the thermal characteristics of most devices. Specifically, the calibration methods are divided into laboratory calibration and online self-calibration. The laboratory calibration method involves running the device at a constant full load in a constant temperature environment, recording the temperature change curve over time, and fitting it with a first-order inertial system to obtain the thermal inertia time constant, which serves as the calibration value for the thermal hysteresis constant. The online self-calibration method involves recording temperature and load changes in real time during device operation, and using a recursive least squares algorithm to fit the device's thermal inertia model online, automatically updating the thermal hysteresis constant value to adapt to thermal characteristic shifts caused by device aging and changes in ambient temperature.

[0023] It should be noted that the reference temperature range is 70 to 100 degrees Celsius. This corresponds to the industrial-grade operating temperature limit of 125 degrees Celsius for the main chips and RF front-end chips used in mainstream portable WiFi devices. When the chip junction temperature exceeds 70 degrees Celsius, slight performance derating begins. When the junction temperature exceeds 85 degrees Celsius, significant derating actions such as carrier aggregation shutdown and standard fallback are triggered. When the junction temperature exceeds 100 degrees Celsius, forced shutdown protection is triggered. The preferred value of 85 degrees Celsius perfectly matches the trigger temperature for strong derating, allowing for accurate prediction of the impact of thermal derating on device performance.

[0024] It should be noted that the thermal derating intensity coefficient ranges from 0.02 to 0.1 degrees Celsius. This means that through testing of mainstream portable WiFi devices, when the device temperature exceeds the reference temperature, for every 1 degree Celsius increase in temperature, the device's actual continuous throughput capacity decreases by approximately 5%. A value of 0.05 degrees Celsius accurately reflects this decrease. Specifically, the calibration method is the load temperature rise test. In a constant temperature environment, the device is connected to a fixed cellular network, maintaining a fixed service load continuously. The ambient temperature is gradually increased, and the correlation between device temperature and actual throughput capacity is recorded. The calibration value of the thermal derating intensity coefficient is obtained through exponential function fitting, adapting to the actual derating characteristics of the device. Furthermore, the device temperature is collected through the portable WiFi device's built-in temperature sensor, which will not be elaborated upon here.

[0025] In one embodiment of the present invention, the signal-to-noise ratio and bandwidth of the candidate network are obtained, and the historical effective capacity and forward predicted capacity are calculated by combining the oblique incidence factor, the rate of change of the incidence factor, and the thermal hysteresis compression factor. Specifically, this includes: Based on the oblique incidence factor, the number of forward sampling steps, the time interval, and the rate of change of the incidence factor, the formula for calculating the predicted value of the oblique incidence factor in the forward window is as follows: in, This is the predicted value of the oblique incidence factor for the forward window. This represents the number of forward sampling steps; The formula for calculating the forward prediction capacity is as follows, based on network bandwidth, candidate network signal-to-noise ratio, window oblique incidence penalty coefficient, forward window oblique incidence factor prediction value, and thermal hysteresis compression factor: in, For forward prediction capacity, For network bandwidth, The signal-to-noise ratio (SNR) of the candidate network. The window oblique incidence penalty coefficient; The formula for calculating historical effective capacity is as follows, based on network bandwidth, historical candidate network signal-to-noise ratio, window oblique incidence penalty coefficient, historical oblique incidence factor, and historical thermal hysteresis compression factor: in, Historical effective capacity The signal-to-noise ratio (SNR) of historical candidate networks. For historical oblique incidence factor, This is the historical thermal hysteresis compressibility factor.

[0026] It should be noted that the forward sampling steps are the total number of sampling points included in the forward prediction window, preferably 10. The forward angle increment is the cumulative predicted increment of the oblique incidence factor over time within the forward prediction window, which is the product of the incidence factor change rate, the forward sampling steps, and the time interval. The predicted value of the forward window oblique incidence factor is the linear extrapolation result of the oblique incidence factor at each future sampling time, which is the sum of the current oblique incidence factor and the forward angle increment. The window oblique incidence penalty coefficient is a fixed parameter used to characterize the penetration penalty enhancement rate when the incident angle deteriorates, preferably 2. The forward penalty exponent is an intermediate variable used to calculate the degree of window penetration attenuation within the forward window, which is the negative value of the product of the window oblique incidence penalty coefficient and the predicted value of the forward window oblique incidence factor. The forward attenuation multiplier is an intermediate variable used to characterize the degree of attenuation of the wireless signal by window penetration within the forward window, which is the calculation result with the natural constant as the base and the forward penalty exponent as the exponent. The candidate network signal-to-noise ratio (SNR) is the ratio of the received signal power to the noise plus interference power of the candidate network, measured in decibels (dB). The forward linear SNR is an intermediate variable that converts the decibel SNR to a linear form; it is calculated by dividing the candidate network SNR by 10 (base 10). The forward base capacity is an intermediate variable representing the basic channel capacity that the candidate network can provide without considering thermal derating. The network bandwidth is the total channel bandwidth currently available to the candidate network, measured in Hertz (Hz). The forward predicted capacity is a predicted value of the effective service capability that the candidate network can provide within a future window, considering the combined effects of radio quality, window penetration, and thermal derating.

