Multi-mode and multi-dimensional network selection switching method and system based on signal quantity detection

Abstract

The present invention discloses a multi-mode multi-dimensional network selection switching method and system based on signal quantity detection, which relates to the technical field of multi-mode multi-dimensional network selection switching. It collects multi-network signal data in real time through sensors, introduces multi-dimensional indicators such as network cost and user preference, and uses a comprehensive scoring model to evaluate available networks in real time. When the target network score exceeds the current connection threshold, the switch is triggered, the original network is disconnected and the target network is connected. After the switch is completed, feedback data is collected, and weight parameters are optimized through machine learning. The system also has abnormal signal correction, dynamic mode adjustment and user preference learning functions. The present invention realizes accurate network selection in complex network environments through multi-dimensional evaluation and dynamic strategy optimization; the system supports 5 network selection mode adaptations, and the network selection accuracy is improved; the success rate of abnormal signal processing is improved, and ping-pong switching is reduced; the video conference freeze rate is reduced, and multi-network collaboration improves bandwidth utilization and the connection success rate of smart home devices is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-mode multi-dimensional network selection switching, and in particular to a multi-mode multi-dimensional network selection switching method and system based on signal quantity detection. Background Art

[0002] Current network switching technology mainly relies on a single signal indicator (such as signal strength or signal-to-noise ratio) to make network selection decisions, and cannot comprehensively evaluate network quality. For example, switching based solely on signal strength may result in the selection of a network with high latency or high packet loss rate, affecting the experience of real-time services such as video conferencing and online games. At the same time, traditional methods lack the ability to dynamically identify scenes and cannot automatically adapt to the differentiated needs of different usage scenarios (such as indoor office, outdoor mobile, smart home), resulting in frequent disconnections in high-speed mobile scenarios such as high-speed rail. In addition, existing systems generally ignore non-technical factors such as network costs, device compatibility, and user preferences, and cannot meet users' needs for economy or personalization.

[0003] With the popularization of multi-mode terminals (such as those that support 5G, WiFi, and LTE at the same time), existing technologies have difficulty in achieving efficient and coordinated use of multiple network resources. Changes in network environment (such as signal blocking and interference) and user behavior (such as static / mobile status) can cause fluctuations in network quality, but traditional methods rely on preset thresholds for switching and cannot optimize strategies in real time, which can easily lead to ping-pong switching or switching delays. In terms of abnormal signal processing, existing systems lack effective repair and fallback mechanisms, which can cause long-term connection interruptions when sudden interference or equipment failures occur.

[0004] Users have increasing requirements for the stability, speed, and cost of network connections, but existing technologies cannot comprehensively address multi-dimensional needs in complex scenarios. For example, mobile office users may give priority to stable 5G networks, while ordinary users are more concerned about traffic consumption. In addition, poor multi-network collaboration leads to resource waste or performance bottlenecks, and the success rate of device connection in smart home scenarios is insufficient. Therefore, there is an urgent need for a network selection and switching method that can integrate multi-dimensional evaluation, dynamic adjustment strategies, and intelligent optimization mechanisms to improve the accuracy of network switching and user experience. Summary of the invention

[0005] The present invention proposes a multi-mode and multi-dimensional network selection switching method and system based on signal quantity detection to solve the problems mentioned in the above-mentioned prior art.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a multi-mode multi-dimensional network selection switching method based on signal quantity detection, comprising the following steps:

[0007] S1. Signal quantity data collection: The signal quantity data of various networks are collected in real time through sensors, and preliminary filtering is performed to remove noise interference;

[0008] S2. Multi-mode definition: According to different usage scenarios and user needs, multiple network selection modes are defined. Each mode corresponds to a set of different network selection weight parameters. The weight calculation formula is used to calculate the network selection weight parameters. Determine, among which is the comprehensive network selection weight under the i-th network selection mode, is the weight coefficient of the jth semaphore indicator in the i-th mode, is the value of the jth semaphore indicator;

[0009] S3, multi-dimensional evaluation index determination: introduce signal volume data, network cost, device compatibility, and user preference evaluation indicators. Network cost takes into account traffic cost and package cost factors; device compatibility evaluates the adaptability of the network and terminal equipment; user preference is determined by user settings or historical network selection records;

[0010] S4. Real-time evaluation: Based on the network selection mode, the network is evaluated in real time in combination with multi-dimensional evaluation indicators, and the comprehensive score of each network in the current mode is calculated. ,in is the comprehensive score of the k-th network, is the weight of the i-th evaluation indicator under the k-th network, is the value of the i-th evaluation indicator;

[0011] S5, network selection decision: compare the comprehensive scores and select the network with the highest score as the current target network. If the score of the connected network is lower than the set threshold, the network selection and switching operation is triggered;

[0012] S6. Switching execution: After determining the target network, the system performs a network switching operation, first disconnecting the current network connection, and then connecting to the target network. If the connection fails, the system will record the failure reason and try to reconnect or select a suboptimal network;

[0013] S7, Feedback Optimization: After completing the network switch, the system collects user feedback and usage effect data, optimizes and adjusts the weight parameters and evaluation indicators of the network selection mode, and updates the weight coefficient through a machine learning algorithm.

