Bidirectional short message communication method used in network-free scene

Abstract

The invention discloses a bidirectional short message communication method used in a network-free scene, relates to the technical field of communication, and is used for solving the problem of poor communication in a network disconnection scene. Through multi-source environment perception and link state fusion modeling, real-time evaluation of communication environment accessibility is realized, the stability of a public network link is dynamically judged, and support is provided for channel switching; a scheduling scoring model is constructed in combination with a link state and task characteristics, so that the terminal preferentially transmits a high-emergency and high-adaptation task under a link limited condition, and the resource utilization efficiency is improved; a multi-time-scale risk integral mechanism is introduced, the recognition capability of sudden states such as falling, static and air pressure sudden change is improved, an emergency short message is automatically generated without human intervention, and quick uploading of SOS information and position is realized, so that the stability, safety and adaptability of communication in a network-free scene are improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and more particularly, to a two-way short message communication method for scenarios without network. Background Art

[0002] In extreme operating environments such as daily power grid inspections, mountain area operation and maintenance, and sudden natural disasters, public network communication infrastructure is often completely unavailable due to geographical occlusion, signal coverage blind spots, or sudden interruptions, resulting in the inability of operation terminals to upload operation status, feedback location information, and transmit alarm events at critical moments, seriously threatening the closed-loop execution of dispatching instructions and the safety of personnel operations. Especially in areas without public network signal coverage such as mountain cliffs, tunnels, and forest areas, traditional communication mechanisms based on cellular networks completely fail and cannot provide a basic two-way information transmission path.

[0003] Deficiencies of the prior art: The Beidou satellite navigation system has a short message communication function, providing a feasible solution for information transmission in environments without public networks. However, most existing short message communication technologies rely on manual triggering or fixed-period reporting, lacking the ability to adaptively perceive environmental changes and an automatic triggering mechanism for emergency states, and are difficult to cope with problems such as "communication preemption", "status overflow", and "link delay" in emergencies. At the same time, existing solutions generally lack a joint evaluation and analysis strategy for communication link status and communication tasks, resulting in problems such as inaccurate communication resource scheduling, unreasonable priority allocation, and untimely response to emergency events. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the following solutions are provided to solve the problem of poor communication in the network interruption scenario in the above-mentioned background art.

[0005] To achieve the above object, the present invention provides the following technical solutions: A two-way short message communication method for scenarios without network, including the following steps: Based on multi-source sensor data integrated by the terminal, construct a perturbation tensor structure, generate a communication reachability score through non-linear mapping, and detect sudden changes in the communication environment in combination with a sliding window; Construct a communication state vector according to the 4G link activity score factor, and execute an asymmetric hysteresis switching strategy according to the combined judgment result of the communication reachability score and the 4G link activity score to complete the primary and secondary channel switching between the public network communication channel and the Beidou short message channel; Extract an attribute vector for the task to be scheduled, construct a non-linear priority scoring function, generate a task scheduling subset under communication resource constraints, and perform structure mapping and format compression processing according to the primary channel type to form a transmission task packet adaptable to the link; Continuously monitor risk factors and perform risk scoring, combine multi-time scale integration windows to determine high-confidence abnormal states in terms of risk trends. When a high-confidence abnormal state is confirmed, generate an emergency short message with the smallest structure and give it priority transmission via the Beidou link.

[0006] In a preferred embodiment, based on the multi-source sensor data integrated by the terminal, construct a perturbation tensor structure. The specific steps include: Collect on-site environmental information through the sensor modules integrated by the terminal. The sensors used include barometric pressure sensors, three-axis accelerometers, geomagnetic sensors, gyroscopes, light sensors, infrared or ultrasonic ranging sensors; Perform mapping processing of the sensing data output by each type of sensor to a unified scale to obtain an environmental standardized feature vector; When the terminal moves, rotates rapidly in attitude, or senses abnormal magnetic disturbances in the spatial direction, use the acceleration magnitude as a perturbation factor. At the same time, use the gradient change of the geomagnetic vector to represent the magnetic space perturbation intensity, and combine the attitude angle change to form a three-dimensional cross mapping to construct a three-dimensional perturbation tensor structure for capturing the combined perturbation trend; In the three-dimensional perturbation tensor structure, each element value is determined by the non-linear combination of three types of perturbation factors. The acceleration factor is placed in the denominator as a perturbation excitation term to enhance the mutation behavior with an exponential function. The attitude angle change and the magnetic space perturbation intensity term are placed in the numerator as weight control terms.

[0007] In a preferred embodiment, generate a communication reachability score through non-linear mapping, and combine a sliding window to detect mutation behaviors in the communication environment. The specific steps include: After completing the construction of the environmental standardized feature vector and the perturbation structure tensor, fuse and map these two types of data into a communication link reachability index space; For each sensor eigenvalue, introduce a corresponding response adjustment factor and use the hyperbolic tangent function form in the saturation interval for mapping; Extract the perturbation extreme value in the perturbation structure tensor and use it as a global perturbation term. Combine it with the sensor eigenvalue term to construct the total input value of the scoring function; The perturbation extreme value is input as a perturbation sensitivity factor, dominating the low-value interval of the scoring function; Input the combined result into a non-linear boundary-preserving function, compress and map the linear superposition result into the interval [0, 1], and output the communication reachability scoring factor; Use a time-series sliding window mechanism to maintain a scoring record window of a fixed length inside the terminal. Each time a new communication reachability score is received, calculate the difference with the communication reachability score result at the previous moment in the window. If the current score drops by more than the set drop threshold, generate a communication mutation flag bit and regard it as a communication environment mutation behavior.