[0027] It should be noted that the historical sampling steps refer to the total number of sampling points included within the historical observation window, preferably 10. The historical candidate network signal-to-noise ratio (SNR) is the SNR value of the candidate network collected at the historical sampling time, i.e., the SNR data collected in the decision loop at the corresponding historical time. The historical thermal hysteresis compression factor is the thermal hysteresis compression factor value calculated at the historical sampling time. The historical penalty exponent is an intermediate variable used to calculate the window penetration attenuation degree within the historical window, i.e., the negative value of the product of the window oblique incidence penalty coefficient and the historical oblique incidence factor. The historical attenuation multiplier is an intermediate variable used to characterize the degree of wireless signal attenuation due to window penetration within the historical window, i.e., the calculation result with the natural constant as the base and the historical penalty exponent as the exponent. The historical linear SNR is an intermediate variable that converts the historical decibel form of the SNR into a linear form, i.e., the calculation result with the historical candidate network SNR divided by 10 with the base 10. The historical base capacity is an intermediate variable representing the basic channel capacity that the candidate network can provide in the past without considering the impact of historical thermal derating. Historical effective capacity is the actual effective service capacity that a candidate network can provide at a historical sampling point, under the combined effects of wireless quality, window penetration, and thermal derating.

[0028] It should be noted that the window angle incidence penalty coefficient ranges from 1 to 3. This means that when electromagnetic waves pass through the coated windows of a high-speed train, the penetration loss increases exponentially with the sine of the incident angle. When the incident angle increases from 0 degrees to 90 degrees, the increase in penetration loss is approximately 10 to 20 dB. The optimal value of 2 accurately fits this loss variation. Specifically, the calibration methods are divided into two types: line pre-calibration and online calibration. The line pre-calibration method involves conducting actual vehicle penetration loss tests on the target operating high-speed train line, recording the window penetration loss data at different incident angles, and obtaining the calibration value of the window angle incidence penalty coefficient through exponential function fitting. The online calibration method involves the equipment recording the changes in received signal strength at different incident angles of the same base station in real time during operation, and automatically calibrating the value of the window angle incidence penalty coefficient through least squares fitting to adapt to the differences in window penetration characteristics of different lines and different train models.

[0029] It should be noted that the number of sampling points in the observation window ranges from 5 to 20. Firstly, a higher number of sampling points results in a longer observation window, covering more complete network state changes. However, the prediction accuracy of linear extrapolation decreases as the window length increases. A 1-second window with 10 sampling points maintains high prediction accuracy while ensuring complete coverage and a smooth switching decision cycle. Secondly, a higher number of sampling points involves more accumulation calculations, increasing the computational burden on the device. The computational complexity of 10 sampling points is extremely low, making it suitable for the embedded computing capabilities of portable WiFi devices. Finally, with too few sampling points, fluctuations in the value of a single sampling point can significantly impact the calculation results of the available service capacity, easily leading to decision jitter. The accumulation calculation of 10 sampling points achieves a smoothing filtering effect, ensuring decision stability. When train speeds are high and network states change rapidly, the number of sampling points can be appropriately reduced to 5; when train speeds are low and network states are stable, the number of sampling points can be appropriately increased to 20.

[0030] It should be noted that the acquisition of candidate network signal-to-noise ratio (SNR) can be achieved through the cellular baseband chip natively integrated into the portable WiFi device. Specifically, this involves: first, configuring the measurement rules for co-frequency and inter-frequency neighboring cells using the cellular baseband chip, setting the measurement period and bandwidth to ensure all candidate networks are included in the measurement range; second, the baseband chip measures the received power of the reference signal and the power of interference plus noise using the received downlink reference signal, calculating the SNR in decibels, and recording the candidate network number and measurement time for each measurement value; finally, a moving average filter is applied to the SNR obtained from three consecutive measurements to eliminate numerical fluctuations caused by fast fading, and the filtered value is used as the final candidate network SNR for subsequent calculations. Furthermore, network bandwidth acquisition can be achieved through the cellular baseband chip natively integrated into the portable WiFi device and its system message parsing function. For example, the default bandwidth of mainstream 5G frequency bands in China is 100 MHz, and the default bandwidth of mainstream 4G frequency bands is 20 MHz; this acquisition method can be implemented through the device's native functions without additional hardware.

