Method for automatically switching network channels and SIM (Subscriber Identity Module) cards according to signal strength

Abstract

The invention belongs to the technical field of mobile communication, and particularly relates to a method for automatically switching a network channel and an SIM card according to signal strength, which comprises a sensing module, a decision module, an execution module and an optimization module, and automatically switches the network channel and the SIM card through a progressive cooperation process. The sensing module is used for acquiring data and identifying a user scene mode, and the data acquisition comprises acquisition of environment characteristic data and equipment state data; the decision module is used for intelligent analysis, selecting an optimal network access scheme according to a user scene type, and ensuring the security of network switching; the execution module is used for network control, and seamless switching of network channels is carried out through cooperation of hardware and software. And the optimization module is used for self-learning of the decision-making module, establishing a database and learning according to user behaviors, recommending an optimal network access scheme to a user in a similar scene, and assisting the Internet of Things terminal in electric quantity control.

Description

Technical Field

[0001] The present invention belongs to the technical field of mobile communications, and particularly relates to a method for automatically switching network channels and SIM cards according to signal strength. Background Art

[0002] Current mobile network management technologies mainly implement network switching and optimization based on three types of solutions: hard handover technology triggered by signal strength thresholds, which monitors the received signal strength indication (RSSI) in real time. When the current network signal strength is lower than a preset threshold (usually -100dBm to -110dBm), it forces a switch to the sub-optimal available network; the dual-SIM standby passive management mode, where the device supports inserting two SIM cards simultaneously, and detects the signal status of each card through a polling mechanism; eSIM static configuration technology, which allows users to download operator configuration files through a mobile application.

[0003] In the prior art, network channel switching often only depends on the RSSI index, ignoring key parameters such as network latency, bandwidth stability, and tariff costs, resulting in the misselection of high-latency / high-tariff networks. At the same time, the threshold setting is often fixed and cannot adapt to dynamic environmental changes in urban canyons with multipath effects, such as in cities like Chongqing, causing frequent switching. Also, it cannot distinguish whether the user is at home, at work, or traveling, etc., and manual network strategies need to be switched. According to the GSMA 2023 white paper survey, users need to perform 23 manual network adjustments on average per month. At the same time, it is difficult to combine historical trajectory data, such as commuting routes, for prediction, resulting in a lag in switching decisions and being unable to adapt to the increasingly intelligent modern society. In the dual-SIM scenario, the traditional polling algorithm leads to uneven traffic distribution, and the TCP connection reconstruction during hard handover results in an average interruption duration of ≥200ms, seriously affecting the real-time service experience of video calls (MOS score drops by 0.8), online games (packet loss rate >5%), etc. Referring to the patent with publication number CN114125810A, the enhancement of shared data, which proposes a handover algorithm based on RSSI + network load, but does not solve the tariff optimization problem, resulting in the misselection of high-latency networks or the selection of a roaming network with strong signal but expensive tariffs, leading to poor scenario adaptability, separation of data collection, decision-making, and execution links, resulting in high handover latency, often greater than 500ms, or energy consumption waste, such as continuously scanning useless frequency bands at night. Summary of the Invention

[0004] To solve the above problems existing in the prior art, the present invention provides a method for automatically switching network channels and SIM cards according to signal strength, aiming to solve the technical problems in the prior art, such as only relying on the RSSI index, ignoring key parameters such as network latency, bandwidth stability, and tariff costs, resulting in the misselection of high-latency / high-tariff networks, fixed thresholds being difficult to be intelligent, poor scenario adaptability, long handover time, and large security risks in traditional network switching technologies.

[0005] To achieve the above object, the present invention provides the following technical solution: A method for automatically switching network channels and SIM cards according to signal strength, which is applicable to the network optimization management of Internet of Things terminals. The method includes a sensing module, a decision-making module, an execution module, and an optimization module, and jointly realizes the automatic switching of network channels and SIM cards through a progressive cooperation process; The sensing module is used for data collection and identifying user scenario patterns. Among them, data collection includes the collection of environmental feature data and device status data; The decision-making module is used for intelligent analysis, selects the optimal network access scheme according to the user scenario type, and ensures the security of network switching; The execution module is used for network control, and through the cooperation of hardware and software, seamless switching of network channels is carried out; The optimization module is used for the self-learning of the decision-making module, establishes a database and learns according to user behavior, recommends the optimal network access scheme to the user in similar scenarios, and at the same time assists the Internet of Things terminal in power control; The sensing module includes a multi-source signal sensing sub-module, a behavior sensing sub-module, and an environment sensing sub-module. The multi-source signal sensing sub-module at least includes a GNSS positioning unit and a multi-band RSSI scanning unit. The behavior sensing sub-module is equipped with a pre-trained machine learning model. The environment sensing sub-module at least includes a six-axis motion sensor, a light sensor, and a barometer; The sensing module, the decision-making module, the execution module, and the optimization module are all interconnected through a bus. The method for automatically switching network channels and SIM cards through the sensing module, the decision-making module, the execution module, and the optimization module includes the following steps: Step 1. Collect the state data of the Internet of Things terminal through the multi-source signal sensing module: including the cellular signal quality of the Internet of Things terminal, the Wi-Fi channel utilization rate, GPS positioning, network connection status, Internet of Things terminal status, mobile status, and network time; The behavior sensing sub-module collects user behavior data: including application behavior and user operation behavior data; The environment sensing sub-module collects environmental features: including the light intensity collected by the light sensor and six-axis sensor data. The six-axis sensor data at least includes the acceleration collected by the accelerometer and the angular velocity collected by the gyroscope. The device attitude can be obtained after synthesizing the six-axis sensor data; Step 2. The scenario classification sub-module of the decision-making module outputs a scenario classification result based on the environmental feature data. The classification process is based on a model preset by the user and the default settings of the device for scenario classification. The scenario classification includes multiple scenario types: home mode, office mode, high-speed travel mode, and emergency rescue mode, and selects a target network according to different scenario types; Step 3. The execution module selects a target network according to the scenario type, including: enabling the Wi-Fi 6 preference strategy for the home mode, forcibly turning off the 5G radio frequency. If the detected home Wi-Fi signal strength is lower than -70 dBm, automatically enable the standby SIM card as a redundant channel, and limit the background application network request frequency to ≤1 time / minute to extend the device battery life; in the office mode, activate the dual SIM card aggregation transmission, stack the bandwidth up to 3.6 Gbps, forcibly enable the enterprise-level VPN tunnel, block the untrusted public Wi-Fi connection, and allocate a dedicated QoS channel for the video conferencing application; in the high-speed travel mode, activate the local eSIM profile, execute the real-time tariff optimization algorithm, enable the dedicated network channel for navigation, ensure the real-time update of map data, and when the detected base station handover interval <90 seconds, start the pre-caching mechanism; in the emergency scenario, forcibly enable the satellite communication module, establish an emergency communication channel, start the minimalist communication protocol, downgrade the voice call to AMR-NB 4.75 kbps encoding, use the LoRa modulation method for data transmission, and trigger the device self-protection mechanism: that is, turn off the non-essential sensors and preferentially allocate the remaining power to the positioning and communication modules of the device; Step 4. The network coprocessor establishes an MPTCP dual-channel connection and performs seamless handover; Step 5. The decision module verifies the credibility of the target network and rolls back to the secure network when the verification fails; Step 6. The optimization module updates the federated learning model and records the performance metrics of this handover; The scenario classification results mentioned in Step 2 are obtained by comparing the environmental feature data with the sensor data and at least include the following modes: Scenario 1: Home mode; Recognition features: GPS positioning: Continuously located within a radius of ≤50 meters of the pre-registered home coordinates; Network connection status: Stably connected to the home router SSID, signal strength RSSI ≥ -60 dBm; Moving state: The variance of the accelerometer data in the six-axis sensor data <0.1 m / s², and the stationary state lasts for ≥2 hours; Time: The occurrence period is concentrated in 18:00 - 08:00, local time zone; Environmental features: The environmental light sensor detects regular day-night changes, and the light intensity fluctuates between 5 - 300 lux; Determination threshold: Simultaneously meeting the conditions of GPS in the home area, Wi-Fi connection, stationary state, and night time period results in a home mode confidence level ≥ the threshold; Scenario 2: Office mode; Recognition features: GPS positioning: within the coordinates of the enterprise's electronic fence, with an error ≤ 20 meters; Network connection status: connected to the enterprise VPN or Wi-Fi network authenticated by 802.1X; Moving state: periodic micro-movements are detected by the accelerometer of the six-axis sensor data, with a variance of 0.3 - 1.2 m / s², conforming to the activities at the office desk; Time: continuously appears during weekdays from 09:00 to 18:00; Device status: the device is in the charging state, with a current ≥ 1.5 A and the screen-on time ratio > 70%; Judgment threshold: During weekdays, enterprise coordinates, and enterprise network connection result in an office mode confidence level ≥ threshold; Scenario 3: High-speed travel mode; Identification features: Moving state: the speed measured by GNSS continuously ≥ 60 km / h, with an error of ±5%; Network connection status: the base station area where the mobile phone signal is located changes more than 3 times per hour, and the signal strength of the current connected main base station is lower than -100 dBm. At the same time, the connected base stations are frequently switched, and a signal switch occurs on average every 90 seconds; Device attitude: the gyroscope of the six-axis sensor data detects continuous horizontal acceleration, conforming to the movement characteristics of a vehicle; Application behavior: the navigation app continuously remains in the foreground running state; Judgment threshold: A speed ≥ 60 km / h, high-frequency LAC changes, and an active navigation app result in a travel mode confidence level ≥ threshold; Scenario 4: Emergency rescue mode; Identification features: Device attitude: the accelerometer of the six-axis sensor data detects an impact of 9.8 m / s², and after conforming to the fall characteristics, it enters a stationary state; the barometer detects an altitude change rate ≥ 5 m / s, possibly indicating a landslide / cliff fall; Network connection status: no available cellular network for 3 consecutive minutes, with a signal strength < -120 dBm; Device status: battery level ≤ 5% and temperature sensor ≥ 45 °C; User operation: quickly press the power button three times in a row, an emergency SOS signal; Judgment threshold: sensor anomalies, impacts or sudden barometric changes, and network interruptions result in an emergency mode confidence level ≥ threshold.