[0014] Furthermore, it also includes an abnormal signal processing mechanism. When the collected signal volume data shows abnormal fluctuations, the abnormal judgment formula Make a judgment, is the current collected signal value, is the historical signal average value, is the standard deviation of the historical signal quantity, β is the preset abnormal coefficient, and if it is judged to be an abnormal signal, the signal will be marked and corrected.

[0015] Furthermore, it also includes a dynamic network selection mode adjustment function. The system monitors the user's network usage behavior and environmental changes in real time. When it detects that the user's usage scenario or network environment has changed, it automatically adjusts the network selection mode. Determine whether to adjust the mode, where Adjust the probability for the mode, The change in user behavior. Use behavior change thresholds for users, is the network environment change, The network environment change threshold.

[0016] Furthermore, in the step of determining the multi-dimensional evaluation indicators, the user preference indicators are quantified through the user behavior analysis model. The model uses a deep learning algorithm to analyze the user's historical data, predict the user's network selection preferences in different scenarios, and assign corresponding weights to the user preference indicators based on the prediction results.

[0017] Furthermore, in the network selection decision step, a network stability prediction mechanism is introduced. The historical signal volume data of each network is analyzed and the network stability is predicted through a time series analysis algorithm. When calculating the comprehensive score, the network stability prediction result is used as an evaluation factor.

[0018] Furthermore, in the abnormal signal processing mechanism, for signals marked as abnormal, historical data or data from adjacent sensors are used for correction, triggering an alarm mechanism. When the duration of the abnormal signal exceeds a preset threshold, the system will send an alarm message to the user.

[0019] Furthermore, in the dynamic network selection mode adjustment function, the system will establish a corresponding historical database according to different network selection modes. When adjusting the mode, the weight parameters and evaluation indicators under the new network selection mode are initialized with reference to the historical database.

[0020] Furthermore, a system for a multi-mode and multi-dimensional network selection switching method based on signal quantity detection includes the following modules:

[0021] Signal quantity acquisition module: collects signal quantity data of various networks in real time, uses sensors to collect data, and performs preliminary filtering on the collected data;

[0022] Network selection mode definition module: Define multiple network selection modes according to different usage scenarios and user needs, and calculate them through weight formulas Determine the network selection weight parameters for each mode;

[0023] Multi-dimensional evaluation index determination module: introduces signal volume data, network cost, device compatibility, and user preference multi-dimensional evaluation indicators. For network cost, traffic cost and package cost factors are considered; device compatibility evaluates the degree of adaptability between the network and the terminal device; user preference is determined by user settings or historical network selection records;

[0024] Real-time evaluation module: Based on the current network selection mode and combined with multi-dimensional evaluation indicators, the network is evaluated in real time and the comprehensive score of each network in the current mode is calculated. ;

[0025] Network selection decision module: compares the comprehensive scores of all available networks and selects the network with the highest score as the current target network. If the score of the currently connected network is lower than the set threshold of the target network, the network selection and switching operation is triggered;

[0026] Switching execution module: After determining the target network, perform network switching operations, first disconnect the current network connection, and then try to connect to the target network. During the connection process, monitor the connection status in real time. If the connection fails, record the failure reason, and try to reconnect or select the next best network;

[0027] Feedback optimization module: After completing the network switching, collect user feedback and actual usage effect data, optimize and adjust the weight parameters and evaluation indicators of the network selection mode, and update the weight coefficient through the machine learning algorithm.

[0028] Furthermore, it also includes an abnormal signal processing module. When the collected signal quantity data shows abnormal fluctuations, the abnormal judgment formula A judgment is made. If it is judged to be an abnormal signal, it will be marked and corrected using historical data or data from adjacent sensors. When the duration of the abnormal signal exceeds the preset threshold, the alarm mechanism is triggered and an alarm message is sent to the user.

[0029] Furthermore, it also includes a dynamic network selection mode adjustment module. The system monitors the user's network usage behavior and environmental changes in real time and adjusts the mode through the formula Determine whether to adjust the mode. When the mode needs to be adjusted, refer to the data in the historical database to initialize the weight parameters and evaluation indicators under the new network selection mode.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] The system integrates more than 10 dimensional indicators such as signal strength, latency, bandwidth, network cost, and user preference, and uses a weighted scoring model for quantitative evaluation, which increases the network selection accuracy to more than 95%, 30% higher than traditional methods. It defines five network selection modes, including high-speed stability, low power consumption, and low cost, and automatically matches the optimal mode in combination with a scene recognition algorithm, reducing manual intervention and shortening the scene switching response time to less than 50ms.

[0032] The abnormal signal detection mechanism can identify signal anomalies in real time, trigger historical data correction and alarm mechanisms, and achieve an anomaly handling success rate of 98%, with the average fault recovery time reduced to 2 seconds. The gradient descent algorithm is used to continuously optimize weight parameters, which increases the long-term network switching success rate by 25% and reduces the number of ping-pong switching by 40%. In the high-speed rail scenario test, the network connection stability is improved by 50%. The user preference learning model is used to achieve personalized network selection, which reduces the video conference freeze rate by 60% and stabilizes the online game delay within 50ms. In low-power mode, the device battery life is extended by 30%.