[0031] In one embodiment of the present invention, the real-time service arrival rate and the backlog queue size are extracted, and the required service rate for the service is calculated using the observation window duration, specifically including: The formula for calculating the required service rate for a service, based on real-time service arrival rate, backlog queue size, number of sampling points in the observation window, and time interval, is as follows: in, Service rate required for business needs For real-time service delivery rate, To address the backlog of pending orders, To determine the number of sampling points in the observation window, The duration of the observation window.

[0032] It should be noted that the real-time service arrival rate is the real-time new service data rate converged from multiple local terminals to the portable WiFi device, measured in bits per second. The backlog queue size is the total length of backlogged data waiting to be sent in the portable WiFi device's routing module, measured in bits. The number of sampling points in the observation window is the total number of sampling points within the historical and forward observation windows, preferably 10. The observation window duration is the total duration of the historical and forward observation windows, which is the product of the number of sampling points and the time interval. The average replenishment service rate is the average additional service rate required to clear the current backlog queue within the observation window duration, calculated by dividing the backlog queue size by the observation window duration. The service required rate is the minimum service capability threshold that the current hotspot's external backhaul needs to provide, which is the sum of the real-time service arrival rate and the average replenishment service rate.

[0033] Specifically, the service requirement rate constructed by this invention includes two dimensions: real-time service arrival rate and average replenishment service rate. The real-time service arrival rate corresponds to the new service traffic demand, while the average replenishment service rate corresponds to the clearing demand of existing backlog queues, fully covering the two key factors influencing user experience. This upgrades the switching decision from a comparison of absolute network capacity to a comparison of effective capacity relative to service demand, more closely aligning with real-world user experience.

[0034] It should be noted that real-time service arrival rate collection can be achieved through the native routing and forwarding module integrated into the portable WiFi device. For example, firstly, traffic statistics rules can be configured between the WAN and LAN interfaces of the routing module to collect all uplink traffic from the local terminal to the cellular backhaul link, and downlink traffic from the cellular backhaul link to the local terminal, covering all service data from all terminals. Then, at the same period as the sampling interval, the total service data volume in each period is counted, in bits. Next, the total service data volume in each period is divided by the sampling interval to obtain the real-time service arrival rate in bits per second for that period. Finally, a moving average filter is applied to the real-time service arrival rates of five consecutive periods to eliminate numerical fluctuations caused by sudden service bursts. The filtered value is then used as the final real-time service arrival rate and input into subsequent calculations. This collection method can be implemented through the native functions of the built-in routing system of the portable WiFi device, and is compatible with all mainstream embedded routing operating systems.

[0035] It should be noted that the collection of the backlog queue size can be achieved through the routing and forwarding module and cellular modem natively integrated into the portable WiFi device. First, the transmission queue length of the WAN interface in the routing module is read, and the total length of Internet Protocol (IP) packets waiting to be forwarded to the cellular link is calculated, in bits. Then, the transmission buffer length inside the cellular modem is read using AT commands, and the total length of packets submitted to the modem but not yet transmitted is calculated. Finally, the queue lengths of the two parts are added together to obtain the final backlog queue size. This collection method can fully cover the backlog data across the entire device link, accurately reflecting the existing queuing pressure of services, and can be implemented through the native system interface of existing devices.

[0036] In one embodiment of the present invention, the historical serviceable capacity and the forward-predicted capacity are calculated using the historical effective capacity, the forward-predicted capacity, and the service rate required by the business, and the arc offset is calculated accordingly, specifically including: Based on historical effective capacity, service requirements, number of sampling points in the observation window, and time interval, the formula for calculating historical serviceable capacity is as follows: in, For historical available service volume; Based on the forward prediction capacity, the service rate required by the business, the number of sampling points in the observation window, and the time interval, the formula for calculating the forward serviceable capacity is as follows: in, Forward serviceability; Based on historical serviceability, forward serviceability, and the smallest positive number, the formula for calculating the arc segment offset is as follows: in, This is the arc segment offset. It is a very small positive number.