[0008] In a preferred embodiment, construct a communication state vector according to the 4G link activity scoring factor, and execute an asymmetric hysteresis switching strategy according to the combined judgment result of the communication reachability score and the 4G link activity score. The specific steps include: Use the communication reachability score as the environmental factor for channel switching, and introduce the 4G link activity scoring factor as the communication state vector to evaluate signal stability and position perturbation trend; The 4G link activity score is used to represent the dynamic communication state of the current public network link. The signal stability of the 4G link activity score is obtained by continuously recording the change of the 4G signal strength received by the terminal within a given monitoring period and calculating the drop rate of the received power per unit time. According to the continuous position point sequence of the terminal positioning module, calculate the spatial displacement amplitude within the same period to obtain the position perturbation trend of the terminal in the physical space, and use a weighted combination structure to construct the 4G link activity scoring function; Combine the two indicators of the communication reachability score and the 4G link activity score to form a link state vector. The communication reachability score reflects the physical feasibility of the environment for the link, while the 4G link activity score reflects the signal behavior stability of the network itself; Set a segmented hysteresis type switching strategy according to the link state vector.

[0009] In a preferred embodiment, complete the primary and backup channel switching between the public network communication channel and the Beidou short message channel, including the following steps: When the communication reachability score exceeds 0.8, it is determined to be in a high reach state, and the current 4G main link communication is maintained; when the communication reachability score is lower than 0.5, it is determined to be in a low reach state, and the channel switching strategy is triggered to switch to the Beidou short message link; when the communication reachability score is less than or equal to 0.8 and greater than or equal to 0.5, it is in the switching hysteresis state, and the switch is not performed immediately, and it enters the monitoring buffer period; The 4G link activity score is used as an auxiliary judgment basis. If the 4G link activity score shows a negative growth trend, trigger the channel switching strategy and enter the standby channel in advance; After switching to the standby channel, adaptively adjust the structure of the current communication task; Construct a communication task structure mapping mechanism, and classify the current tasks to be sent into type-one tasks, type-two tasks, and type-three tasks according to their priorities; When switching to the Beidou link, extract the necessary fields in a type of task to form a structure. The necessary fields include the terminal number, longitude and latitude position, alarm type identifier, timestamp, and scene feature encoding, to avoid exceeding the short message capacity limit.

[0010] In a preferred embodiment, extract the attribute vector for the task to be scheduled and construct a non-linear priority scoring function, including the following steps: Abstract the task structure into an attribute vector, which includes the task generation time, maximum allowable delay time, data volume size, task urgency level, and link reachability score; After the task attribute vector is constructed, perform task scheduling sorting and construct a non-linear priority scoring function; The priority scoring function is used to quantify the scheduling value of each task under the current communication state and is calculated through joint modeling of task attributes and network status; Extract the urgency level, generation time, maximum allowable delay time, and communication reachability score of each task to be scheduled; Use a non-linear function to construct a time modulation term. Taking the difference between the task generation time and the current system time as the input, multiply it by the adjustment parameter and use it as the variable of the exponential function to generate the time effect score of the task; Evaluate the transmittability of the task under the link condition, multiply the maximum allowable delay time of the task by the link reachability score, and input the result into the non-linear response function to map it to the transmittability score, which reflects whether the task can be transmitted currently; Structurally combine the time effect score and the transmission feasibility score, and output the normalized score value as the priority score through the boundary compression function, which is limited between 0 and 1.

[0011] In a preferred embodiment, generate a task scheduling subset under communication resource constraints, and perform structure mapping and format compression processing according to the main channel type to form a transmission task packet adaptable to the link, including the following steps: Set the available transmission capacity threshold for the current scheduling period according to the main communication channel type. The available transmission capacity threshold is the upper bound of the scheduling window, which reflects the total amount of data that can be transmitted per unit time under the current link condition. The available transmission capacity threshold under the 4G link can be given according to the real-time bandwidth evaluation result, and it is the maximum fixed message length under the Beidou short message link; Multiply the bandwidth capacity identifier of the current communication channel by the scheduling period time length to obtain the allocable communication budget as the transmission capacity; The bandwidth capacity identifier is the 4G network rate estimate value or the maximum load of the Beidou short message, and the scheduling period time length is the maximum transmission time available for the current communication scheduling; Sort the calculated task priority score results in descending order to form a task waiting to be scheduled queue; sequentially select the task with the highest score value from the head of the queue, accumulate the data volume until the accumulated total exceeds the current window capacity limit, then stop task selection, and form the subset of the current scheduling period as the task scheduling subset; Identify the current main communication channel type to form a transmission task packet with an adaptable link. If it is a 4G public network link, no compression mapping processing is required; If the current main channel is a Beidou short message link, schedule tasks for structure compression and field reduction processing to form a transmission task packet. The field reduction processing includes field screening and reconstruction, field encoding compression, field alignment and byte control.