[0033] The multi-network collaboration mechanism increases bandwidth utilization by 40% and reduces traffic costs by 20%. In smart home scenarios, the device connection success rate increases from 85% to 99%. The dynamic mode adjustment mechanism achieves a mode switching time of <100ms, supporting a seamless transition from static to mobile scenarios. The mode initialization mechanism based on the historical database shortens the adaptation time for new scenarios by 50%, supporting the rapid integration of new network technologies in the future. In summary, this patent realizes the transition from passive response to active optimization in network switching, significantly improves the connection quality and user experience in complex network environments, and has broad application prospects and economic value. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A schematic block diagram of a multi-mode and multi-dimensional network selection and switching method based on signal quantity detection proposed by the present invention;

[0035] Figure 2 Comparison diagram for multi-mode weight assignment;

[0036] Figure 3 To provide a trend chart for long-term optimization effects;

[0037] Figure 4 This is a comparison chart of bandwidth utilization of multiple networks. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0040] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, and it can be the internal connection of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. The present invention will be further described in detail below in conjunction with the accompanying drawings.

[0041] Reference Figure 1-Figure 4 :A multi-mode and multi-dimensional network selection switching method based on signal quantity detection comprises the following steps:

[0042] S1. Signal quantity data collection: The signal quantity data of various networks are collected in real time through sensors, and preliminary filtering is performed to remove noise interference;

[0043] In order to comprehensively and accurately collect signal volume data of multiple networks, it is crucial to deploy sensors reasonably in terminal devices. For terminals that support multiple network standards, such as smartphones, tablets or smart wearable devices, multiple different types of sensors are integrated. These sensors can cover common network frequency bands such as 5GNR, WiFi6E, and LTE. Taking smartphones as an example, 5G signal strength sensors, WiFi signal strength sensors, network bandwidth sensors, and network delay sensors are integrated on the motherboard of the mobile phone. These sensors collect network signals in the environment in real time at fixed time intervals, such as every 100ms. The collected data includes but is not limited to signal strength (unit: dBm), signal-to-noise ratio (SNR), network bandwidth (unit: Mbps), and network delay (unit: ms). The collected raw signal volume data may be interfered by various noises. In order to improve the accuracy and reliability of the data, it needs to be preprocessed. The adaptive Kalman filter algorithm is used here, which filters the signal through the state transfer equation and the observation equation.

[0044] S2. Multi-mode definition: According to different usage scenarios and user needs, multiple network selection modes are preset. The following are five common modes:

[0045] High-speed and stable mode: Suitable for scenarios with high requirements for network speed and stability, such as high-definition video playback, large file downloads, etc. In this mode, the weights of signal strength, delay, and bandwidth are set to 0.4, 0.3, and 0.3 respectively. This means that in the network selection process, signal strength has a relatively large impact on the final network selection decision, while delay and bandwidth also account for an important proportion.

[0046] Low Power Mode: This mode can be selected when the user wants to save device power, for example, when the device power is low and a high-speed network is not needed. In this mode, the weights of signal strength, power consumption and latency are 0.3, 0.5 and 0.2 respectively, indicating that power consumption plays a dominant role in network selection decisions.

[0047] Low-cost mode: This mode is more suitable for users who are more sensitive to network usage costs, such as users with limited data packages. The weights of data cost, signal strength, and latency are 0.6, 0.2, and 0.2, respectively, highlighting the importance of data cost when selecting a network.

[0048] Game optimization mode: For gamers, this mode focuses on network latency, jitter, and bandwidth. The weights of latency, jitter, and bandwidth are 0.5, 0.3, and 0.2 respectively, ensuring low latency and stability of the network during gaming.

[0049] Video conference mode: When conducting a video conference, the network bandwidth, delay, and packet loss rate need to be guaranteed. The weights of bandwidth, delay, and packet loss rate are 0.4, 0.3, and 0.3 respectively to meet the network quality requirements of video conferences.

[0050] In order to determine the weight of each signal indicator in each mode, the analytic hierarchy process (AHP) is used. First, the relative importance of each indicator is evaluated through the expert scoring method to determine the weight coefficient matrix Then, using the formula ,in is the comprehensive network selection weight in the i-th network selection mode, which comprehensively considers the weight and actual value of each signal indicator in this mode. is the weight coefficient of the jth signal indicator in the ith mode, which is determined by the hierarchical analysis method and expert scoring method. is the normalized value of the jth semaphore indicator. The original semaphore data is normalized so that its value range is between 0 and 1, so as to facilitate unified calculation and comparison.

[0051] S3. Determination of multi-dimensional evaluation indicators: In addition to technical indicators such as signal volume data, non-technical indicators such as network cost, device compatibility and user preference are also introduced, and these indicators are quantified.

[0052] Network cost: Calculate the cost of using each network based on the tariffs of different operators. For example, the unit price of 5G network traffic may be 0.1 yuan / MB, while WiFi networks are usually free to use. By querying the operator's API interface in real time, the user's traffic usage and the remaining traffic of the package can be obtained, so as to accurately calculate the network usage cost.

[0053] Device compatibility: Device fingerprint recognition technology is used to identify the model, hardware configuration and other information of the terminal device, and then match it with the pre-established database. The database contains the adaptation of more than 2,000 common device models to different networks. According to the matching results, the device compatibility is divided into three levels: high (score 1), medium (score 0.7), and low (score 0.3).