[0037] It should be noted that the first capacity difference is the difference between the historical effective capacity and the service rate required by the business. The historical single-step surplus is the total effective service that the candidate network can provide exceeding the business demand within a single historical sampling time, which is the product of the positive portion of the first capacity difference and the time interval. The historical serviceable capacity is the total surplus service capacity of the candidate network relative to the current business demand within the historical observation window, which is the sum of the historical single-step surplus corresponding to all historical sampling steps. The second capacity difference is the difference between the forward predicted capacity and the service rate required by the business. The forward single-step surplus is the total effective service that the candidate network can provide exceeding the business demand within a single forward sampling time, which is the product of the positive portion of the second capacity difference and the time interval. The forward serviceable capacity is the total surplus service capacity of the candidate network relative to the current business demand within the forward observation window, which is the sum of the forward single-step surplus corresponding to all forward sampling steps. The bias numerator is the numerator used to calculate the arc bias, which is the result of subtracting the forward serviceable capacity from the historical serviceable capacity. The minimum positive number is a fixed parameter used to avoid the denominator being zero during calculation, and its preferred value is 0.000001. The bias denominator is the denominator term used to calculate the arc segment bias, which is the sum of the historical serviceable quantity, the forward serviceable quantity, and the minimum positive number. The arc segment bias is a quantitative indicator used to characterize the asymmetry of the serviceable arc segments of the candidate network before and after, which is the result of dividing the bias numerator by the bias denominator.

[0038] Specifically, existing handover schemes typically only focus on network capacity metrics at a single moment, failing to determine the network's sustainable service capacity along its trajectory. In contrast, the serviceable capacity defined in this invention is the cumulative sum of capacity surplus at each sampling moment within the observation window. Effective serviceable capacity is only generated when the effective capacity exceeds the service rate required by the business. Historical serviceable capacity reflects the network's recent service capacity, while forward serviceable capacity reflects the network's future service capacity. The comparison between these two is the core basis for determining the asymmetry of serviceable capacity across different serviceable segments.

[0039] Specifically, the arc bias quantifies the asymmetry between the front and back of the serviceable arcs of a candidate network, determining whether the network has entered the latter half of the serviceable arc's collapse zone. Existing handover schemes typically use instantaneous optimal metrics as the basis for handover, failing to identify whether the network is about to experience rapid capacity collapse. The arc bias defined in this invention, however, is calculated by normalizing the difference between historical serviceable capacity and forward serviceable capacity, achieving a quantitative assessment of the network's remaining service life. This upgrade from instantaneous optimality to optimal remaining service life fundamentally solves the problem of frequent handovers that involve immediate termination upon entry.

[0040] Specifically, existing wireless communication solutions typically define the available area of ​​a network based on cell coverage, without considering the impact of train trajectory, window penetration characteristics, and equipment thermal status on the available area. The serviceable arc segment defined in this invention refers to the continuous path interval within the carriage where a candidate network can continuously provide effective service according to the current service load requirements, after considering the train trajectory as a continuous path. That is, it focuses not on the point-like network state at a single moment, but on the continuous interval along the movement path, while also considering the multiple influences of window penetration, thermal derating, and service demands, which will not be elaborated upon here.

[0041] In one embodiment of the present invention, the handover interruption cost is calculated using the real-time service arrival rate and handover interruption time, and the final revenue value is calculated by combining the forward serviceability, arc offset, and handover interruption cost, specifically including: The formula for calculating the handover interruption cost, based on real-time service arrival rate, handover interruption time, candidate network number, and current serving network number, is as follows: in, To compensate for the cost of switching interrupts, This is the current service network number. To switch the interrupt time; The formula for calculating the final revenue value is as follows, based on the forward serviceability, arc offset, and handover interruption cost: in, For the final profit value, This is the function for extracting the maximum value.

[0042] It should be noted that the current serving network number is the unique identification number corresponding to the cellular network currently providing backhaul service to the portable WiFi device, which is the target network number finally determined in the previous decision loop. The handover interruption cost is a quantified value of the service interruption loss incurred during the handover from the current serving network to the target candidate network. The handover interruption cost for the current serving network is 0, and the handover interruption cost for non-current serving networks is the total amount of new services accumulated during the handover interruption time. The handover interruption time is the total interruption duration required to switch from the current serving network to the target candidate network, preferably 100 milliseconds. The bias penalty term is an intermediate variable used to suppress the revenue of candidate networks that have entered the latter half of the collapse zone; it is the maximum value between the arc bias and 0. The bias compensation coefficient is an intermediate variable used to correct the forward serviceable quantity; it is the result of 1 minus the bias penalty term.