[0012] In a preferred embodiment, continuously monitor risk factors and perform risk scoring, and combine multi-time scale integration windows to determine a high-confidence abnormal state through risk trend determination, including the following steps: Construct abnormal detection factors for different risk scenarios, construct and process a non-linear combined risk scoring function for the abnormal detection factors. The risk scoring function is used to evaluate the potential abnormal state level of the terminal in the current environment; First perform a normalization operation on all abnormal detection factors, and use the hyperbolic tangent function for non-linear amplification. Combine the output values of all factors after non-linear amplification processing, and aggregate them through exponential mapping to obtain the risk score; Based on time, set three time windows, and the three time windows correspond to time window lengths of seconds, tens of seconds, and minutes respectively; Within each time window, accumulate the risk scores over the entire time period covered by the window, and set a risk threshold for each time window for continuous risk behavior detection; If the accumulated risk scores in at least two of the three time windows exceed their respective set risk thresholds at the same time, it is determined that the current terminal is in a high-confidence abnormal state.

[0013] In a preferred embodiment, when a high-confidence abnormal state is confirmed, generate an emergency short message with the smallest structure and give priority to sending it through the Beidou link, including the following steps: When the triggering condition of the high-confidence abnormal state is established, immediately start the emergency short message construction process, generate the smallest structured communication data content, and initiate a high-priority emergency sending request through the Beidou short message channel.

[0014] The technical effects and advantages of a two-way short message communication method for a network-free scenario according to the present invention: The present invention realizes real-time quantitative evaluation of the accessibility of the communication environment through multi-source environmental perception and link status fusion modeling, can dynamically judge the stability and effectiveness of the public network link, and provide data support for the switching of communication channels. Secondly, the scheduling scoring model constructed based on the link status and task characteristics enables the terminal to give priority to the transmission of high-urgency and high-adaptability tasks under link-restricted conditions, avoids low-value tasks from occupying key communication resources, and improves the utilization efficiency of communication resources. In addition, by introducing a multi-time scale risk integration mechanism, the recognition accuracy of sudden states such as falling, stillness, and sudden changes in air pressure is improved, and structured emergency short messages can be automatically generated without human intervention, ensuring that when the public network is interrupted or personnel lose contact, SOS information can be quickly uploaded and the location is returned, thereby improving the stability, security and engineering adaptability of two-way communication in network-free scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 The present invention is a flowchart of a two-way short message communication method in a non-network scenario. DETAILED DESCRIPTION

[0016] 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.

[0017] In order to achieve the above objectives, Figure 1 A structural schematic diagram of a two-way short message communication method for a non-network scenario of the present invention is provided, which specifically includes the following steps: Based on the multi-source sensor data integrated by the terminal, a disturbance tensor structure is constructed, and a communication reachability score is generated through nonlinear mapping, and the sudden change behavior of the communication environment is detected by combining the sliding window; The communication state vector is constructed according to the 4G link activity score factor. Based on the joint judgment result of the communication reachability score and the 4G link activity score, an asymmetric hysteresis switching strategy is implemented to complete the primary and backup channel switching between the public network communication channel and the Beidou short message channel. Extract attribute vectors from the scheduled tasks, construct a nonlinear priority scoring function, generate task scheduling subsets under communication resource constraints, and perform structure mapping and format compression processing according to the main channel type to form a transmission task package that can adapt to the link; Continuously monitor risk factors and perform risk scoring, combine multi-time scale integration windows to determine risk trends and high-confidence abnormal states. Once a high-confidence abnormal state is confirmed, generate a minimum-structured emergency short message and send it preferentially through the Beidou link.

[0018] In a communication environment without public network access, grid operation and maintenance terminals face complex and uncertain factors such as sudden channel interruptions, terrain occlusion, and environmental interference. Therefore, it is necessary to implement an adaptive communication reachability prediction mechanism inside the terminal. This mechanism takes multiple original sensing signals as inputs and generates real-time link status scores through fusion modeling, providing a basis for subsequent link switching, emergency short message triggering, and resource scheduling strategies; Step 1: Perform multi-source environmental data collection and fusion modeling: First, collect on-site environmental information through multiple types of sensor modules integrated in the terminal. The sensors used include, but are not limited to: barometric pressure sensors (reflecting altitude changes and tightness), three-axis accelerometers (capturing sudden changes in personnel status), geomagnetic sensors (perceiving spatial disturbances), gyroscopes (estimating spatial attitude changes), light sensors (judging occlusion levels), and infrared or ultrasonic ranging sensors (used to identify local narrow spaces). The original data output by each type of sensor has problems such as unit differences, different dimensions, and inconsistent dynamic ranges. Therefore, it is necessary to perform mapping processing of all sensing data to a unified scale for subsequent modeling use; For each original sensor output value, first compress its numerical span through a logarithmic function with a constant base, and then superimpose a stretching factor related to the sensor response sensitivity (set differently according to different sensors), so as to maintain the dynamic distribution characteristics of signals in each dimension and avoid boundary compression distortion problems. All the mapped sensing features are encapsulated into an environmental standardized feature vector, which is used as the first-level data output of this step.