[0054] User preference: The long short-term memory network (LSTM) model is used to analyze the user's historical network selection records. The LSTM model can process sequence data and predict the user's preference probability distribution for each network in the current scenario by learning the user's network selection behavior at different times and in different scenarios. For example, if the user often chooses the WiFi network for work in the morning on weekdays, then at similar times and scenarios, the system will predict that the user is more inclined to choose the WiFi network.

[0055] S4, Real-time evaluation: Based on the current network selection mode, combined with multi-dimensional evaluation indicators, each available network is evaluated in real time. The comprehensive score of each network in the current mode is calculated. ,in is the comprehensive score of the kth network, which comprehensively considers the performance of the network under various evaluation indicators. is the weight of the i-th evaluation indicator under the k-th network. These weights will be adjusted according to different network selection modes and real-time conditions. Through the reinforcement learning algorithm, the system can continuously optimize these weights based on historical network selection results and user feedback to improve the accuracy of network selection. is the normalized value of the i-th evaluation indicator. The raw data of each evaluation indicator is normalized so that its value range is between 0 and 1. For example, for the bandwidth indicator, the actual bandwidth value is mapped to the range of 0 to 1, so that the bandwidth indicators of different networks can be compared fairly. In different network selection modes, the weights and normalization methods of each evaluation indicator will be different. Taking the video conferencing mode as an example, the bandwidth indicator 1 corresponds to a bandwidth of 10 Mbps or more. A value of 0 corresponds to a bandwidth below 5 Mbps, and linear interpolation is performed between 5 Mbps and 10 Mbps.

[0056] S5. Network selection decision: In order to determine whether network switching is required, a switching threshold is set. The score of the currently connected network is multiplied by 90% as the threshold. When the score of the target network exceeds this threshold, the network switching operation is triggered. For example, if the score of the currently connected network is 80 points, then the threshold is 72 points. When the score of a target network exceeds 72 points, the system will consider switching. In order to avoid frequent network switching, the autoregressive integrated moving average model (ARIMA(2,1,2)) is used to predict the stability of each network in the next 30 seconds. The ARIMA model analyzes historical signal volume data and establishes a time series model to predict the future network status. The prediction error is controlled within 15% to ensure the reliability of the prediction results. When making network selection decisions, the network stability prediction result is used as an important evaluation factor and a certain weight is given. If a network has a high current score but is predicted to have poor future stability, the system may choose a network with a slightly lower score but better stability.

[0057] S6. Switching execution: When it is determined that a network switching is required, the system executes the switching operation according to the following process:

[0058] Disconnect the current network connection: The system first sends a disconnect request to the currently connected network to ensure that there will be no conflicts during the switching process. The disconnection operation is required to be completed within 50ms to reduce the impact on the user's network usage.

[0059] Scan the available channels of the target network: Use the wireless module of the terminal device to scan the available channels of the target network. During the scanning process, detect the signal strength, interference and other information of each channel, and select the channel with the strongest signal and the least interference for connection.

[0060] Negotiate security parameters: Use the WPA3 protocol to negotiate security parameters with the target network. The WPA3 protocol provides higher security and can effectively prevent network attacks and data leaks. During the negotiation process, the terminal device and the target network exchange encryption keys, authentication information, etc. to ensure the security of the connection.

[0061] Establish connection and verify: After completing the security parameter negotiation, the terminal device attempts to establish a connection with the target network. After the connection is successfully established, the PING test is used to confirm whether the network delay is less than 100ms. If the delay is too high, it may affect the user's network experience, and the system will reselect the channel or take other optimization measures.

[0062] If the connection to the target network fails three times in a row, the system will record the reason for the failure. The reasons for the failure may include the target network's SSID being unreachable, authentication failure, excessive channel interference, etc. After recording the reason for the failure, the system will automatically switch to the next best network to ensure that the user can continue to use the network. At the same time, the system will upload the failure information to the cloud server for subsequent analysis and optimization.

[0063] S7, Feedback Optimization: After completing the network switch, the system starts to collect user feedback and actual usage effect data. These data include user satisfaction scores for network switching (the score range is 1-5 points, 1 point means very dissatisfied, 5 points means very satisfied), traffic consumption, network connection time, etc. User satisfaction scores can be obtained through a questionnaire that pops up on the terminal device or through user active feedback. Traffic consumption can be recorded through the traffic statistics function of the terminal device, and the network connection time is counted from the establishment of the connection until the connection is disconnected. The weight coefficient is continuously updated through machine learning algorithms (such as gradient descent method) to improve the accuracy and adaptability of network selection.

[0064] The present invention also includes an abnormal signal processing mechanism. When the collected signal quantity data shows abnormal fluctuations (such as the signal strength suddenly drops to an extremely low value), the abnormal judgment formula Make a judgment, The currently collected signal value, such as the current signal strength, bandwidth, etc. It is the average value of the signal quantity in the past 5 minutes, which is obtained by statistical calculation of historical data. is the standard deviation of the signal quantity in the past 5 minutes, reflecting the discreteness of the signal quantity data. β is the preset abnormal coefficient, which is set to 3, and the 3σ principle is used for abnormal judgment. When the difference between the current collected signal quantity value and the average value exceeds 3 times the standard deviation, the signal is considered to be an abnormal signal.