[0043] It should be noted that the estimated remaining revenue is an intermediate variable representing the remaining service revenue that the candidate network can provide after deducting the bias penalty; it is the product of the forward serviceable amount and the bias compensation coefficient. The final revenue value is the quantified value of the comprehensive service revenue of the candidate network after deducting the handover interruption cost; it is the calculated result of subtracting the handover interruption cost from the estimated remaining revenue. The maximum value extraction function is an operation function used to extract the maximum value from multiple values. After inputting multiple values, it outputs the maximum value. The target network number is the unique identification number corresponding to the candidate network with the largest final revenue value; it is the network number with the largest final revenue value among all candidate networks. The candidate network set is the total set of all accessible cellular networks; it is all cellular networks that the portable WiFi device can search for and that meet the access requirements. The independent variable maximum value extraction function is an operation function used to extract the independent variable that maximizes the function value. After inputting the final revenue value corresponding to each candidate network, it outputs the candidate network number that maximizes the final revenue value.

[0044] It should be noted that the handover interruption time ranges from 50 milliseconds to 500 milliseconds. Specifically, the 3GPP standard defines intra-frequency handover interruption time as less than 50 milliseconds, inter-frequency handover interruption time as less than 100 milliseconds, and inter-carrier inter-system handover interruption time as typically between 200 and 500 milliseconds. A preferred value of 100 milliseconds is suitable for mainstream intra-carrier inter-frequency handover scenarios. Specifically, the calibration method is a real-network handover test. The portable WiFi device performs multiple handovers between different standards and carriers within the target network, recording the actual interruption duration for each handover. The average of these multiple tests is used as the calibration value for the handover interruption time. For example, if the average interruption time for intra-carrier 5G to 4G handover is 80 milliseconds, the corresponding handover interruption time can be calibrated as 80 milliseconds; if the average interruption time for inter-carrier handover is 300 milliseconds, the corresponding handover interruption time can be calibrated as 300 milliseconds, achieving accurate calibration for different handover scenarios.

[0045] In one embodiment of the present invention, a target network number is generated by selecting the candidate network with the highest final revenue value, a handover execution flag is generated by combining it with the current serving network, and the stable maintenance duration is obtained by dividing the forward serviceable quantity by the service rate required by the business. Specifically, this includes: The formula for calculating the target network ID based on the final revenue value is as follows: in, Number the target network. For the candidate network set, This is the function for extracting the independent variable; The formula for calculating the handover execution flag, based on the target network ID and the current serving network ID, is as follows: in, To switch the execution flag; Based on the forward serviceability of the target network ID and the service rate required by the business, the formula for calculating the stable maintenance duration is as follows: in, To maintain a stable duration, The forward serviceable quantity corresponding to the target network number.

[0046] It should be noted that the handover execution flag is a binary flag used to control whether the handover action is executed. Its value is either 0 or 1. When the target network number is the same as the current serving network number, the value is 0; when the target network number is different from the current serving network number, the value is 1. The stable maintenance duration is the estimated duration for which the target network can stably support the service under the current service load. It is calculated by dividing the forward service capacity of the target network by the service rate required by the service.

[0047] Specifically, the handover execution flag is a custom binary control flag used in this invention to achieve closed-loop execution of handover decisions. When the target network number is the same as the current serving network number, the flag value is 0, indicating that no handover action is performed and the current serving network remains unchanged. When the target network number is different from the current serving network number, the flag value is 1, indicating that the handover process of the cellular module is immediately triggered, and the backhaul link is switched to the target network.

[0048] Specifically, the stable maintenance duration is a custom timescale estimate used in this invention to quantify the sustainable service capability of a target network. The engineering applications of this metric are mainly twofold: first, it provides a basis for decision reassessment, allowing adjustment of the trigger time for the next decision calculation based on the stable maintenance duration, avoiding unnecessary frequent calculations; second, it provides a reference for service scheduling, allowing for advance adjustment of service coding rates or transmission strategies when the stable maintenance duration is short, preventing service interruptions during handover. This metric elevates handover decisions from a single action to a proactive, end-to-end management process.

[0049] It should be noted that the final output of this invention is the integrated handover execution flag, target network number, and stable maintenance duration. This output can be directly input into the device's cellular modem execution module. When the handover execution flag is 1, the modem immediately executes the handover process to the target network; when the handover execution flag is 0, the modem maintains its current network connection state. The stable maintenance duration can be synchronously input into the device's routing management module to adjust traffic scheduling strategies and decision reassessment cycles, ensuring the continuity of service transmission.