[0019] Perform perturbation behavior modeling and structured tensor generation: The impact of environmental factors on the communication link is often not static single-point interference, but the result of multi-variable joint perturbation. Therefore, after completing the scale unification, it is necessary to further establish a perturbation feature tensor structure from the perspective of dynamic changes. This structure is used to characterize the coupling behavior between the current attitude of the terminal (three-dimensional angle changes represented by the combination of Euler angles), geomagnetic gradient perturbation, and acceleration mutation intensity; Specifically, when the terminal undergoes violent movement, rapid attitude rotation, or abnormal magnetic interference in the spatial direction perception, these factors jointly indicate that there may be severe disturbances in the communication environment or channel interference. To achieve modeling, the acceleration magnitude is used as a disturbance factor, and at the same time, the local gradient change of the geomagnetic vector is used to represent the intensity of the magnetic space disturbance. Combined with the attitude angle change, a three-dimensional cross-mapping is formed to construct a three-dimensional disturbance tensor structure that can be used to capture the joint disturbance trend. In this disturbance tensor structure, each element value is determined by the non-linear combination of three types of disturbance factors. Among them, the acceleration factor is placed in the denominator as a disturbance excitation term, and the exponential function is used to enhance the mutation behavior, while the attitude angle change and the magnetic space disturbance intensity term are placed in the numerator as weight control terms; For example, when grid operation and maintenance personnel carry a smart terminal to inspect high-voltage towers in the mountains and suddenly slip downhill, the terminal will experience the following typical physical characteristics: the acceleration magnitude rises rapidly: from a stable 1.0g to 2.5g rapidly, indicating the existence of slipping or falling behavior, the gyroscope detects that the roll angle changes from 0° to 70° within a short time, indicating a sudden change in the terminal direction, and the change rate of the geomagnetic vector rises sharply, which may be due to magnetic field disturbance caused by approaching a metal structure or entering an enclosed space; The above three changes jointly indicate that the current environment may have changed from stable inspection to an abnormal state, and it is very likely that the stability of the communication channel has been affected. It is necessary to immediately make communication link adaptation and short message preparation; To quantify this joint disturbance trend, the following tensor unit expression can be constructed to represent the disturbance intensity of an element in the three-dimensional tensor: , where E is the disturbance value (dimensionless) of the tensor unit at the current moment; is the attitude angle change rate (such as the roll angle change rate), representing the terminal flipping intensity; is the gradient magnitude of the geomagnetic vector within the time window, representing the degree of spatial magnetic field disturbance; is the acceleration magnitude (e.g., multiples of the gravitational acceleration); is the disturbance excitation response constructed by using the exponential decay function, so that the greater the acceleration, the closer the denominator is to 1, thereby amplifying the overall output of the above formula; In the above expression, the acceleration magnitude The greater it is, the more violent the movement is, and the exponential term The smaller it is, the denominator tends to 1, thereby enhancing the entire disturbance value; if there are sudden changes in the attitude angle ( increases) and magnetic field mutation ( increases) at the same time, the value E of the entire disturbance tensor element will rise rapidly; once multiple tensor elements E show a collective rise within the same time window, it can be determined that the communication environment may have entered a state of "high risk and severe channel disturbance".

[0020] The generation target of the tensor is not only for visual analysis, but to extract the extreme points in the structure as the sensitivity characterization index of potential interference in the current communication state.

[0021] The generation logic of the communication reachability score is as follows: After completing the construction of the environmental standardization eigenvector and the perturbation structure tensor, these two types of data need to be fused and mapped into a communication link reachability index space; Perform individual mapping on each component in the environmental standardization eigenvector. The core idea of this mapping function is: for each sensor eigenvalue, introduce a corresponding response adjustment factor (referred to as the excitation coefficient), which is used to control the response amplitude of this feature to the final score result. The mapping method adopts the form of a hyperbolic tangent function with a saturation interval. By adjusting the steepness of the response curve, the eigenvalue can have a stable response ability to the scoring system; Extract the perturbation extreme value in the perturbation structure tensor as the global perturbation term. By combining it with the above sensor eigenvalue term, construct the total input value of the scoring function. This combination method adopts a splicing mechanism based on the extreme value extraction and suppression function, where the perturbation extreme value is input as the perturbation sensitivity factor, dominating the low-value interval of the scoring function; Finally, input the above combination result into a non-linear boundary-preserving function. The goal of this function is to compress and map the linear superposition result into the interval [0, 1], while maintaining the response continuity at the boundary. Non-Sigmoid activation methods such as the SoftSign function or the Swish function can be used. The output result is the communication reachability score factor. The closer this score factor is to 1 numerically, the more stable the current communication environment and the more controllable the channel quality; the closer it is to 0, the greater the risk that the current scenario poses to the communication link; Introduce a time-series sliding window mechanism: maintain a scoring record window with a fixed length inside the terminal. Each time a new communication reachability score is received, calculate the difference between it and the communication reachability score result at the previous moment in the window. If the current score drops by more than the set drop amplitude threshold, generate a communication mutation flag bit as the communication environment mutation behavior; It should be noted that the setting of the threshold is based on the empirical environmental drift tolerance, usually in the interval of 0.1 to 0.2. A large change in the score indicates that the environment has changed rapidly, such as entering a tunnel, closing an elevator car, or being blocked by a cliff wall. At this time, switching to the Beidou short message channel in advance can significantly improve the link stability and the timeliness of alarm transmission.