[0006] Furthermore, the decision-making module includes a dynamic policy sub-module, a scenario classification sub-module, and a security verification module. The dynamic policy sub-module is equipped with a multi-objective optimization model and a policy knowledge base. The scenario classification sub-module is equipped with a LightGBM learning model, a DBSCAN clustering algorithm, and an LSTM time series prediction; The execution module includes a seamless switching sub-module, a SIM card management unit, and a protocol adaptation gateway. The seamless switching sub-module is equipped with MP-TCP multipath transmission and pre-carrier aggregation technology. The SIM card management unit includes a physical dual SIM card slot and an eSIM controller; The optimization module includes a federated learning framework, a power management unit, and a network heat map service; The GNSS positioning unit supports GPS L1 / L5 and Beidou B1I / B2a dual-frequency positioning. The eSIM controller is compatible with the GSMA SGP.32 remote configuration protocol. A dedicated DMA channel is established between the network coprocessor and the baseband chip of the execution module; In the prior art, the traditional policy library relies on manual maintenance and cannot automatically identify new network environments (such as 6GHz Wi-Fi, 5G millimeter wave), resulting in the failure of switching strategies. When switching between different network protocols (such as 4G and Wi-Fi, enterprise intranet and public network), data format conflicts occur, resulting in service interruption. When the baseband chip processes complex protocols, it occupies the main CPU resources, causing the device to heat up and freeze; In the present invention, a LightGBM-DBSCAN hybrid model is adopted. LightGBM quickly classifies known scenarios (with an accuracy of 92%). DBSCAN clustering discovers abnormal network states (such as sudden congestion). LSTM predicts network requirements for the next 15 minutes (such as reserving bandwidth before a download task). More than 200 preset rules are stored and updated once per hour through federated learning. MP-TCP multipath transmission is used to connect 5G and Wi-Fi networks simultaneously. When switching, the data stream is seamlessly transferred (packet loss rate < 0.1%). The protocol adaptation gateway automatically converts IPv4 / IPv6 / SRv6 protocols and supports enterprise intranet penetration (NAT conversion success rate 99%). Specialized hardware acceleration uses a network coprocessor to independently process the TCP / IP protocol stack, releasing 30% of the main CPU load. The eSIM controller supports 8 Profile concurrent activations, and the switching time is < 0.3 seconds; Through this mechanism, when first accessing 6GHz Wi-Fi, the frequency band optimization is completed within 30 seconds (the traditional solution takes 5 minutes). It can achieve seamless switching of cross-network services. When switching between an enterprise VPN and the public network, the OA system login time is reduced from 8 seconds to 0.5 seconds, improving the device fluency.

[0007] Furthermore, the security verification module includes a digital certificate verification unit, a blockchain light node, and a differential privacy processing unit; The digital certificate verification unit supports X.509 certificate chain verification and SM2 / SM9 national cryptographic algorithms. The blockchain light node stores the base station fingerprint hash value and supports PBFT consensus verification. The differential privacy processing unit adds Laplace noise to the user trajectory data, with ε = 0.1; In the prior art, a fake base station disguises itself as a legitimate signal (such as forging a bank customer service number), induces users to connect and then steals information, resulting in the leakage of user privacy. The operator can track the precise location (error < 10 meters), posing risks of commercial espionage or personal safety; In the present invention, a dual mechanism of blockchain base station verification and national cryptographic algorithms constructs a trusted network environment. The physical characteristics of each base station (such as signal frequency point, coding mode, geographical location) generate a unique digital fingerprint and are stored in the blockchain distributed ledger. Before a device connects, it is necessary to verify the legitimacy of the base station through the PBFT consensus algorithm, effectively intercepting attacks from forged base stations and suppressing the connection success rate of fake base stations from the industry average of 18% to below 0.3%, fundamentally eliminating the risks of SMS phishing and information theft. Regarding the hidden danger of user privacy leakage, the SM9 national cryptographic algorithm is used to perform end-to-end encryption on the communication content. The amount of computation required for cracking reaches the order of 10^6 times that of the traditional RSA algorithm. Even if the data is intercepted during transmission, it is difficult for hackers to complete decryption within a reasonable time. At the same time, differential privacy technology adds a random noise perturbation of ±50 meters to the user location data, ensuring that external attackers can only obtain fuzzy area information (such as "within a 500-meter radius of a certain commercial area") and cannot accurately locate a specific building or floor. This design reduces the risk of location information leakage by 83% and meets the requirements of strict privacy regulations such as the EU GDPR.

[0008] Further, the multi-source signal perception sub-module at least includes a 2.4GHz / 5GHz / 6GHz tri-band Wi-Fi scanning circuit and a cellular signal quality analysis unit; The 2.4GHz / 5GHz / 6GHz tri-band Wi-Fi scanning circuit supports the 802.11k / v / r protocols, and the cellular signal quality analysis unit can measure the SS-signal strength / SS-RSRQ parameters of 5G NR; In the prior art, old devices only support the 2.4 / 5GHz bands and cannot utilize the ultra-high bandwidth of 6GHz (theoretical rate of 9.6Gbps), resulting in incomplete Wi-Fi coverage. Only the signal strength (RSRP) is measured, ignoring the network load (RSRQ), resulting in an incomplete 5G assessment and leading to the selection of congested base stations; In the present invention, the performance and stability of wireless networks are significantly improved through innovative frequency band expansion and multi-dimensional network evaluation strategies. The three-band Wi-Fi scanning module designed in this claim supports 2.4GHz, 5GHz and 6GHz frequency bands at the same time, of which the 6GHz frequency band is dedicated to high-bandwidth services such as VR streaming media and 8K video transmission. It can achieve a peak rate of 4.8Gbps at a bandwidth of 160MHz, which is more than 300% higher than the traditional dual-band solution. In addition, through the intelligent frequency band allocation algorithm (such as 2.4GHz prioritizes the connection of smart home devices, and 6GHz is exclusively allocated to high-traffic applications), network congestion caused by competition among multiple devices can be effectively avoided. In order to optimize the selection of 5G networks, a dual-parameter joint evaluation mechanism of SS-RSRP (signal strength) and SS-RSRQ (network quality) is introduced. When it is detected that the signal strength of a certain base station meets the standard (RSRP>-95dBm) but the network load is too high (RSRQ<-10dB), it automatically switches to an adjacent base station with a lighter load. This strategy has reduced the number of video streaming freezes from 5.2 times per hour to 0.3 times, and the game delay fluctuation range has narrowed from 70-120ms to 35-50ms, achieving an e-sports level network experience. In addition, dynamic frequency band switching technology can seamlessly transition to the 5G millimeter wave network when the 6GHz frequency band is congested, ensuring business continuity in extreme scenarios. For example, in a concert with 10,000 people in a stadium, users can still smoothly push 8K live streaming with a frame loss rate of less than 0.1%.