[0050] In one embodiment of the present invention, a portable WiFi network switching system based on multi-parameter dynamic adjustment includes: The first module obtains the incident angle of the candidate network, calculates the oblique incident factor, and calculates the rate of change of the incident factor in combination with the time interval. The second module collects equipment temperature and calculates the equipment temperature rise rate by combining the time interval, integrates the equipment temperature and the equipment temperature rise rate to calculate the equivalent hysteresis temperature, and calculates the thermal hysteresis compressibility factor accordingly. The third module obtains the signal-to-noise ratio and bandwidth of the candidate network, and calculates the historical effective capacity and forward predicted capacity by combining the oblique incidence factor, the rate of change of the incidence factor and the thermal hysteresis compression factor. The fourth module extracts the real-time service arrival rate and the backlog queue size, and calculates the service rate required for the service using the observation window duration. The fifth module uses historical effective capacity, forward predicted capacity, and service rate required by the business to calculate historical serviceable capacity and forward serviceable capacity, and calculates arc offset accordingly. The sixth module calculates the handover interruption cost using real-time service arrival rate and handover interruption time, and combines the forward serviceability, arc offset, and handover interruption cost to calculate the final revenue value. The seventh module selects the candidate network with the highest final benefit value to generate the target network number, combines it with the current service network to generate a switching execution flag, and divides the forward serviceable quantity by the service rate required by the business to obtain the stable maintenance duration.

[0051] Specifically, this invention can be directly deployed in a portable WiFi device, with the device's antenna facing the vehicle window, to provide local wireless access services to multiple mobile terminals inside the vehicle. After the device is deployed, the basic parameters are calibrated first, including the thermal hysteresis constant, reference temperature, thermal derating coefficient, window oblique incidence penalty coefficient, and switching interruption time. After calibration, the cyclic execution process of the method can be started.

[0052] The execution cycle of the method is consistent with the preset time interval. Within each execution cycle, the basic data collection is completed first. The device collects the signal-to-noise ratio and network bandwidth of all candidate networks through the natively integrated cellular baseband chip, calculates the incident angle of candidate networks through the native multi-antenna array or satellite positioning module, collects the real-time temperature of the device through the native built-in temperature sensor, and collects the real-time service arrival rate and backlog queue size through the native routing module. After all collected data is smoothed and filtered, it enters the subsequent calculation process.

[0053] Within each execution cycle, the device completes the entire calculation process according to preset steps, ultimately outputting three results: a handover execution flag, the target network number, and the stable maintenance duration. For example, in a certain execution cycle, the device calculates that the final gain value of the current serving network is 15 megabits, the final gain value of another candidate network is 22 megabits, and the final gain values ​​of the remaining candidate networks are all lower than that of the current serving network. The device identifies this candidate network as the target network, sets the handover execution flag to 1, and calculates the stable maintenance duration of the target network to be 1.2 seconds. Based on the output results, the device immediately triggers the handover process from the cellular module to the target network, and simultaneously adjusts the trigger time for the next decision execution to 1 second later, based on the stable maintenance duration.

[0054] For example, in another execution cycle, if the device calculates that the final benefit values ​​of all candidate networks are lower than those of the current serving network, the device sets the target network number to the current serving network number, sets the switching execution flag to 0, and calculates that the stable maintenance time of the current network is 2.5 seconds. The device maintains the current network connection state and, based on the stable maintenance time, adjusts the trigger time for the next decision execution to 2 seconds to reduce unnecessary frequent calculations.

[0055] It should be noted that this invention is adaptable to the unique environmental characteristics of high-speed rail scenarios, reducing unnecessary frequent switching actions, lowering the probability of service interruption, and improving the user experience of multiple terminals within the carriage. Secondly, the computational complexity of this invention is relatively low, allowing direct deployment in existing commercial portable WiFi devices without hardware modifications, thus possessing strong scalability. Finally, the decision-making logic of this invention can be extended to other high-speed mobile scenarios, including in-vehicle wireless access scenarios such as rail transit and long-distance passenger transport, demonstrating a wide range of applicability.

[0056] It should be noted that the interval and threshold sizes are set for ease of comparison. The size of the threshold depends on the amount of sample data and the base number set by those skilled in the art for each set of sample data, as long as it does not affect the proportional relationship between the parameter and the quantized value. Furthermore, the above formulas are all dimensionless calculations, and the formulas are derived from software simulations using a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0057] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A portable WiFi network switching method based on multi-parameter dynamic adjustment, characterized in that, Includes the following steps: Step S1: Obtain the incident angle of the candidate network, calculate the oblique incident factor, and calculate the rate of change of the incident factor in combination with the time interval; Step S2: Collect the equipment temperature and calculate the equipment temperature rise rate by combining the time interval; merge the equipment temperature and the equipment temperature rise rate to calculate the equivalent hysteresis temperature; and calculate the thermal hysteresis compressibility factor accordingly. Step S3: Obtain the signal-to-noise ratio and network bandwidth of the candidate network, and calculate the historical effective capacity and forward predicted capacity by combining the oblique incidence factor, the rate of change of the incidence factor and the thermal hysteresis compression factor. Step S4: Extract the real-time service arrival rate and the backlog queue size, and calculate the service rate required for the service using the observation window duration; Step S5: Calculate the historical serviceable capacity and forward serviceable capacity using the historical effective capacity, forward predicted capacity and service rate required by the business, and calculate the arc offset accordingly. Step S6: Calculate the handover interruption cost using the real-time service arrival rate and handover interruption time, and combine the forward serviceable quantity, arc offset, and handover interruption cost to calculate the final revenue value. Step S7: Select the candidate network with the largest final benefit value to generate the target network number, combine it with the current service network to generate the switching execution flag, and divide the forward serviceable quantity by the service rate required by the business to obtain the stable maintenance time.