[0022] After completing the multi-source perception and fusion modeling of the communication environment, it is necessary to further determine the primary and backup paths of the communication channel based on the environmental perception results and link state estimation results, so that when the 4G public network signal quality deteriorates or is unavailable, it automatically switches to the Beidou short message channel to ensure the most basic two-way short message communication function; after the 4G network is restored, quickly determine whether to switch back to the public network channel to support large data volume transmission tasks, such as image summaries, remote scheduling commands, etc. To avoid resource consumption and communication interruption caused by frequent switching, an asymmetric hysteresis mechanism is adopted, and structured decision rules are established based on two core indicators: perception score and 4G link activity score. Step 2, perform link state judgment and primary and backup communication channel switching, and the specific content is as follows: Construct a joint judgment index system for communication status: Use the communication reachability score generated in Step 1 as the main environmental factor for channel switching. This score comprehensively considers various environmental state factors such as air pressure disturbance, attitude change, magnetic field disturbance, and acceleration fluctuation, and conducts a comprehensive evaluation of whether it is suitable to establish a stable communication link currently. The closer the value of this score is to 1, the more stable the environment, the less channel interference, and it is suitable to maintain high-speed communication; conversely, the lower the value, the worse the communication environment, and it is necessary to consider degradation strategies or activate the backup channel. In addition to the environmental factor, introduce the 4G link activity score factor as the communication status vector, that is, construct a dynamic evaluation function that includes signal stability and position disturbance trend. Specifically, the 4G link activity score is used to characterize the dynamic communication status of the current public network link, comprehensively reflecting the fluctuation trend and potential degradation risk of the communication link. This indicator does not rely on the received signal strength or signal-to-interference ratio at a single time point, but is constructed based on the change rate of 4G received power and the disturbance amplitude of the terminal geographical location within a certain time window. Specifically, first continuously record the change of the 4G signal strength received by the terminal within a given monitoring period, and calculate the decline rate of the received power per unit time; at the same time, according to the continuous position point sequence provided by the terminal positioning module (such as BDS / GPS), calculate the spatial displacement amplitude within the same period, and quantify the position disturbance trend of the terminal in the physical space. To integrate the dynamic change characteristics of the above two dimensions, a weighted combination structure is introduced to construct a 4G link activity scoring function. Among them, signal stability reflects the fluctuations at the signal level, and the position perturbation trend reflects the link degradation risk caused by environmental occlusion or geographical distance. The function form adopts a linear coupling structure, allowing the influence degrees of signals and positions on the 4G link activity score to be adjusted separately, so as to adapt to the sensitivity requirements in different application scenarios. For example, if the power decline rate is significant and the terminal displacement rapidly approaches a known occlusion area (such as a canyon or a tunnel entrance), the 4G link activity score will rise rapidly, indicating that the current link is in a high degradation risk state; for example, the calculation method of the 4G link activity scoring function is as follows: , where, is the 4G link activity score, which is used to characterize the link degradation trend. The larger the value, the higher the instability degree of the link; is the change amount of the 4G received power per unit time, usually negative (indicating signal attenuation), and its absolute value represents the decline rate; is the monitoring period length, with the unit of second; is the spatial movement distance of the terminal within the monitoring period, which can be obtained by calculating the Euclidean distance or path integral of the starting and ending GPS coordinate points; The calculation of the 4G link activity score aims to dynamically perceive the immediate stability and failure risk of the public network communication link. As a key input parameter for the main and backup channel switching judgment, compared with the traditional evaluation method based on instantaneous signal strength, this method effectively improves the foresight and stability of link state judgment by introducing two dynamic dimensions of signal trend and position change. Especially in the inspection scenarios with intermittent coverage, high mobility or complex environment, this processing mechanism can significantly enhance the real-time response ability of the communication switching strategy and the accuracy of link selection.

[0023] Combining the communication reachability score and the 4G link activity score forms a link state vector. The first component reflects the physical feasibility of the environment for the link, and the second component reflects the signal behavior stability of the network itself.

[0024] Set a segmented hysteresis switching strategy: Based on the above two-factor link state vector, construct an asymmetric threshold judgment model to reduce the frequent channel oscillation switching caused by signal fluctuations. This model divides the communication reachability score into three intervals: high reach area, switching hysteresis area and low reach area. The high reach area indicates that the environment is unobstructed and the 4G signal is stable, and the 4G channel is preferred; the low reach area means that the environment is complex and the public network quality deteriorates, and it is necessary to actively switch to the Beidou short message link. The middle switching hysteresis area is used to keep the current communication state unchanged to avoid repeated switching caused by small fluctuations; The specific values ​​are set as follows: If the communication accessibility score exceeds 0.8, it is judged as a high accessibility state, and the current 4G main link communication is maintained; if it is lower than 0.5, it is judged as a low accessibility state, triggering the channel switching strategy and switching to the Beidou short message link; if it is between 0.8 and 0.5, it is a switching hysteresis state, and it will not switch immediately and enter the monitoring buffer period. The 4G link activity score is an auxiliary judgment basis. If the value shows a negative growth trend, that is, the signal strength decreases within a unit time and the terminal position moves significantly, it indicates that the signal decay trend is strong. Even if the communication score does not fall below 0.5, the channel switching strategy can be triggered to enter the Beidou short message link backup channel in advance.