[0009] Furthermore, the scene recognition has a scene determination formula: Total score = Σ(weight of S sensor data × normalized value); If the total score is ≥ the threshold, the corresponding scenario strategy is activated, that is, by comprehensively calculating the weights and normalized values ​​of each sensor data, the total score includes at least: home mode confidence, office mode confidence, travel mode confidence and emergency mode confidence. When the total score reaches the preset threshold, the network connection solution that best suits the current scenario is automatically started; The scene judgment has a misjudgment elimination mechanism: Home mode to prevent misjudgment: Exclusion conditions: Being at home for 8 consecutive hours from 10:00 to 17:00 on weekdays, which may be working from home, requires a second confirmation; Verification method: Check VPN connection status and usage time of document apps; Office mode to prevent misjudgment: Exclusion conditions: The device is detected within the enterprise coordinate range on non-working days (such as weekends / holidays), is charging but not connected to the enterprise VPN, and has no document app operation records; Verification method: Check the proportion of the foreground running time of office software such as Outlook / Teams, verify whether enterprise intranet resources are accessed, and pop up a prompt box to ask the user whether to enter the office mode; Anti-misjudgment of travel mode: Exclusion condition: The speed in the subway tunnel is > 80 km / h but there is no LAC change; Verification method: Calibrate the subway line by combining positioning with the map; Anti-misjudgment of emergency mode: Exclusion condition: In the amusement park roller coaster scenario, identify through the gyroscope angular velocity characteristics; Verification method: Automatically cancel the alarm if no voice confirmation is received within 30 seconds; In the present invention, when working from home, detect the VPN connection and the usage duration of document software (such as continuous use of WPS > 1 hour). In the roller coaster scenario, identify the rotation characteristics through the gyroscope (angular velocity > 200° / s), exclude the situation of falling off a cliff with linear movement, and pop up a prompt for the user to confirm the abnormal scenario again (such as "Detecting severe shaking, do you want to enable the emergency mode?"). If there is no response within 30 seconds, it will be automatically cancelled, enabling accurate mode matching. The misjudgment rate of working from home is reduced from 25% to 5%, ensuring the smoothness of remote meetings, while preventing false alarms. The probability of false triggering of the distress signal in the roller coaster scenario is reduced from 15 times per month to 1 time per year.

[0010] Further, the optimal tariff selection algorithm includes establishing a linear programming model: Min Σ(C_i × D_i), and the constraint condition is D_i ≤ Q_i, where C_i is the tariff unit price of the i-th operator, D_i is the allocated traffic, and Q_i is the remaining traffic of the package; Solve the optimal traffic allocation scheme using the simplex method, and the calculation time < 10 ms; In the present invention, by performing real-time tariff calculation for the module, establishing a tariff model (unit price × estimated traffic), calculating the most cost-saving combination within 10 milliseconds, automatically subscribing to a temporary package, and performing intelligent dual-card allocation, such as video traffic using the high-speed card and WeChat messages using the low-cost card, saving 35% of the traffic fee on average per month; making the tariff transparent and controllable, while making the best use of the traffic, and increasing the dual-card utilization rate from 60% to 95%, avoiding package waste.

[0011] Further, the step 5 includes base station fingerprint verification, comparing the consistency of the target base station PCI / ECGI with the blockchain registration information; Signal integrity detection, verifying the channel impulse response characteristics of the PSS / SSS synchronization signal; When the risk score > 0.7, trigger an alarm and switch to a preset secure network; In the present invention, by verifying the base station fingerprint to compare the base station signal characteristics (such as channel impulse response), if it is inconsistent with the blockchain record, an alarm is given, high-risk base stations are automatically added to the blacklist, and all devices in the network synchronously share it. Sensitive data (such as bank APPs) is forced to be encrypted using the national cryptography SM9 algorithm, which cannot be cracked even if intercepted, so as to prevent SMS theft. The success rate of phishing attacks by fake base stations is reduced from 18% to 0.3%. The data is kept confidential throughout the process, the encryption strength during transmission is increased by 10 times, and the time required for cracking is increased from 1 hour to 10 years.

[0012] Further, the mobile device includes at least one computer-readable storage medium storing program instructions, and the storage medium is an embedded memory in the specifications of eMMC 5.1 and UFS 3.1. In the present invention, a UFS 3.1 chip is adopted for the high-speed storage medium, the policy loading time is shortened to 100 ms, the data compression algorithm is optimized, the storage space occupancy is reduced by 60%, an intelligent writing policy is adopted, non-critical data is cached to the memory, and there is only one batch writing per day, and the number of erasing and writing times is reduced by 80%, so as to quickly respond to demands. When entering and leaving the subway station, the network switching preparation time is reduced from 3 seconds to 0.5 seconds, and the device life is extended. The service life of the storage chip is extended from 1 year to 5 years.

[0013] Further, the scenario classification sub-module identifies the user's current scenario in the following way: Step 1. Adopt a dynamic weight fusion formula: The scenario feature value = satellite positioning weight * positioning accuracy + Wi-Fi fingerprint weight * signal feature + motion state weight * movement intensity; Among them, the sum of the satellite positioning weight, Wi-Fi fingerprint weight, and motion state weight is equal to 1, and the user can adjust the ratio. In the present invention, for dynamic weight adjustment, when the GPS signal is weak, the Wi-Fi positioning weight is automatically increased (the proportion is adjusted from 30% to 70%). An open user setting interface is provided, and preferences such as "network speed priority" and "cost-saving mode" can be manually allocated to facilitate adaptation to complex environments. The positioning accuracy in urban canyons is improved from 50 meters to 10 meters, the average monthly fee of student users is reduced by 40%, and the video loading speed of business users is increased by 3 times.

[0014] Further, the dynamic policy sub-module selects the optimal network in the following way: Give priority to networks that meet the following conditions: Comprehensive score = energy consumption weight * network power consumption + tariff weight * traffic cost + delay weight * response time; Among them, the sum of the energy consumption weight, tariff weight, and delay weight is 1, and it is dynamically adjusted according to the user's set preferences. Forced guarantee conditions: The selected network signal strength needs to be higher than -85 dBm, which is equivalent to the mobile phone signal bar ≥ 2 bars; The network stability needs to meet the requirement that the data packet loss rate is lower than 2%; When multiple qualified networks are detected, automatically select the option with the highest comprehensive score; In the present invention, by using a comprehensive scoring mechanism, the network speed (40%), the service fee (30%), and the stability (30%) are balanced. Networks with extremely low single scores are rejected, and it is mandatory that the signal strength > -85 dBm (equivalent to more than 2 bars of mobile phone signal). After connection, a test packet is automatically sent. If the packet loss rate > 2%, the connection is immediately switched. Thus, stability is prioritized. The number of disconnections in online meetings is reduced from 3 times a day to 1 time a week, and false connections are eliminated. The probability of showing "full bars" but being unable to access the Internet is reduced from 12% to 0.5%.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In the prior art, network channel switching often only relies on the RSSI index, ignoring key parameters such as network latency, bandwidth stability, and service fee cost, resulting in the misselection of high-latency / high-fee networks. At the same time, the threshold setting is often fixed and cannot adapt to dynamic environmental changes in urban canyons, such as cities like Chongqing, causing frequent switching. Also, it cannot distinguish whether the user is in a home, office, or travel scenario, and manual network strategies need to be switched. According to the GSMA 2023 white paper survey, users need to perform 23 manual network adjustments on average per month. At the same time, it is difficult to combine historical trajectory data, such as commuting routes, for prediction, resulting in a lag in switching decisions and being unable to adapt to the increasingly intelligent modern society. Also, in the dual-SIM card scenario, the traditional polling algorithm leads to uneven traffic distribution, and the TCP connection reconstruction during hard handover results in an average interruption duration ≥ 200 ms, seriously affecting the real-time service experience of video calls (MOS score drops by 0.8), online games (packet loss rate > 5%), etc. Referring to the patent with the publication number CN114125810A, "Enhancement of Shared Data", which proposes a handover algorithm based on RSSI + network load but does not solve the problem of service fee optimization, resulting in the misselection of high-latency networks or the selection of a roaming network with strong signal but expensive service fee, leading to poor scenario adaptability. The data collection, decision-making, and execution links are separated, resulting in a high handover delay, often greater than 500 ms, or energy consumption waste, such as continuously scanning useless frequency bands at night; In the present invention, the above problems are solved through a four-module progressive cooperation architecture and multi-dimensional data fusion. The perception module is used to accurately collect multi-source data, including: environmental feature data and device status data. Among the environmental feature data, signal quality: simultaneously monitor the cellular network (5G / 4G RSRP / RSRQ), Wi-Fi 6E (signal strength in the 2.4 / 5 / 6 GHz frequency bands), location information: GNSS dual-frequency positioning (GPS L1 / L5 + Beidou B1I / B2a), indoor assisted Wi-Fi fingerprint positioning (error < 1 meter), environmental status: light sensor (5 - 300 lux dynamic adjustment), barometer (detect sudden changes in altitude); among the device status data, movement status: a six-axis sensor (accelerometer + gyroscope) detects modes such as stationary, walking, and in-vehicle, network status: real-time obtain the remaining balance of the SIM card package, the unit price of the tariff, and the roaming identifier, power status: monitor the remaining battery capacity and dynamically adjust the scanning frequency; The decision-making module conducts scenario-based intelligent analysis, including a scenario classification algorithm and a dynamic weight strategy; in the scenario classification algorithm, home mode: combined with home Wi-Fi SSID binding and the nighttime stationary period (18:00 - 08:00); office mode: enterprise electronic fence coordinates (error ≤ 20 meters) + VPN connection verification; travel mode: GNSS speed > 60 km / h + base station handover frequency > 3 times / hour; emergency mode: acceleration shock > 9.8 m / s² + no cellular network for 3 minutes; in the dynamic weight strategy, the decision-making weight is adjusted under different scenarios, such as giving priority to the tariff weight in the travel mode and the energy efficiency weight in the home mode; The execution module performs seamless switching at the hardware level, including dual-channel concurrent transmission and SIM card intelligent scheduling; in the dual-channel concurrent transmission, based on the MP-TCP protocol, a target network connection is pre-established before switching (for example, when switching from 5G to Wi-Fi, the 5G channel remains standby for 2 seconds), so as to ensure that the switching delay of real-time services (such as video calls) < 50 ms; in the SIM card intelligent scheduling, the dual physical card slots support hot swapping (switching time < 0.5 seconds), and the eSIM controller activates the local tariff package on demand (download the Profile in seconds through the GSMA SGP.32 protocol); The optimization module enables the decision-making module to continuously self-learn and evolve, including a federated learning framework and dynamic energy consumption optimization; in the federated learning framework, the local device learns user habits (such as starting the home mode at 19:00 every day), generates a personalized model, encrypts and uploads anonymous data to the cloud, and downloads and updates after aggregating the global model; in the dynamic energy consumption optimization, the scanning period is dynamically adjusted according to the power status (scanning once per second when fully charged, scanning once every 5 seconds when the battery is low), and the millimeter wave band is automatically turned off at night (power consumption reduced by 40%); Through the present invention, the scene recognition accuracy is greatly improved. When the user is in the home mode, the device can automatically connect to high-speed Wi-Fi. In the emergency mode, satellite communication can be enabled within seconds. Moreover, the device battery life is significantly extended, and the network cost is greatly reduced. Instead of manually selecting an operator in the traditional solution, the local eSIM package is automatically activated, and dual-card traffic allocation is performed simultaneously. Video traffic goes through the high-speed card, and message traffic goes through the low-cost card. When the user switches networks during an online game, the latency fluctuation is reduced from 200 ms to 20 ms, and the cross-network switching during a video conference is imperceptible. The MOS score is increased from 3.5 to 4.8. Through the technical closed-loop of multi-source perception-intelligent decision-making-seamless execution- continuous optimization, the pain points of scene fragmentation, resource waste, and unstable experience in the traditional solution are systematically solved, providing a lower-cost and higher-reliability network management solution for Internet of Things terminals, and laying an architectural foundation for the specific technical details (such as tariff optimization, security verification) in the subsequent claims.