2. The portable WiFi network switching method based on multi-parameter dynamic adjustment according to claim 1, characterized in that, Obtain the incident angle of the candidate network, calculate the oblique incidence factor, and combine it with the time interval to calculate the rate of change of the incident factor, including: Obtain the incident angle of the candidate network, calculate the absolute value of the sine of the incident angle of the candidate network, and obtain the oblique incidence factor; Obtain the historical oblique incidence factor corresponding to the historical moment, and divide the difference between the oblique incidence factor and the historical oblique incidence factor by the time interval to obtain the rate of change of the incidence factor.

3. The portable WiFi network switching method based on multi-parameter dynamic adjustment according to claim 1, characterized in that, The equipment temperature is collected and the time interval is used to calculate the equipment temperature rise rate. The equipment temperature and the equipment temperature rise rate are then combined to calculate the equivalent hysteresis temperature, and the thermal hysteresis compressibility factor is calculated based on this, including: Obtain the historical device temperature corresponding to a historical moment, divide the difference between the device temperature and the historical device temperature by the time interval, and obtain the device temperature rise rate. Obtain the thermal hysteresis constant, multiply the thermal hysteresis constant by the equipment temperature rise rate to obtain the first product, add the first product to the equipment temperature to obtain the equivalent hysteresis temperature; Obtain the reference temperature and the heat derating intensity coefficient, calculate the temperature difference between the equivalent hysteresis temperature and the reference temperature, extract the positive part of the temperature difference that is greater than zero, multiply the positive part by the heat derating intensity coefficient and take the negative value to obtain the exponential variable, calculate the exponent of the exponential variable with the natural constant as the base, and obtain the thermal hysteresis compressibility factor.

4. The portable WiFi network switching method based on multi-parameter dynamic adjustment according to claim 1, characterized in that, Obtain the signal-to-noise ratio (SNR) and bandwidth of candidate networks, and calculate the historical effective capacity and forward predicted capacity by combining the oblique incidence factor, the rate of change of the incidence factor, and the thermal hysteresis compression factor, including: Extract the forward sampling steps, multiply the forward sampling steps, time interval, and rate of change of incident factor in sequence to obtain the forward angle increment, and add the oblique incident factor to the forward angle increment to obtain the predicted value of the oblique incident factor of the forward window. Extract the window oblique incidence penalty coefficient, multiply the predicted value of the forward window oblique incidence factor by the window oblique incidence penalty coefficient and take the negative value to obtain the forward penalty exponent, calculate the exponent of the forward penalty exponent with the natural constant as the base to obtain the forward decay multiplier. Divide the candidate network signal-to-noise ratio by 10, calculate the exponent with 10 as the base, obtain the forward linear signal-to-noise ratio, multiply the forward linear signal-to-noise ratio by the forward attenuation multiplier and add one, calculate the logarithm with 2 as the base, obtain the forward basic capacity, multiply the forward basic capacity, network bandwidth and thermal hysteresis compression factor in sequence to obtain the forward predicted capacity. Extract the historical sampling steps, the historical oblique incidence factor corresponding to the historical moment, the historical candidate network signal-to-noise-interference ratio corresponding to the historical moment, and the historical thermal hysteresis compression factor corresponding to the historical moment. Multiply the historical oblique incidence factor by the window oblique incidence penalty coefficient and take the negative value to obtain the historical penalty exponent. Calculate the exponent of the historical penalty exponent with the natural constant as the base to obtain the historical attenuation multiplier. Divide the historical candidate network signal-to-noise ratio by ten and calculate the exponent with the base of ten to obtain the historical linear signal-to-noise ratio. Multiply the historical linear signal-to-noise ratio by the historical attenuation multiplier and add one. Calculate the logarithm with the base of two to obtain the historical base capacity. Multiply the historical base capacity, network bandwidth, and historical thermal hysteresis compression factor in sequence to obtain the historical effective capacity.