[0025] After switching the main channel to the backup channel, the structure of the current communication task needs to be adapted and adjusted. Since the byte limit of Beidou short messages in one transmission is relatively strict, usually within about 120 characters, it is impossible to carry large-capacity task packages such as redundant data, image summaries, and inspection logs; Build a communication task structure mapping mechanism to classify the current pending tasks according to priority, including: Class I tasks: such as personnel positioning, fault alarm, emergency assistance and other short-content tasks, which are sent through the Beidou short message channel first; Class II tasks: such as device status change records and log data summaries, which are sent via 4G links; The third type of tasks: such as image transmission and operation record upload, require a file transmission channel supported by a 4G link. If the Beidou short message link is switched, it will be temporarily cached; When switching to the Beidou link, only the necessary fields in one type of task are extracted to form a streamlined structure, including the terminal number, longitude and latitude position, alarm type identifier, timestamp and scene feature code, to avoid exceeding the short message capacity limit.

[0026] In summary, link status judgment and primary-backup channel switching are the core processes for achieving robust communication assurance in actual complex environments. By constructing an environment-driven scoring mechanism and a network link dynamic response model, and combining it with an asymmetric hysteresis strategy, an intelligent downgrade switching from 4G to Beidou and an automatic recovery mechanism to the public network are achieved, ensuring reliable minimum communication capabilities in extreme operating scenarios.

[0027] Step 3: Perform communication task identification and priority scheduling. According to the current communication channel capability (such as 4G public network or Beidou short message) and communication reachability score results, perform task identification, priority classification and scheduling queue management. The specific contents are as follows: Perform communication task structure analysis and task attribute vector extraction: The first step in communication task recognition is the structured analysis of tasks. Each task, when generated, not only contains its original content but also carries task meta-information related to it. For unified processing, the task structure needs to be abstracted into multiple attribute vectors. Typically, the attribute vectors include: task generation time, maximum allowable delay time, data volume size, task urgency level, and link reachability score (i.e., whether it is allowed to be transmitted through the Beidou short message channel); These attribute vectors do not directly participate in the scheduling and sorting but will be used as input factors to participate in the calculation of the subsequent priority scoring function. There are interactive effects among the attribute vectors. For example, a task with a large data volume but a high urgency level may still need to be scheduled first under a bandwidth-limited channel.

[0028] The construction and calculation logic of the priority scoring function include: After the construction of the task attribute vectors is completed, to achieve task scheduling and sorting, a set of non-linear priority scoring functions need to be established to convert the multi-attribute vector input into comparable scoring values. This priority scoring function needs to meet the following technical requirements: For urgent tasks (such as alarms, rescue positioning), it should have an exponential response ability to ensure that the score is significantly higher than that of ordinary tasks; For expired tasks or tasks with low delay tolerance, the priority scoring function should drop rapidly; Considering the interval between the task generation time and the current time, if the time is too long and the task has not been transmitted, the priority scoring function gradually increases to prevent long-term backlog; The priority scoring function is used to quantify the scheduling value of each task in the current communication state and is calculated through the joint modeling of task attributes and network states. Specifically, first, the urgency level, generation time, maximum allowable delay time, and communication reachability score of each task to be scheduled are extracted. Among them, the task urgency level is used as a static scheduling factor to reflect the priority level of the task; the task generation time is used to construct a time decay term to evaluate the time sensitivity of the task at the current moment; the task delay tolerance and the link reachability score together constitute the transmission feasibility term, which is used to evaluate whether the task has reasonable transmission conditions under the current link; To better reflect the immediate scheduling value of the task, a non-linear function is used to construct the time modulation term, that is, the difference between the task generation time and the current system time is used as the input, and after multiplying by the adjustment parameter, it is used as the variable of the exponential function, so as to generate the time effect score of the task. This method simulates the behavior that the scheduling pressure rises sharply when the task is approaching timeout and overcomes the problem of insensitive response of the linear time decay model; Meanwhile, to evaluate the transmissibility of tasks under link conditions, the maximum allowable delay time of the task is multiplied by the link reachability score, and the result is input into a non-linear response function to be mapped into a transmissibility score. This score establishes a dynamic mapping relationship between the time constraints of the task itself and the current link state, and can effectively reflect whether the task is suitable for the current transmission; Finally, the time effect score and the transmission feasibility score are combined structurally, and a normalized score value is output through a boundary compression function, which is restricted between 0 and 1. This compression function takes the form of dividing the input value by the sum of its absolute value and 1, and has good boundary continuity and intermediate value resolution ability. It can retain the scoring details and avoid the distortion of sorting caused by extreme values; For example, a specific example expression combining the structure of the priority scoring function and the boundary compression processing method is: , where PF represents the final normalized priority score, and its value range is (-1, 1). If it needs to be restricted to [0, 1], a further linear mapping can be performed; SJ is the time effect scoring item of the task, indicating the urgency of the task; CS is the transmission feasibility scoring item of the task, which is obtained by multiplying the delay tolerance by the link reachability score.