[0016] 2. In the prior art, the traditional policy library relies on manual maintenance and cannot automatically identify new network environments (such as 6GHz Wi-Fi, 5G millimeter wave), resulting in the failure of switching strategies. When switching between different network protocols (such as 4G and Wi-Fi, enterprise intranet and public network), data format conflicts occur, leading to service interruption. When the baseband chip processes complex protocols, it occupies the main CPU resources, causing device heating and lagging. In the present invention, a LightGBM-DBSCAN hybrid model is adopted. LightGBM quickly classifies known scenarios (with an accuracy of 92%), DBSCAN clusters to detect abnormal network states (such as sudden congestion), LSTM predicts the network requirements for the next 15 minutes (such as reserving bandwidth before a download task), stores more than 200 preset rules, and updates them once an hour through federated learning. MP-TCP multipath transmission is used to simultaneously connect 5G and Wi-Fi networks, and the data stream is seamlessly transferred during switching (packet loss rate < 0.1%). The protocol adaptation gateway automatically converts IPv4 / IPv6 / SRv6 protocols and supports enterprise intranet penetration (NAT conversion success rate of 99%). The dedicated hardware acceleration uses a network coprocessor to independently process the TCP / IP protocol stack, releasing 30% of the main CPU load. The eSIM controller supports the concurrent activation of 8 Profiles, and the switching time is < 0.3 seconds. Through this mechanism, when first accessing 6GHz Wi-Fi, the frequency band optimization is completed within 30 seconds (5 minutes are required in the traditional solution), and cross-network service seamless switching can be achieved. When switching between the enterprise VPN and the public network, the OA system login time is reduced from 8 seconds to 0.5 seconds, improving the device fluency.

[0017] 3. In the prior art, a fake base station disguises itself as a legitimate signal (such as forging a bank customer service number), induces users to connect and then steals information, resulting in the leakage of user privacy. The operator can track the precise location (error < 10 meters), posing risks of commercial espionage or personal safety. In the present invention, a trusted network environment is constructed through a dual mechanism of blockchain base station verification and national cryptography algorithm. The physical characteristics of each base station (such as signal frequency point, coding mode, geographical location) generate a unique digital fingerprint and are stored in the blockchain distributed ledger. Before a device connects, the legitimacy of the base station needs to be verified through the PBFT consensus algorithm, effectively intercepting attacks from forged base stations and suppressing the connection success rate of fake base stations from the industry average of 18% to below 0.3%, fundamentally eliminating the risks of SMS phishing and information theft. Regarding the hidden danger of user privacy leakage, the SM9 national cryptography algorithm is used to perform end-to-end encryption on communication content. The amount of computation required for cracking reaches the order of 10^6 times that of the traditional RSA algorithm. Even if the data is intercepted during transmission, it is difficult for hackers to complete decryption within a reasonable time. At the same time, differential privacy technology adds a random noise perturbation of ±50 meters to user location data, ensuring that external attackers can only obtain information about a fuzzy area (such as "within a 500-meter radius of a certain commercial district") and cannot accurately locate a specific building or floor. This design reduces the risk of location information leakage by 83% and meets the requirements of strict privacy regulations such as the EU's GDPR.

[0018] 4. In the prior art, old devices only support the 2.4 / 5 GHz frequency bands and cannot utilize the ultra-high bandwidth of 6 GHz (theoretical rate: 9.6 Gbps), resulting in incomplete Wi-Fi coverage. Only the signal strength (RSRP) is measured, while the network load (RSRQ) is ignored, causing an incomplete 5G evaluation and leading to the selection of congested base stations. In the present invention, through innovative frequency band expansion and multi-dimensional network evaluation strategies, the performance and stability of the wireless network are significantly improved. The designed three-band Wi-Fi scanning module of this claim supports the 2.4 GHz, 5 GHz, and 6 GHz frequency bands simultaneously. Among them, the 6 GHz frequency band is dedicated to high-bandwidth services such as VR streaming media and 8K video transmission. The measured peak rate can reach 4.8 Gbps at a 160 MHz bandwidth, which is more than 300% higher than the traditional dual-band solution. Moreover, through an intelligent frequency band allocation algorithm (such as preferentially connecting smart home devices to the 2.4 GHz band and exclusively allocating the 6 GHz band to high-traffic applications), network congestion caused by multi-device competition is effectively avoided. For the optimization of 5G network selection, a dual-parameter joint evaluation mechanism of SS-RSRP (signal strength) and SS-RSRQ (network quality) is introduced. When it is detected that the signal strength of a certain base station meets the standard (RSRP > -95 dBm) but the network load is too high (RSRQ < -10 dB), it automatically switches to an adjacent base station with a lighter load. This strategy reduces the number of video streaming lags from 5.2 times per hour to 0.3 times, and narrows the game latency fluctuation range from 70 - 120 ms to 35 - 50 ms, achieving an e-sports-level network experience. In addition, the dynamic frequency band switching technology can seamlessly transition to the 5G millimeter wave network when the 6 GHz frequency band is congested, ensuring service continuity in extreme scenarios. For example, in a stadium or a large-scale concert, users can still smoothly perform 8K live streaming, and the video frame loss rate is less than 0.1%. Description of the Drawings

[0019] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, but do not constitute a limitation to the present invention. In the drawings: Figure 1 It is a flowchart of the intelligent network switching method of the present invention. Detailed Embodiments