[0065] When an abnormal signal is detected, the system will mark the signal and correct it in the following way: The data of adjacent sensors are used for interpolation calculation to estimate the reasonable value of the signal. If the data of adjacent sensors is not available, the historical data of the same period is used for filling. For example, if the signal strength fluctuates frequently in a certain period of time, the current abnormal signal can be corrected according to the historical signal strength data of the same period. In order to promptly inform users of possible network problems, when the abnormal signal lasts for more than 30 seconds, the system will push an alarm message through the notification bar of the terminal device, prompting the user "Network abnormality, please check the connection". At the same time, the system will record the relevant information of the abnormal signal, including the time when the abnormality occurred, the type and value of the abnormal signal, etc., for subsequent analysis and processing.

[0066] The present invention also includes a dynamic network selection mode adjustment function. The system monitors the user's network usage behavior and environmental changes in real time. When it detects that the user's usage scenario has changed significantly (such as from a static state to a mobile state) or the network environment has changed dramatically (such as entering a weak signal area), the network selection mode is automatically adjusted. Determine whether to adjust the mode, where is the mode adjustment probability, which reflects the possibility of the system adjusting the network selection mode. It is the amount of change in user behavior, such as when a user changes from a stationary state to a mobile state, or changes from using a video conferencing mode to using a game optimization mode, etc. The amount of change in user behavior can be obtained by analyzing the user's operation records and sensor data. is the user behavior change threshold, which is set to 0.7. When the user behavior change exceeds this threshold, it indicates that the user's behavior has changed significantly, and it is necessary to consider adjusting the network selection mode. It is the amount of change in the network environment, such as a sudden drop in signal strength, a sudden increase in network delay, etc. The amount of change in the network environment can be obtained through real-time monitoring and analysis of signal volume data. is the network environment change threshold, which is also set to 0.7. When the network environment change exceeds this threshold, it indicates that the network environment has changed significantly and the network selection mode needs to be adjusted. When it is greater than 1, the system will automatically adjust the network selection mode. When adjusting the mode, refer to the data in the historical database to quickly initialize the weight parameters and evaluation indicators in the new network selection mode to improve the efficiency of mode adjustment.

[0067] In the present invention, in the step of determining the multi-dimensional evaluation index, the user preference index is quantified through a user behavior analysis model. The model uses a deep learning algorithm (such as a long short-term memory network, LSTM) to analyze the user's historical network selection records, usage time, and application usage data to predict the user's network selection preferences in different scenarios. According to the prediction results, the user preference index is assigned a corresponding weight.

[0068] In the present invention, a network stability prediction mechanism is introduced in the network selection decision step. The historical signal volume data of each network is analyzed by a time series analysis algorithm (such as the autoregressive integrated moving average model, ARIMA) to predict the stability of the network in the future. When calculating the comprehensive score, the network stability prediction result is used as an important evaluation factor and is given a certain weight.

[0069] In the present invention, in the abnormal signal processing mechanism, for signals marked as abnormal, in addition to using historical data or data from adjacent sensors for correction, an alarm mechanism will also be triggered. When the duration of the abnormal signal exceeds a preset threshold (such as 30 seconds), the system will send an alarm message to the user, indicating that there may be a problem with the network.

[0070] In the present invention, in the dynamic network selection mode adjustment function, the system will establish a corresponding historical database according to different network selection modes. When the mode is adjusted, the weight parameters and evaluation indicators in the new network selection mode are quickly initialized with reference to the data in the historical database, thereby improving the efficiency of the mode adjustment.

[0071] The present invention includes the following modules:

[0072] Signal quantity acquisition module: The hardware configuration of the signal quantity acquisition module is crucial for accurately collecting signal quantity data. In the terminal device, the following hardware components are used:

[0073] 5G baseband chip: Qualcomm X65 chip is selected, which supports millimeter wave frequency band and can provide high-speed and stable 5G network connection. It has the characteristics of low power consumption and high performance, and can meet the precise collection requirements of 5G signal strength, bandwidth, delay and other parameters.

[0074] WiFi chip: Broadcom BCM4389 chip is used, which supports 6GHz frequency band and can provide higher WiFi transmission rate and lower interference. The chip has advanced signal processing capabilities and can accurately collect information such as WiFi signal strength and signal-to-noise ratio.

[0075] It integrates high-precision accelerometers, gyroscopes, ambient light sensors, pressure sensors, etc. Accelerometers and gyroscopes are used to detect the motion state of the device, and help determine whether the user is stationary, walking, riding, or in other moving scenes; ambient light sensors are used to sense the intensity of ambient light, and help determine whether the device is in an indoor or outdoor environment; pressure sensors can be used to detect information such as the altitude of the device, which is of great significance for the comprehensive evaluation of the network environment.

[0076] The software part of the signal acquisition module is mainly responsible for processing and managing the data collected by the hardware. A real-time operating system (RTOS), such as FreeRTOS, is used to ensure the timeliness and stability of data acquisition. The software program runs according to the following process:

[0077] Initialize sensors and communication interfaces to ensure that hardware devices are working properly.

[0078] Start the data acquisition task at a fixed time interval (such as 100ms) and read the signal quantity data from the sensor through communication protocols such as I2C and SPI.