5. The portable WiFi network switching method based on multi-parameter dynamic adjustment according to claim 1, characterized in that, Extract real-time service arrival rate and backlog queue size, and calculate the required service rate for the service using the observation window duration, including: Obtain the number of sampling points in the observation window, multiply the number of sampling points in the observation window by the time interval, and obtain the duration of the observation window; Divide the backlog queue size by the observation window duration to obtain the average replenishment rate; The required service rate is obtained by adding the real-time service arrival rate to the average supplementary service rate.

6. The portable WiFi network switching method based on multi-parameter dynamic adjustment according to claim 1, characterized in that, Using historical effective capacity, forward-predicted capacity, and the service rate required by the business, calculate the historical serviceable capacity and forward serviceable capacity, and calculate the arc offset accordingly, including: Subtract the required service rate from the historical effective capacity to obtain the first capacity difference. Extract the positive part of the first capacity difference that is greater than zero. Multiply the positive part by the time interval to obtain the historical single-step surplus. Sum the historical single-step surplus corresponding to all historical sampling steps to obtain the historical serviceable capacity. The second capacity difference is obtained by subtracting the service rate required by the business from the forward predicted capacity. The positive part of the second capacity difference that is greater than zero is extracted. The positive part is multiplied by the time interval to obtain the forward single-step surplus. The forward single-step surplus corresponding to all forward sampling steps is accumulated and summed to obtain the forward serviceable capacity. Use the difference between historical available service volume and forward available service volume as the bias numerator; The bias denominator is obtained by adding the historical serviceability, the forward serviceability, and the smallest positive number. Divide the bias numerator by the bias denominator to obtain the arc segment bias.

7. The portable WiFi network switching method based on multi-parameter dynamic adjustment according to claim 1, characterized in that, The handover interruption cost is calculated using real-time service arrival rate and handover interruption time. Combined with forward serviceability, arc offset, and handover interruption cost, the final revenue value is calculated, including: Determine if the candidate network number is the same as the current serving network number. If the candidate network number is the same as the current serving network number, assign the handover interruption cost to zero. When the candidate network number is different from the current serving network number, the real-time service arrival rate is multiplied by the handover interruption time to obtain the handover interruption cost; Extract the maximum value between the arc offset and zero to obtain the offset penalty term. Subtract the offset penalty term from one to obtain the offset compensation coefficient. Multiply the forward serviceable amount by the offset compensation coefficient to obtain the estimated remaining revenue. Subtract the handover interruption cost from the estimated remaining revenue to obtain the final revenue value.

8. The portable WiFi network switching method based on multi-parameter dynamic adjustment according to claim 1, characterized in that, The candidate network with the highest final benefit value is selected to generate the target network number. This is combined with the current service network to generate a handover execution flag. The forward serviceable capacity is divided by the service rate required by the business to obtain the stable maintenance duration, including: Compare the final revenue values ​​of all candidate networks, and extract the candidate network number corresponding to the maximum final revenue value as the target network number; Compare the target network number with the current service network number. If the target network number and the current service network number are the same, set the switch execution flag to zero; if the target network number and the current service network number are different, set the switch execution flag to one. Extract the forward serviceable quantity corresponding to the target network number, divide the forward serviceable quantity corresponding to the target network number by the service rate required by the business, and obtain the stable maintenance duration.

9. A portable WiFi network switching system based on multi-parameter dynamic adjustment, characterized in that, Performing a portable WiFi network switching method based on multi-parameter dynamic adjustment as described in any one of claims 1 to 8 includes: The first module obtains the incident angle of the candidate network, calculates the oblique incident factor, and calculates the rate of change of the incident factor in combination with the time interval. The second module collects equipment temperature and calculates the equipment temperature rise rate by combining the time interval, integrates the equipment temperature and the equipment temperature rise rate to calculate the equivalent hysteresis temperature, and calculates the thermal hysteresis compressibility factor accordingly. The third module obtains the signal-to-noise ratio and bandwidth of the candidate network, and calculates the historical effective capacity and forward predicted capacity by combining the oblique incidence factor, the rate of change of the incidence factor and the thermal hysteresis compression factor. The fourth module extracts the real-time service arrival rate and the backlog queue size, and calculates the service rate required for the service using the observation window duration. The fifth module uses historical effective capacity, forward predicted capacity, and service rate required by the business to calculate historical serviceable capacity and forward serviceable capacity, and calculates arc offset accordingly. The sixth module calculates the handover interruption cost using real-time service arrival rate and handover interruption time, and combines the forward serviceability, arc offset, and handover interruption cost to calculate the final revenue value. The seventh module selects the candidate network with the highest final benefit value to generate the target network number, combines it with the current service network to generate a switching execution flag, and divides the forward serviceable quantity by the service rate required by the business to obtain the stable maintenance duration.

Images