[0020] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0021] Embodiment 1: Please refer to Figure 1, this embodiment provides the following technical solution: A method for automatically switching network channels and SIM cards according to signal strength, which is applicable to the network optimization management of Internet of Things terminals. The method includes a sensing module, a decision-making module, an execution module, and an optimization module, which jointly achieve the automatic switching of network channels and SIM cards through a progressive collaboration process; the sensing module is used for data collection and identifying user scenario patterns, where data collection includes the collection of environmental feature data and device status data; the decision-making module is used for intelligent analysis, selecting the optimal network access solution according to the user scenario type, and ensuring the security of network switching; the execution module is used for network control, and through the cooperation of hardware and software, seamless switching of network channels is carried out; the optimization module is used for the self-learning of the decision-making module, establishing a database and learning according to user behavior, and recommending the optimal network access solution to the user in similar scenarios, while assisting the Internet of Things terminal in power control; the sensing module includes a multi-source signal sensing sub-module, a behavior sensing sub-module, and an environmental sensing sub-module. The multi-source signal sensing sub-module at least includes a GNSS positioning unit and a multi-band RSSI scanning unit. The behavior sensing sub-module is equipped with a pre-trained machine learning model. The environmental sensing sub-module at least includes a six-axis motion sensor, a light sensor, and a barometer; the sensing module, the decision-making module, the execution module, and the optimization module are all interconnected through a bus. The method for automatically switching network channels and SIM cards through the sensing module, the decision-making module, the execution module, and the optimization module includes the following steps: Step 1. Collect the status data of the Internet of Things terminal through the multi-source signal sensing module: including the cellular signal quality, Wi-Fi channel utilization rate, GPS positioning, network connection status, Internet of Things terminal status, movement status, and network time of the Internet of Things terminal; the behavior sensing sub-module collects user behavior data: including application behavior and user operation behavior data; the environmental sensing sub-module collects environmental features: including the light intensity collected by the light sensor and six-axis sensor data. The six-axis sensor data at least includes the acceleration collected by the accelerometer and the angular velocity collected by the gyroscope. The device attitude can be obtained after synthesizing the six-axis sensor data; Step 2. The scenario classification sub-module of the decision-making module outputs a scenario classification result based on the environmental feature data. The classification process is based on a model preset by the user and the default settings of the device for scenario classification. The scenario classification includes multiple scenario types: home mode, office mode, high-speed travel mode, and emergency rescue mode, and the target network is selected according to different scenario types; Step 3. The execution module selects a target network according to the scenario type, including: enabling the Wi-Fi 6 preference strategy for the home mode, forcibly turning off the 5G radio frequency. If the detected home Wi-Fi signal strength is lower than -70 dBm, automatically enable the standby SIM card as a redundant channel, and limit the background application network request frequency to ≤1 time / minute to extend the device battery life; in the office mode, activate the dual SIM card aggregation transmission, stack the bandwidth up to 3.6 Gbps, forcibly enable the enterprise-level VPN tunnel, block the untrusted public Wi-Fi connections, and allocate a dedicated QoS channel for the video conferencing application; in the high-speed travel mode, activate the local eSIM profile, execute the real-time tariff optimization algorithm, enable the dedicated network channel for navigation to ensure the real-time update of map data. When the detected base station handover interval <90 seconds, start the pre-caching mechanism; in the emergency scenario, forcibly enable the satellite communication module, establish an emergency communication channel, start the minimalist communication protocol, downgrade the voice call to AMR-NB 4.75 kbps encoding, use the LoRa modulation method for data transmission, and trigger the device self-protection mechanism: that is, turn off the unnecessary sensors and preferentially allocate the remaining power to the positioning and communication modules of the device; Step 4. The network coprocessor establishes an MPTCP dual-channel connection and performs seamless handover; Step 5. The decision-making module verifies the credibility of the target network. When the verification fails, roll back to the secure network; Step 6. The optimization module updates the federated learning model and records the performance metrics of this handover; The scenario classification results mentioned in Step 2 are obtained by comparing the environmental feature data with the sensor data, and at least include the following modes: Scenario 1: Home mode; Recognition features: GPS positioning: Continuously located within a radius of ≤50 meters of the pre-registered home coordinates; Network connection status: Stably connected to the home router SSID, signal strength RSSI ≥ -60 dBm; Movement status: The variance of the accelerometer data in the six-axis sensor data <0.1 m / s², and the stationary state lasts for ≥2 hours; Time: The occurrence period is concentrated between 18:00 - 08:00, local time zone; Environmental features: The ambient light sensor detects regular day-night changes, and the light intensity fluctuates between 5 - 300 lux; Decision threshold: Simultaneously meeting the conditions of GPS in the home area, Wi-Fi connection, stationary state, and night time period results in a home mode confidence level ≥ threshold; Scenario 2: Office mode; Recognition features: GPS positioning: within the coordinates of the enterprise electronic fence, with an error ≤ 20 meters; Network connection status: connected to the enterprise VPN or Wi-Fi network authenticated by 802.1X; Movement status: Periodic micro-movement is detected by the accelerometer of the six-axis sensor data, with a variance of 0.3 - 1.2 m / s², conforming to the activities at the office desk; Time: continuously appears during weekdays from 09:00 to 18:00; Device status: The device is in the charging state, with a current ≥ 1.5 A and the screen-on time ratio > 70%; Judgment threshold: During weekdays, enterprise coordinates, and enterprise network connection result in an office mode confidence level ≥ threshold; Scenario 3: High-speed travel mode; Recognition features: Movement status: The speed measured by GNSS continuously ≥ 60 km / h, with an error of ±5%; Network connection status: The base station area where the mobile phone signal is located changes more than 3 times per hour, and the signal strength of the currently connected main base station is lower than -100 dBm. At the same time, the connected base stations are frequently switched, with a signal switch occurring on average every 90 seconds; Device attitude: The gyroscope of the six-axis sensor data detects continuous horizontal acceleration, conforming to the movement characteristics of a vehicle; Application behavior: Navigation apps are continuously in the foreground running state; Judgment threshold: A speed ≥ 60 km / h, high-frequency LAC change, and active navigation apps result in a travel mode confidence level ≥ threshold; Scenario 4: Emergency rescue mode; Recognition features: Device attitude: The accelerometer of the six-axis sensor data detects an impact of 9.8 m / s², and after conforming to the fall characteristics, it enters a stationary state; The barometer detects an altitude change rate ≥ 5 m / s, possibly indicating a landslide / cliff fall; Network connection status: No available cellular network for 3 consecutive minutes, with a signal strength < -120 dBm; Device status: Battery level ≤ 5% and temperature sensor ≥ 45 °C; User operation: Press the power button quickly three times in a row, sending an emergency SOS signal; Judgment threshold: Sensor anomalies, impact or sudden barometric changes, and network interruption result in an emergency mode confidence level ≥ threshold.

[0022] The decision-making module includes a dynamic policy sub-module, a scenario classification sub-module, and a security verification module. The dynamic policy sub-module is equipped with a multi-objective optimization model and a policy knowledge base. The scenario classification sub-module is equipped with a LightGBM learning model, a DBSCAN clustering algorithm, and an LSTM time series prediction; The execution module includes a seamless switching sub-module, a SIM card management unit, and a protocol adaptation gateway. The seamless switching sub-module is equipped with MP-TCP multi-path transmission and pre-carrier aggregation technology. The SIM card management unit includes a physical dual SIM card slot and an eSIM controller; The optimization module includes a federated learning framework, a power management unit, and a network heat map service; In the present invention, a LightGBM-DBSCAN hybrid model is adopted. LightGBM quickly classifies known scenarios (accuracy rate of 92%), DBSCAN clustering discovers abnormal network states (such as sudden congestion), LSTM predicts network requirements in the next 15 minutes (such as reserving bandwidth before a download task), stores more than 200 preset rules, and updates them once per hour through federated learning. MP-TCP multi-path transmission connects 5G and Wi-Fi networks simultaneously, and the data stream is seamlessly transferred during switching (packet loss rate < 0.1%). The protocol adaptation gateway automatically converts IPv4 / IPv6 / SRv6 protocols and supports enterprise intranet penetration (NAT conversion success rate of 99%). Specialized hardware acceleration uses a network coprocessor to independently process the TCP / IP protocol stack, releasing 30% of the main CPU load. The eSIM controller supports the concurrent activation of 8 Profiles, and the switching time is < 0.3 seconds; Through this mechanism, when first accessing 6GHz Wi-Fi, the frequency band optimization is completed within 30 seconds (the traditional solution takes 5 minutes), and cross-network service seamless switching can be achieved. When switching between an enterprise VPN and the public network, the OA system login time is reduced from 8 seconds to 0.5 seconds, improving the device fluency.