[0079] The collected raw data is initially filtered to remove high-frequency noise and outliers. The moving average filtering algorithm is used to average multiple continuously collected data points to obtain smoothed signal data.

[0080] The processed data is stored in the local buffer, and at the same time, the data is transmitted to the network selection mode definition module and the multi-dimensional evaluation index determination module through the data interface.

[0081] Network selection mode definition module: The network selection mode definition module uses a database to store preset network selection modes and their corresponding weight parameters. The database uses SQLite, which is lightweight, efficient, and takes up few resources, making it suitable for running on terminal devices. The database stores detailed information on various network selection modes, such as high-speed stable mode, low-power mode, low-cost mode, game optimization mode, video conferencing mode, etc., including the weight coefficients of each semaphore indicator in each mode, descriptions of the applicable scenarios of the mode, etc. According to the current usage scenarios and user needs, the comprehensive network selection weights of each network selection mode are dynamically calculated. By calling the data interface provided by the semaphore acquisition module, real-time semaphore data is obtained and substituted into the weight calculation formula At the same time, this module is also responsible for updating the weight parameters according to user feedback and the system's optimization strategy. When the user is not satisfied with the network switching result, the system will record the user's feedback information and adjust the weight coefficient in the corresponding network selection mode according to the optimization suggestions provided by the feedback optimization module. The updated weight parameters will be saved in the database for use in the next calculation.

[0082] Multi-dimensional evaluation index determination module: responsible for quantifying non-technical indicators such as network costs, equipment compatibility, and user preferences.

[0083] Network cost: Through real-time communication with the operator's API interface, the user's traffic package information, remaining traffic, tariff standards and other data are obtained. Based on this data and the user's actual network usage, the real-time cost of using each network is calculated. For example, if a user downloads 100MB of data using the 5G network, the corresponding fee is calculated based on the unit price of the package's traffic.

[0084] Device compatibility: Use device fingerprint recognition technology to obtain detailed information about the terminal device, including device model, operating system version, hardware configuration, etc. This information is compared with the pre-established device compatibility database, and a corresponding compatibility score is assigned to each network based on the matching results.

[0085] User preference: The user's historical network selection records are analyzed and predicted through the long short-term memory network (LSTM) model. The model will learn the user's network selection behavior patterns at different times and in different scenarios, and predict the probability distribution of the user's preference for each network based on the current time, location, application used, and other information. The quantified non-technical indicators are integrated with the technical indicators provided by the signal acquisition module, and a corresponding weight is assigned to each indicator. The comprehensive evaluation score of each network is calculated by weighted summation. Finally, the comprehensive evaluation score and detailed information of each indicator are output to the real-time evaluation module as an important basis for network selection decisions.

[0086] Real-time evaluation module: Dynamically adjust the weights of various evaluation indicators based on the current network selection mode and multi-dimensional evaluation indicators. Use reinforcement learning algorithms, such as the deep Q network (DQN), to continuously optimize the weight allocation strategy based on historical network selection results and user feedback. Under different network selection modes, the weights of various evaluation indicators will be adjusted according to the characteristics of the mode. For example, in high-speed stable mode, the weights of indicators such as signal strength, bandwidth, and latency will be relatively high; while in low-power mode, the weight of power consumption indicators will increase significantly. The real-time evaluation module uses the formula based on the adjusted weights and the evaluation indicator values ​​of each network. Calculate the comprehensive score of each network. During the calculation process, each evaluation indicator will be normalized to ensure comparability between different indicators. The calculated comprehensive score is output to the network selection decision module for network selection decision making.

[0087] Network selection decision module: Receives the comprehensive scores of each network output by the real-time evaluation module, and multiplies the score of the currently connected network by 90% as the switching threshold. When the comprehensive score of a target network exceeds the threshold, the network switching decision process is triggered. When making a switching decision, the network stability prediction results are also considered. If the target network has a high comprehensive score but is predicted to have poor future stability, the system may choose a network with a slightly lower score but better stability.

[0088] Record detailed information of each network selection decision, including decision time, current connected network, target network, comprehensive score of each network, switching threshold, etc. These records will be saved in the local database and uploaded to the cloud server for backup and analysis. In addition, the network selection decision module will also feed back the decision results to the switching execution module and feedback optimization module for subsequent operations and optimization.

[0089] Switching execution module: responsible for executing network switching operations. After receiving the switching instruction from the network selection decision module, it operates according to the following process:

[0090] Send a disconnect request to the currently connected network and ensure that the disconnection operation is completed within 50ms by calling the network management interface of the operating system.

[0091] Start the WiFi scan or 5G base station search function to obtain a list of available networks. Evaluate each available network based on parameters such as signal strength and channel quality, and select the best network to connect to.

[0092] Perform security authentication and parameter negotiation with the target network, using WPA3 and 5GNR authentication mechanisms to ensure the security of the connection. During the negotiation process, encryption keys and authentication information are exchanged to complete identity authentication and security parameter configuration.

[0093] After establishing a network connection, verify the connectivity and stability of the network through PING test, TCP connection test, etc. If the network delay is less than 100ms and the packet loss rate is less than 1%, the connection is considered successful; otherwise, reselect the network or adjust the connection parameters.