[0029] Perform scheduling window construction and subset generation: The scheduling window and subset generation mechanism is used to screen the task set with the highest current priority score and the most suitable link state under the condition of limited communication resources, ensuring the fast scheduling and transmission of high-priority tasks within the established transmission capacity range. Based on the task priority scoring results and the communication channel transmission capacity indicators, a task scheduling subset is constructed to form the smallest task combination that can be executed within the current scheduling period; First, set the available transmission capacity threshold for the current scheduling period according to the type of the main communication channel (such as 4G public network or Beidou short message). This threshold is the upper bound of the scheduling window and reflects the total amount of data that can be transmitted per unit time under the current link conditions. Under the 4G link, this capacity can be given according to the real-time bandwidth evaluation results, and under the Beidou short message link, it is the maximum length of the fixed message, such as 120 bytes. The transmission capacity is denoted as RL and is used as the upper limit constraint for subsequent task subset screening; Specifically, multiply the bandwidth capacity identifier of the current communication channel (which can be the estimated value of the 4G network rate or the maximum load of the Beidou short message) by the length of the scheduling period (for example, the maximum transmission time available for the current communication scheduling), and a meaningful allocable communication budget can be obtained as the transmission capacity; Sort the calculated task priority score results in descending order to form a task scheduling queue. Select the task with the highest score value from the queue head in sequence, and accumulate its data volume until the cumulative total is about to exceed the current window capacity limit. Then stop task selection to form a task subset for the current scheduling period, which is denoted as Ts and is the optimal task set that meets the score priority constraint and transmission capacity constraint.

[0030] Perform communication channel adaptation and task structure mapping: The communication channel adaptation and task structure mapping mechanism is used to dynamically adjust the data structure format of tasks according to the capability characteristics of the current main communication link, ensuring the executability and transmission efficiency of the scheduled task subset during the actual sending process. Since there are significant differences in transmission rate, maximum packet capacity, and continuity guarantee between 4G links and Beidou short message links, the task structure needs to be processed differently to meet the transmission requirements of different links; Specifically, first identify the type of the current main communication channel. If it is a 4G public network link, it can support the transmission of task packets in standard structure format, including multi-field data structures such as image summaries, job logs, positioning trajectories, and multi-parameter status quantities, without the need for compression mapping processing. The task can be directly encapsulated in the original structure and submitted to the communication interface; If the current main channel is a Beidou short message link, due to the limitation of the maximum number of bytes per sent message (for example, within 120 bytes), the scheduled task must undergo structure compression and field reduction processing. The task structure mapping process is as follows: Perform field screening and reconstruction, retaining the necessary fields in the task, including the terminal unique identifier, longitude and latitude coordinates, task number, event type code, timestamp, and several status bit flags. Non-keyword fields (such as image summaries, complete logs, auxiliary status quantities, etc.) are excluded; Perform field encoding compression, converting the task fields using compact encoding, such as fixed-length encoding, predefined index mapping, or bitmask structures, to compress fields such as task type, status label, and alarm level into integer or binary bit strings; Perform field alignment and byte control, perform byte alignment processing on all retained fields to ensure that the total length of the structure is within the available length of the link. If it exceeds the limit, perform differential compression processing on the location information, or split the task into multiple short messages for asynchronous sending; To ensure the effectiveness of the task after structure mapping, construct a set of task structure adaptation template libraries, automatically select applicable structure compression templates according to different link types, and bind the templates to each task in the scheduling subset before the scheduling link, ensuring that the structure adjustment of the task is completed before transmission, and avoiding message truncation or communication failure caused by structure mismatch; In addition, the task structure mapping can also be equipped with a fragment recombination mechanism in the short message link. When a task needs to be split into multiple short message fragments for transmission, a task identifier and a sequence index are attached to each fragment. The receiving end reorganizes and restores the fields in sequence to ensure the logical integrity and service consistency of the task content.