[0023] The security verification module includes a digital certificate verification unit, a blockchain light node, and a differential privacy processing unit; The digital certificate verification unit supports X.509 certificate chain verification and SM2 / SM9 national cryptographic algorithms. The blockchain light node stores the base station fingerprint hash value and supports PBFT consensus verification. The differential privacy processing unit adds Laplace noise to user trajectory data, with ε = 0.1; A fake base station disguises itself as a legitimate signal (such as forging a bank customer service number), induces users to connect and then steals information, resulting in the leakage of user privacy. The operator can track the precise location (error < 10 meters), posing risks of commercial espionage or personal safety; In the present invention, a trusted network environment is constructed through a dual mechanism of blockchain base station verification and national cryptography algorithm. The physical characteristics of each base station (such as signal frequency point, coding mode, geographical location) generate a unique digital fingerprint and are stored in the blockchain distributed ledger. Before a device connects, the legitimacy of the base station needs to be verified through the PBFT consensus algorithm, effectively intercepting attacks from forged base stations, suppressing the connection success rate of fake base stations from the industry average of 18% to below 0.3%, and fundamentally eliminating the risks of SMS phishing and information theft. Regarding the hidden danger of user privacy leakage, the SM9 national cryptography algorithm is used to encrypt communication content end-to-end. The computational effort required for cracking reaches the order of 10^6 times that of the traditional RSA algorithm. Even if the data is intercepted during transmission, it is difficult for hackers to complete decryption within a reasonable time. At the same time, differential privacy technology adds a random noise perturbation of ±50 meters to user location data, ensuring that external attackers can only obtain fuzzy area information (such as "within a radius of 500 meters of a certain commercial area"), and cannot accurately locate specific buildings or floors. This design reduces the risk of location information leakage by 83% and meets the requirements of strict privacy regulations such as the EU's GDPR.

[0024] The multi-source signal perception sub-module includes at least a 2.4GHz / 5GHz / 6GHz tri-band Wi-Fi scanning circuit and a cellular signal quality analysis unit; The 2.4GHz / 5GHz / 6GHz tri-band Wi-Fi scanning circuit supports the 802.11k / v / r protocols, and the cellular signal quality analysis unit can measure the SS-signal strength / SS-RSRQ parameters of 5G NR; Old devices only support the 2.4 / 5GHz bands and cannot utilize the ultra-high bandwidth of 6GHz (theoretical rate of 9.6Gbps), resulting in incomplete Wi-Fi coverage. They only measure the signal strength (RSRP) and ignore the network load (RSRQ), causing an incomplete 5G assessment and leading to the selection of congested base stations; In the present invention, the performance and stability of wireless networks are significantly improved through innovative frequency band expansion and multi-dimensional network evaluation strategies. The three-band Wi-Fi scanning module designed in this claim supports 2.4GHz, 5GHz and 6GHz frequency bands at the same time, of which the 6GHz frequency band is dedicated to high-bandwidth services such as VR streaming media and 8K video transmission. It can achieve a peak rate of 4.8Gbps at a bandwidth of 160MHz, which is more than 300% higher than the traditional dual-band solution. In addition, through the intelligent frequency band allocation algorithm (such as 2.4GHz prioritizes the connection of smart home devices, and 6GHz is exclusively allocated to high-traffic applications), network congestion caused by competition among multiple devices can be effectively avoided. In order to optimize the selection of 5G networks, a dual-parameter joint evaluation mechanism of SS-RSRP (signal strength) and SS-RSRQ (network quality) is introduced. When it is detected that the signal strength of a certain base station meets the standard (RSRP>-95dBm) but the network load is too high (RSRQ<-10dB), it automatically switches to an adjacent base station with a lighter load. This strategy has reduced the number of video streaming freezes from 5.2 times per hour to 0.3 times, and the game delay fluctuation range has narrowed from 70-120ms to 35-50ms, achieving an e-sports level network experience. In addition, dynamic frequency band switching technology can seamlessly transition to the 5G millimeter wave network when the 6GHz frequency band is congested, ensuring business continuity in extreme scenarios. For example, in a concert with 10,000 people in a stadium, users can still smoothly push 8K live streaming with a frame loss rate of less than 0.1%.

[0025] Scene recognition has a scene determination formula: Total score = Σ(weight of S sensor data × normalized value); If the total score is ≥ the threshold, the corresponding scenario strategy is activated, that is, by comprehensively calculating the weights and normalized values ​​of each sensor data, the total score includes at least: home mode confidence, office mode confidence, travel mode confidence and emergency mode confidence. When the total score reaches the preset threshold, the network connection solution that best suits the current scenario is automatically started; Scene judgment has a misjudgment elimination mechanism: Home mode to prevent misjudgment: Exclusion conditions: Being at home for 8 consecutive hours from 10:00 to 17:00 on weekdays, which may be working from home, requires a second confirmation; Verification method: Check VPN connection status and usage time of document apps; Office mode to prevent misjudgment: Exclusion conditions: The device is detected within the enterprise coordinate range on non-working days (such as weekends / holidays), is charging but not connected to the enterprise VPN, and has no document app operation records; Verification method: Check the percentage of time that office software such as Outlook / Teams is running in the foreground, verify whether it accesses corporate intranet resources, and pop up a prompt box to ask the user whether to enter the office mode; Travel mode prevents misjudgment: Exclusion conditions: speed > 80 km / h in subway tunnels but no LAC change; Verification method: Positioning combined with map subway line calibration; Emergency mode to prevent misjudgment: Exclusion conditions: The roller coaster scene in the amusement park is identified by the gyroscope angular velocity feature; Verification method: If no voice confirmation is received within 30 seconds, the alarm will be automatically cancelled; In the present invention, when working from home, the VPN connection and the usage time of the document software are detected (such as continuous use of WPS for > 1 hour). In the roller coaster scene, the rotation characteristics are identified by the gyroscope (angular velocity > 200° / s) to exclude the situation of falling off a cliff in a straight line. The user confirms the abnormal scene pop-up prompt for a second time (such as "violent shaking is detected, do you want to enable emergency mode?"). If there is no response within 30 seconds, it is automatically cancelled, so that accurate pattern matching can be achieved. The misjudgment rate of working from home is reduced from 25% to 5%, ensuring smooth remote meetings and preventing false alarms. The probability of false triggering of rescue in the roller coaster scene is reduced from 15 times / month to 1 time / year.

[0026] The optimal tariff selection algorithm includes establishing a linear programming model: Min Σ(C_i × D_i), with the constraint D_i≤ Q_i, where C_i is the tariff unit price of the i-th operator, D_i is the allocated traffic, and Q_i is the remaining traffic of the package; The simplex method is used to solve the optimal flow distribution plan, and the calculation time is less than 10ms; In the present invention, by performing real-time tariff calculation for the module and establishing a tariff model (unit price × estimated traffic), the most cost-saving combination is calculated within 10 milliseconds, temporary packages are automatically ordered, and dual cards are intelligently allocated, such as video traffic using high-speed cards and WeChat messages using low-priced cards, saving an average of 35% of traffic fees per month; making the tariff transparent and controllable, while making full use of traffic, and increasing the utilization rate of dual cards from 60% to 95%, avoiding package waste.

[0027] Step 5 includes base station fingerprint verification, comparing the consistency of the target base station PCI / ECGI with the blockchain registration information; Signal integrity detection, verifying the channel impulse response characteristics of the PSS / SSS synchronization signal; When the risk score is > 0.7, an alarm is triggered and the network is switched to the preset safety network; In the present invention, by verifying the base station fingerprint to compare the base station signal characteristics (such as channel impulse response), if it is inconsistent with the blockchain record, an alarm will be triggered. High-risk base stations are automatically added to the blacklist and shared synchronously by all devices in the network. Sensitive data (such as bank APPs) is forced to use the national cryptographic SM9 algorithm for encryption, which cannot be cracked even if intercepted, so as to prevent SMS theft. The success rate of phishing attacks by fake base stations has dropped from 18% to 0.3%. The data is kept confidential throughout the process, the encryption strength during transmission is increased by 10 times, and the time required for cracking has increased from 1 hour to 10 years.

[0028] The mobile device includes at least one computer-readable storage medium storing program instructions. The storage medium is an embedded memory in the specifications of eMMC5.1 and UFS 3.1. In the present invention, the high-speed storage medium adopts a UFS 3.1 chip, the policy loading time is shortened to 100 ms, the data compression algorithm is optimized, the storage space occupancy is reduced by 60%, and the intelligent write policy is adopted. Non-critical data is cached to the memory, and there is only one batch write per day, and the number of erase-write cycles is reduced by 80%, so as to quickly respond to demands. When entering and leaving the subway station, the network switching preparation time is reduced from 3 seconds to 0.5 seconds, and the device life is extended. The service life of the storage chip is extended from 1 year to 5 years.

[0029] The scene classification sub-module identifies the user's current scene in the following way: Step 1. Adopt the dynamic weight fusion formula: The scene feature value is the satellite positioning weight * positioning accuracy + Wi-Fi fingerprint weight * signal feature + motion state weight * movement intensity; Among them, the sum of the satellite positioning weight, Wi-Fi fingerprint weight, and motion state weight is equal to 1, and the user can adjust the ratio. In the present invention, for dynamic weight adjustment, when the GPS signal is weak, the Wi-Fi positioning weight is automatically increased (the proportion is adjusted from 30% to 70%). An open user setting interface is provided, and preferences such as "network speed priority" and "cost-saving mode" can be manually allocated to facilitate adaptation to complex environments. The positioning accuracy in urban canyons is improved from 50 meters to 10 meters, the average monthly fee of student users is reduced by 40%, and the video loading speed of business users is increased by 3 times.