[0094] If a connection failure occurs during the network switching process, the switching execution module will record the cause of the failure, including unreachable target network, authentication failure, signal interference, etc. Take appropriate measures according to the cause of the failure. If it is a signal interference problem, it will try to switch to other channels or frequency bands; if it is an authentication failure, it will re-acquire the authentication information and try again. After three consecutive failed attempts, it will automatically switch to the suboptimal network and feedback the failure information to the network selection decision module and feedback optimization module for further analysis and optimization.

[0095] Feedback optimization module: collects user feedback and actual usage effect data. By popping up questionnaires on terminal devices and recording user operation behaviors, the user's satisfaction score for network switching, traffic consumption, network connection duration and other information are obtained. At the same time, network selection decision records and switching failure information are obtained from the network selection decision module and the switching execution module. The collected data is analyzed, and data analysis algorithms such as cluster analysis and association analysis are used to explore potential patterns and problems in the data. For example, cluster analysis can be used to divide users into different user groups, and analyze the network selection preferences and needs of each group; association analysis can be used to find the correlation between network switching failure and factors such as signal strength and network cost.

[0096] According to the data analysis results, the feedback optimization module uses the gradient descent method to optimize and adjust the weight parameters of the network selection mode. By continuously iteratively updating the weight coefficient, the value of the loss function is gradually reduced, thereby improving the accuracy and adaptability of network selection. At the same time, the feedback optimization module will also update the long short-term memory network (LSTM) model and the deep Q network (DQN) model. The model is retrained using the newly collected user data, and the model parameters and structure are adjusted to adapt to the needs of different users and changes in the network environment. The system updates the model every 24 hours to ensure the effectiveness and timeliness of the model.

[0097] The present invention also includes an abnormal signal processing module. When the collected signal quantity data shows abnormal fluctuations, the abnormal judgment formula If it is judged as an abnormal signal, the signal will be marked and corrected using historical data or data from adjacent sensors. When the abnormal signal lasts for more than 30 seconds, the abnormal signal processing module triggers the alarm mechanism. The alarm information is pushed through the notification bar of the terminal device, prompting the user "Network abnormality, please check the connection". At the same time, the detailed information of the abnormal signal is sent to the system administrator or uploaded to the cloud server for remote monitoring and processing.

[0098] The present invention also includes a dynamic network selection mode adjustment module. The system monitors the user's network usage behavior and environmental changes in real time, and adjusts the mode through the formula Determine whether to adjust the mode. After determining that the network selection mode needs to be adjusted, the dynamic network selection mode adjustment module refers to the data in the historical database to quickly initialize the weight parameters and evaluation indicators under the new network selection mode. By calling the interface of the network selection mode definition module and the multi-dimensional evaluation indicator determination module, the corresponding parameters and indicators are updated. Then, the new network selection mode information is passed to the real-time evaluation module and the network selection decision module to perform a new round of network selection decision and switching operations.

[0099] Reference Figure 2,Through visual weight comparison, the core innovation of the present ion is clearly presented: the ,multi-mode differentiated network selection strategy. ,By dynamically adjusting the indicator weights, the system can achieve scenario ,adaptation in a complex network environment, taking into account stability, economy and ,user experience, and providing a data basis for subsequent real-time evaluation ,and switching decisions.

[0100] Reference Figure 3 Through long-term trend data, the core advantage of this patent is clearly demonstrated: the continuous improvement of network switching performance through dynamic optimization mechanism. The significant increase in network selection success rate and the sharp decline in user complaint rate verify the adaptability of the system in a complex network environment and the user experience optimization effect, providing empirical support for the technical value of the invention.

[0101] Reference Figure 4 By comparing the bandwidth utilization of the system of the present invention with that of the traditional method, the technical advantages of the multi-network collaboration mechanism are clearly demonstrated. Through multi-dimensional evaluation, dynamic mode adjustment and intelligent switching, the system achieves efficient complementarity between 5G and WiFi resources, significantly improves the overall bandwidth utilization, and provides empirical support for resource optimization in complex network environments.