[0030] The dynamic policy sub-module selects the optimal network in the following way: Give priority to networks that meet the following conditions: The comprehensive score = energy consumption weight * network power consumption + tariff weight * traffic cost + latency weight * response time; Among them, the sum of the energy consumption weight, tariff weight, and latency weight is 1, which is dynamically adjusted according to the user's set preferences. Mandatory guarantee conditions: The signal strength of the selected network needs to be higher than -85 dBm, which is equivalent to the mobile phone signal strength ≥ 2 bars; Network stability must meet the requirement that the packet loss rate is less than 2%; When multiple eligible networks are detected, the option with the highest combined score is automatically selected.

[0031] Embodiment 2: See also Figure 1 , this embodiment provides the following technical solutions: the staff uses the method disclosed in the present invention to verify, and installs the system of the present invention on the smart phone of a colleague in the department. The colleague flew from Beijing to Berlin to attend an international exhibition in June. Before the flight landed, the perception module detected a rapid drop in altitude (500 meters per minute) through the barometer, and combined with GNSS positioning to confirm that it was about to arrive at Berlin Tegel Airport, the decision module triggered the "cross-border roaming mode", and the tariff optimization engine compared the local operator packages in real time: Vodafone (€15 / 10GB), Deutsche Telekom (€20 / 15GB), O2 (€12 / 8GB), based on the user's preset "price-performance priority" strategy, selected O2's €12 package and passed the GSMA The SGP.32 protocol activates the eSIM in seconds, and the execution module simultaneously establishes an MP-TCP dual-channel connection. The main channel uses O2's 5G network to carry the live streaming of the exhibition (20Mbps bit rate), and the backup channel retains the original SIM card on standby to receive domestic emergency text messages. The security verification module verifies the fingerprint of the local base station through the blockchain light node and intercepts two pseudo base stations that are forged as "China Mobile roaming cooperation stations". The optimization module records that this switching took 0.8 seconds and saved €183 in roaming fees, and learns user preferences: video transmission was used frequently during the exhibition, and dedicated bandwidth channels were automatically reserved later. When leaving the exhibition hall, the perception module detected that the user took a taxi (GNSS speed 65km / h), and the decision module enabled the navigation dedicated network. Even if passing through an underground tunnel (cellular signal loss), the positioning was quickly restored through the pre-cached subway base station information, and the navigation deviation throughout the journey was less than 5 meters.

[0032] Embodiment 3: See also Figure 1 In this embodiment, the staff drove from Hebei to Chongqing. The smart device used for navigation was equipped with the patented system and passed through the dense tunnel group of the Wulong section of the Baomao Expressway. When the vehicle set out from Hebei, the system had preloaded the base station map and tunnel 3D model of the Chongqing area according to the itinerary plan. When the vehicle approached the Baimashan Tunnel (total length of 6.2 kilometers), the perception module matched the GNSS positioning with the high-precision map and triggered the "tunnel prediction mode" 1 kilometer in advance: the network coprocessor pre-activated the dual SIM cards of China Mobile and China Telecom, started the MP-TCP dual-link connection (primary card 5G NSA, backup card 4G VoLTE), and cached the Baidu map offline navigation data (including the lane-level 3D model in the tunnel). When entering the tunnel, the GPS signal is lost, and the system immediately switches to multi-sensor fusion positioning: The six-axis IMU calculates the track through the wheel speed (continuous acceleration of 0.3g) and the steering wheel angle (gyroscope angular velocity of 15° / s). The lidar scans the feature points on the tunnel wall for auxiliary correction, and the positioning error is controlled within 0.5 meters. At the same time, the cellular signal analysis unit detects that the RSRP of the main card drops sharply to -110 dBm, and the decision-making module completes the following operations within 200 ms: ① Switch to the China Telecom 4G network (private network coverage in the tunnel, RSRP = -92 dBm); ② Enable the Beidou RDSS satellite short message function (send encrypted location messages to the cloud server every 30 seconds); ③ Limit the bandwidth of the entertainment system (the video APP is downgraded to 480P), and prioritize navigation and emergency communication. Suddenly, there is a mobile network fluctuation in the middle section of the tunnel (the base station handover interval is shortened to 45 seconds), and the execution module starts the "resilient transmission mode": The navigation data is packetized and transmitted concurrently through the 5G, 4G, and satellite three channels. The forward error correction (FEC) technology is used to reconstruct the data packets to ensure that the navigation instruction delay is less than 800 ms in a complex electromagnetic environment. When exiting the tunnel, the system captures the sudden change in brightness (rising from 5 lux to 2000 lux within 0.3 seconds) through the optical sensor, immediately resumes the 5G main link, and uploads the full network quality log to the optimization module. When passing through the 1.2-kilometer-long Fairy Mountain Tunnel later, the system optimizes the strategy based on historical learning results: ① Download the base station fingerprint library in the tunnel 500 meters in advance, and compress the handover time from 0.8 seconds to 0.3 seconds; ② Dynamically adjust the satellite message sending interval to 2 minutes (saving 40% power); ③ Activate the exemption from the bandwidth limit of the passenger entertainment system (recognize the need for children to watch cartoons). Among the 28 tunnels in the whole journey, the total navigation interruption time is reduced from 46 minutes in the traditional scheme to 1.2 minutes, and the traffic cost is saved by 57%.

[0033] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for automatically switching network channels and SIM cards according to signal strength, which is applicable to network optimization management of Internet of Things terminals, and is characterized by: The method includes a perception module, a decision module, an execution module and an optimization module, which jointly realize automatic switching of network channels and SIM cards through a progressive collaboration process; The perception module is used for data collection and identification of user scenario patterns, wherein data collection includes collection of environmental feature data and device status data; The decision module is used for intelligent analysis, selecting the optimal network access solution according to the user scenario type, and ensuring the security of network switching; The execution module is used for network control, and performs seamless switching of network channels through the cooperation of hardware and software; The optimization module is used for self-learning of the decision-making module, establishes a database and learns according to user behavior, and recommends the best network access solution to users in similar scenarios, while assisting the IoT terminal in power control; The perception module includes a multi-source signal perception submodule, a behavior perception submodule and an environment perception submodule. The multi-source signal perception submodule includes at least a GNSS positioning unit and a multi-band RSSI scanning unit. The behavior perception submodule is equipped with a pre-trained machine learning model. The environment perception submodule includes at least a six-axis motion sensor, a light sensor and a barometer. The perception module, decision module, execution module and optimization module are all interconnected through a bus, and the method for automatically switching network channels and SIM cards performed by the perception module, decision module, execution module and optimization module includes the following steps: Step 1. Collect IoT terminal status data through the multi-source signal perception module: including cellular signal quality, Wi-Fi channel utilization, GPS positioning, network connection status, IoT terminal status, mobile status and network time of the IoT terminal; the behavior perception submodule collects user behavior data: including application behavior and user operation behavior data; the environment perception submodule collects environmental features: including light intensity collected by the light sensor and six-axis sensor data, the six-axis sensor data at least includes acceleration collected by the accelerometer and angular velocity collected by the gyroscope, and the six-axis sensor data can be integrated to obtain the device posture; Step 2. The scene classification submodule of the decision module outputs the scene classification result based on the environmental feature data. The classification process is based on the user's preset model and the device's default scene classification. The scene classification includes multiple scene types: home mode, office mode, high-speed travel mode and emergency rescue mode, and the target network is selected according to the different scene types; Step 3. The execution module selects the target network according to the scenario type, including: enabling the WiFi 6 optimization strategy for the home mode, forcibly shutting down the 5G radio frequency, automatically enabling the backup SIM card as a redundant channel if the home Wi-Fi signal strength is detected to be lower than -70dBm, and limiting the frequency of background application network requests to ≤1 time / minute to extend the device life; in the office mode, activating dual SIM card aggregation transmission, superimposing bandwidth to a maximum of 3.6Gbps, forcibly enabling the enterprise-level VPN tunnel, shielding untrusted public Wi-Fi connections, and allocating dedicated QoS channels for video conferencing applications; in the high-speed travel mode, activating the local eSIM profile, executing the real-time tariff optimization algorithm, enabling the navigation dedicated network channel, ensuring the real-time update of map data, and starting the pre-caching mechanism when it is detected that the base station switching interval is <90 seconds; in the emergency scenario, forcibly enabling the satellite communication module, establishing an emergency communication channel, starting the minimalist communication protocol, downgrading voice calls to AMR-NB 4.75kbps encoding, and using LoRa modulation for data transmission, triggering the device self-protection mechanism: that is, turning off non-essential sensors and allocating the remaining power to the device's positioning and communication modules first; Step 4. The network coprocessor establishes an MPTCP dual-channel connection and performs seamless switching. Step 5. The decision module verifies the credibility of the target network and rolls back to the safe network if the verification fails. Step 6. The optimization module updates the federated learning model and records the performance indicators of this switch; The scene classification result mentioned in step 2 is obtained by comparing the environmental feature data with the sensor data, and includes at least the following modes: Scenario 1: Home mode; Identifying features: GPS positioning: Continuously located within a radius of ≤50 meters of the pre-registered home coordinates; Network connection status: Stable connection to home router SSID, signal strength RSSI ≥ -60dBm; Mobile state: The accelerometer data variance of the six-axis sensor data is <0.1m / s², and the static state lasts ≥2 hours; Time: The occurrence period is concentrated between 18:00-08:00, local time zone; Environmental characteristics: The ambient light sensor detects regular day-night changes, and the light intensity fluctuates between 5-300 lux; Judgment threshold: The home mode confidence level of GPS must be ≥ the threshold value when it is in the home area, connected to Wi-Fi, stationary, and at night. Scenario 2: Office mode; Identifying features: GPS positioning: located within the coordinate range of the enterprise's electronic fence, with an error of ≤20 meters; Network connection status: connected to the enterprise VPN or 802.1X authenticated Wi-Fi network; Movement status: The accelerometer of the six-axis sensor data detected periodic micro-movements with a variance of 0.3-1.2m / s², which is consistent with office activities; Time: Continuously appears from 09:00 to 18:00 on weekdays; Device status: The device is in charging state, the current is ≥1.5A and the screen is on for >70% of the time; Decision threshold: Working day time period, enterprise coordinates, and enterprise network connection to obtain office mode confidence ≥ threshold value; Scenario 3: High-speed travel mode; Identifying features: Mobile status: GNSS measurement speed is continuously ≥ 60km / h, with an error of ±5%; Network connection status: The base station area where the mobile phone signal is located changes more than 3 times per hour, and the signal strength of the currently connected main base station is lower than -100dBm. At the same time, the connected base station is frequently switched, and the signal switching occurs once every 90 seconds on average; Device posture: The gyroscope of the six-axis sensor data detects continuous horizontal acceleration, which is consistent with the movement characteristics of the vehicle; Application behavior: Navigation apps are continuously running in the foreground; Decision threshold: Speed ​​≥ 60km / h, LAC changes frequently, navigation APP is active, and the travel mode confidence ≥ threshold value; Scenario 4: Emergency rescue mode; Identifying features: Device posture: The accelerometer of the six-axis sensor data detected an impact of 9.8m / s², which meets the characteristics of a fall and then enters a state of stillness; the barometer detects an altitude change rate of ≥5m / s, which may indicate a landslide / fall off a cliff; Network connection status: No cellular network available for 3 consecutive minutes, signal strength <-120dBm; Device status: power ≤ 5% and temperature sensor ≥ 45°C; User operation: Press the power button three times quickly to send an emergency SOS signal; Judgment threshold: sensor abnormality, impact or sudden change in air pressure, network interruption, emergency mode confidence ≥ threshold.

2. The method of automatically switching network channels and SIM cards according to signal strength according to claim 1, characterized in that: The decision-making module includes a dynamic strategy submodule, a scenario classification submodule and a safety verification module. The dynamic strategy submodule is equipped with a multi-objective optimization model and a strategy knowledge base, and the scenario classification submodule is equipped with a LightGBM learning model, a DBSCAN clustering algorithm and LSTM time series prediction; The execution module includes a seamless switching submodule, a SIM card management unit and a protocol adaptation gateway, the seamless switching submodule is equipped with MP-TCP multipath transmission and pre-carrier aggregation technology, and the SIM card management unit includes a physical dual card slot and an eSIM controller; The optimization module includes a federated learning framework, a power management unit, and a network heat map service; The GNSS positioning unit supports GPS L1 / L5 and Beidou B1I / B2a dual-frequency positioning, the eSIM controller is compatible with the GSMASGP.32 remote configuration protocol, and a dedicated DMA channel is established between the network coprocessor of the execution module and the baseband chip.

3. The method of automatically switching network channels and SIM cards according to signal strength according to claim 1, characterized in that: The security verification module includes a digital certificate verification unit, a blockchain light node and a differential privacy processing unit; The digital certificate verification unit supports X.509 certificate chain verification and SM2 / SM9 national encryption algorithm. The blockchain light node stores the base station fingerprint hash value and supports PBFT consensus verification. The differential privacy processing unit adds Laplace noise to the user trajectory data, ε=0.

1.

4. The method of automatically switching network channels and SIM cards according to signal strength according to claim 1, characterized in that: The multi-source signal sensing submodule includes at least a 2.4GHz / 5GHz / 6GHz tri-band Wi-Fi scanning circuit and a cellular signal quality analysis unit; The 2.4GHz / 5GHz / 6GHz tri-band Wi-Fi scanning circuit supports 802.11k / v / r protocols, and the cellular signal quality analysis unit can measure the SS-signal strength / SS-RSRQ parameters of 5G NR.

5. The method of automatically switching network channels and SIM cards according to signal strength according to claim 1, characterized in that: The scene recognition has a scene determination formula: Total score = Σ(weight of S sensor data × normalized value); If the total score is ≥ the threshold, the corresponding scenario strategy is activated, that is, by comprehensively calculating the weights and normalized values ​​of each sensor data, the total score includes at least: home mode confidence, office mode confidence, travel mode confidence and emergency mode confidence. When the total score reaches the preset threshold, the network connection solution that best suits the current scenario is automatically started; The scene judgment has a misjudgment elimination mechanism: Home mode to prevent misjudgment: Exclusion conditions: Being at home for 8 consecutive hours from 10:00 to 17:00 on weekdays, which may be working from home, requires a second confirmation; Verification method: Check VPN connection status and usage time of document apps; Office mode to prevent misjudgment: Exclusion conditions: The device is detected within the enterprise coordinate range on non-working days (such as weekends / holidays), is charging but not connected to the enterprise VPN, and has no document app operation records; Verification method: Check the percentage of time that office software such as Outlook / Teams is running in the foreground, verify whether it accesses corporate intranet resources, and pop up a prompt box to ask the user whether to enter the office mode; Travel mode prevents misjudgment: Exclusion conditions: speed > 80 km / h in subway tunnels but no LAC change; Verification method: Positioning combined with map subway line calibration; Emergency mode to prevent misjudgment: Exclusion conditions: The roller coaster scene in the amusement park is identified by the gyroscope angular velocity feature; Verification method: If no voice confirmation is received within 30 seconds, the alarm will be automatically cancelled.

6. The method of automatically switching network channels and SIM cards according to signal strength according to claim 5, characterized in that: The optimal tariff selection algorithm includes establishing a linear programming model: Min Σ(C_i × D_i), with the constraint that D_i ≤ Q_i, where C_i is the tariff unit price of the i-th operator, D_i is the allocated traffic, and Q_i is the remaining traffic of the package; The simplex method is used to solve the optimal flow distribution solution, and the calculation time is <10ms.

7. The method of automatically switching network channels and SIM cards according to signal strength according to claim 5, characterized in that: The step 5 includes base station fingerprint verification, comparing the consistency of the target base station PCI / ECGI with the blockchain registration information; Signal integrity detection, verifying the channel impulse response characteristics of the PSS / SSS synchronization signal; When the risk score is > 0.7, an alarm is triggered and the system switches to the preset safety network.

8. The method of automatically switching network channels and SIM cards according to signal strength according to claim 1, characterized in that: The mobile device includes at least one computer-readable storage medium storing program instructions, and the storage medium is an embedded memory of eMMC 5.1 and UFS 3.1 specifications.

9. The method of automatically switching network channels and SIM cards according to signal strength according to claim 1, characterized in that: The scene classification submodule identifies the scene where the user is located by: Step 1. Use the dynamic weight fusion formula: The scene feature value is satellite positioning weight * positioning accuracy + Wi-Fi fingerprint weight * signal feature + motion state weight * movement intensity; The sum of the satellite positioning weight, Wi-Fi fingerprint weight, and motion status weight is equal to 1, and the user can adjust the ratio.

10. The method of automatically switching network channels and SIM cards according to signal strength according to claim 1, characterized in that: The dynamic strategy submodule selects the optimal network in the following way: Priority is given to networks that meet the following conditions: Comprehensive score = energy consumption weight * network power consumption + tariff weight * traffic cost + delay weight * response time; The sum of the energy consumption weight, tariff weight, and delay weight is 1, which is dynamically adjusted according to the user's setting preferences; Mandatory guarantee conditions: The selected network signal strength must be higher than -85dBm, which is equivalent to a mobile phone signal grid ≥ 2 grids; Network stability must meet the requirement that the packet loss rate is less than 2%; When multiple eligible networks are detected, the option with the highest combined score is automatically selected.

Images