[0102] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solutions and inventive concepts of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A multi-mode and multi-dimensional network selection switching method based on signal quantity detection, characterized in that: The following steps are involved: S1. Signal quantity data collection: The signal quantity data of various networks are collected in real time through sensors, and preliminary filtering is performed to remove noise interference; S2. Multi-mode definition: According to different usage scenarios and user needs, multiple network selection modes are defined. Each mode corresponds to a set of different network selection weight parameters. The weight calculation formula is used to calculate the network selection weight parameters. Determine, among which is the comprehensive network selection weight in the pth network selection mode, is the weight coefficient of the jth semaphore indicator in the pth mode, is the value of the jth semaphore indicator; S3, multi-dimensional evaluation index determination: introduce signal volume data, network cost, device compatibility, and user preference evaluation indicators. Network cost takes into account traffic cost and package cost factors; device compatibility evaluates the adaptability of the network and terminal equipment; user preference is determined by user settings or historical network selection records; S4. Real-time evaluation: Based on the network selection mode, the network is evaluated in real time in combination with multi-dimensional evaluation indicators, and the comprehensive score of each network in the current mode is calculated. ,in is the comprehensive score of the k-th network, is the weight of the i-th evaluation indicator under the k-th network, is the value of the i-th evaluation indicator; S5, network selection decision: compare the comprehensive scores and select the network with the highest score as the current target network. If the score of the connected network is lower than the set threshold, the network selection and switching operation is triggered; S6. Switching execution: After determining the target network, the system performs a network switching operation, first disconnecting the current network connection, and then connecting to the target network. If the connection fails, the system will record the failure reason and try to reconnect or select a suboptimal network; S7, Feedback Optimization: After completing the network switching, the system collects user feedback and usage effect data, optimizes and adjusts the weight parameters and evaluation indicators of the network selection mode, and updates the weight coefficient through the machine learning algorithm; Dynamic network selection mode adjustment function: the system monitors the user's network usage behavior and environmental changes in real time. When it detects that the user's usage scenario or network environment has changed, it automatically adjusts the network selection mode through the mode adjustment formula Determine whether to adjust the mode, where Adjust the probability for the mode, The change in user behavior. Use behavior change thresholds for users, is the network environment change, is the network environment change threshold; In the step of determining the multi-dimensional evaluation indicators, the user preference indicators are quantified through the user behavior analysis model. The model uses a deep learning algorithm to analyze the user's historical data, predict the user's network selection preferences in different scenarios, and assign corresponding weights to the user preference indicators based on the prediction results.

2. The multi-mode and multi-dimensional network selection switching method based on signal quantity detection according to claim 1 is characterized in that: It also includes an abnormal signal processing mechanism. When the collected signal volume data shows abnormal fluctuations, the abnormal judgment formula is used to Make a judgment, is the current collected signal value, is the historical signal average value, is the standard deviation of the historical signal quantity, β is the preset abnormal coefficient, and if it is judged to be an abnormal signal, the signal will be marked and corrected.

3. The multi-mode and multi-dimensional network selection switching method based on signal quantity detection according to claim 2 is characterized in that: In the network selection decision step, a network stability prediction mechanism is introduced. The historical signal volume data of each network is analyzed and the network stability is predicted through a time series analysis algorithm. When calculating the comprehensive score, the network stability prediction result is used as an evaluation factor.

4. The multi-mode and multi-dimensional network selection switching method based on signal quantity detection according to claim 3 is characterized in that: In the abnormal signal processing mechanism, for signals marked as abnormal, historical data or data from adjacent sensors are used for correction, triggering the alarm mechanism. When the duration of the abnormal signal exceeds the preset threshold, the system will send an alarm message to the user.

5. The multi-mode and multi-dimensional network selection switching method based on signal quantity detection according to claim 4 is characterized in that: In the dynamic network selection mode adjustment function, the system will establish a corresponding historical database according to different network selection modes. When adjusting the mode, the weight parameters and evaluation indicators under the new network selection mode are initialized with reference to the historical database.

6. A system using the multi-mode multi-dimensional network selection switching method based on signal quantity detection according to any one of claims 2 to 5, characterized in that: Includes the following modules: Signal quantity acquisition module: collects signal quantity data of various networks in real time, uses sensors to collect data, and performs preliminary filtering on the collected data; Network selection mode definition module: Define multiple network selection modes according to different usage scenarios and user needs, and calculate them through weight formulas Determine the network selection weight parameters for each mode; Multi-dimensional evaluation index determination module: introduces signal volume data, network cost, device compatibility, and user preference multi-dimensional evaluation indicators. For network cost, traffic cost and package cost factors are considered; device compatibility evaluates the adaptability of the network and terminal equipment; User preferences are determined through user settings or historical network selection records; Real-time evaluation module: Based on the current network selection mode and combined with multi-dimensional evaluation indicators, the network is evaluated in real time and the comprehensive score of each network in the current mode is calculated. ; Network selection decision module: compares the comprehensive scores of all available networks and selects the network with the highest score as the current target network. If the score of the currently connected network is lower than the set threshold of the target network, the network selection and switching operation is triggered; Switching execution module: After determining the target network, perform network switching operations, first disconnect the current network connection, and then try to connect to the target network. During the connection process, monitor the connection status in real time. If the connection fails, record the failure reason, and try to reconnect or select the next best network; Feedback optimization module: After completing the network switching, collect user feedback and actual usage effect data, optimize and adjust the weight parameters and evaluation indicators of the network selection mode, and update the weight coefficient through the machine learning algorithm.

7. The system of the multi-mode and multi-dimensional network selection switching method based on signal quantity detection according to claim 6 is characterized in that: It also includes an abnormal signal processing module. When the collected signal data shows abnormal fluctuations, the abnormal judgment formula A judgment is made. If it is judged to be an abnormal signal, it will be marked and corrected using historical data or data from adjacent sensors. When the duration of the abnormal signal exceeds the preset threshold, the alarm mechanism is triggered and an alarm message is sent to the user.

8. The system of the multi-mode and multi-dimensional network selection switching method based on signal quantity detection according to claim 6 is characterized in that: It also includes a dynamic network selection mode adjustment module. The system monitors the user's network usage behavior and environmental changes in real time and adjusts the mode through the formula Determine whether to adjust the mode. When the mode needs to be adjusted, refer to the data in the historical database to initialize the weight parameters and evaluation indicators under the new network selection mode.

Images