[0031] Continuously monitor the running state of the terminal and the environmental disturbance state, construct an emergency state discrimination model and a short message active triggering strategy. In the case of no public network coverage, the closed-loop communication goal of information being sendable, receivable, and feedbackable can still be guaranteed. Step 4: Construct an emergency trigger and a short message quick response mechanism. The specific steps are as follows: Preprocess the input data of multiple types of sensors of the terminal (such as acceleration, air pressure, gyro angular velocity, geomagnetic disturbance, and distance change). Construct multiple anomaly detection factors for different risk scenarios, such as factors for indicating a person's fall, a too-long stationary time, the intensity of spatial vibration, and a sudden change in air pressure. Adopt a combined form of the high-order derivative, disturbance gradient, and trend amplitude of continuous time series changes to enhance the response sensitivity to emergencies. Taking the fall factor as an example, instead of directly judging that the acceleration exceeds the threshold, calculate the deviation between the instantaneous growth rate of the acceleration modulus and the integral of the amplitude change per unit time. If this deviation accumulates rapidly and is higher than the set change threshold, it is marked as a potential fall behavior. The construction of the stationary time factor does not use the average acceleration, but is based on whether the change rate of the acceleration at multiple moments approaches zero and continuously exceeds the time threshold to conduct a stable state discrimination. Construct and process a non-linear combination risk scoring function for multiple anomaly detection factors: The non-linear combination risk scoring function for multiple anomaly detection factors is used to evaluate the potential anomaly state level of the terminal in the current environment, and realizes the quantitative analysis of the overall risk by fusing key indicators in various sensor data. Specifically, first define a number of independent anomaly detection factors, and each anomaly detection factor represents the response intensity of a certain type of sensor anomaly event, such as acceleration mutation, attitude flip rate, geomagnetic disturbance gradient, and sudden air pressure drop amplitude. To achieve unified dimension processing, all anomaly detection factors are first normalized to ensure that their values are within a consistent standard interval. Introduce a non-linear amplification mechanism to process the normalized factors. Specifically, take each normalized anomaly detection factor as an input variable and pass it into a modulation function with a saturation response characteristic, such as the hyperbolic tangent function (tanh). This type of function has a gentle output change when the input value is small, which can suppress false signals brought by minor disturbances; while when the input value approaches the anomaly critical area, the output response will rise rapidly, realizing a strong amplification of emergencies and forming a non-linear gain structure. Combine all the output values of the factors after non-linear amplification processing and achieve aggregation through exponential mapping. For example, first raise each amplified value to the power of its corresponding sensitivity index, then accumulate all the results, and finally take the maximum power index value as the normalization factor to construct the overall risk score result. Thus, the scoring logic that emphasizes the dominance of strong anomalies and the cooperation of weak anomalies is emphasized in the structure, ensuring that when one or more key factors are significantly abnormal, the total score rises rapidly; at the same time, when multiple medium-strength factors act simultaneously, the effective accumulation of risk scores can also be achieved; For example, assume that the set of anomaly detection factors is: , denotes the k-th anomaly detection factor, and define the risk score function as follows: , where is the i-th anomaly detection factor (normalized); is the sensitivity amplification value of the i-th anomaly detection factor, which adjusts the steepness of its contribution to the overall risk and is set according to actual needs; is the sensitivity index; For example, under the condition of existing data accumulation, by analyzing the historical true risk events (such as falls, slips, and static coma) and the change trends of the response intensities of each anomaly detection factor before the event occurs, the average early lift amplitude and peak response ratio of each type of factor before the risk appears can be statistically obtained. The sensitivity index can be defined accordingly as: , where is the average value of the i-th anomaly detection factor before the risk event; is the average value of this anomaly detection factor during non-risk periods; is the empirical adjustment factor, which controls the overall amplification multiple and is set according to the actual scenario; is a small constant to prevent the denominator from being zero.

[0032] The construction of the multi-factor non-linear combination risk score function aims to dynamically perceive the coupling effect of the terminal state and external environmental disturbances, and provide a quantifiable, adjustable, and interpretable risk level output indicator. This function avoids over-reliance on a single anomaly detection factor, reduces the probability of false triggering of risk judgments caused by accidental disturbances, the introduction of the non-linear response function enhances the mutation perception ability of the score, and the maximum exponent normalization strategy ensures the stability of the score boundary.

[0033] Construct risk trend windows at multiple time scales: Based on time, divide the risk evolution behavior into three levels: short time, medium time, and long time, corresponding to the risk change perception abilities at the second level, ten-second level, and minute level respectively, and can simultaneously identify different risk patterns with strong suddenness, medium persistence, or slow evolution; In specific implementation, three time windows are set: the short time window is used to capture instantaneously occurring high-intensity risks, such as terminal drops, severe collisions, etc.; the medium time window focuses on the change trends at the level of dozens of seconds, such as short-term stillness of a person in a special state, abnormal postures, or magnetic field interference; while the long time window pays attention to the cumulative state of the risk score within several minutes, mainly used to identify low-frequency risks that evolve slowly but may have potential hazards, such as a person losing the ability to move, long-term stillness, or continuous air pressure drop, etc.; Within each time window, the change of the risk score over time is continuously tracked, and the risk score is cumulatively processed throughout the time period covered by the window. This cumulative process does not only focus on the instantaneous numerical value of the score, but also considers the duration of its maintaining a high level. Simply put, if the risk score remains at a medium to high level for a certain period of time, even if it does not reach the peak, it will be regarded as threatening due to its persistence; In order to determine whether the cumulative risk reaches the emergency trigger standard, a risk threshold is set for each time window. This threshold represents the risk integral standard that is considered sufficient to trigger an actual safety event at this time scale. If the cumulative value of the risk score within the window exceeds the corresponding set threshold, it can be considered that there is a continuous risk behavior during this time period; A dual judgment mechanism is set. Only when the cumulative risk scores in at least two of the three time windows simultaneously exceed their respective set risk thresholds, it is determined that the current terminal is in a high-confidence abnormal state. This dual-window collaborative trigger strategy can, on the one hand, filter out short-term accidental disturbances and avoid false reports caused by sensor noise or instantaneous anomalies, and on the other hand, it can also ensure the timely response to truly dangerous events. Especially when risks appear simultaneously on two time scales, it indicates that they have sufficient intensity and persistence and should trigger the emergency handling process preferentially; Furthermore, this mechanism supports setting different window lengths and judgment thresholds according to different usage scenarios. For example, for a high-intensity operation site, a shorter window response time and a lower judgment threshold can be set to achieve rapid perception of severe drops or impacts; for an environment with long-term presence, the threshold of the short-term window can be increased and the trigger threshold of the long-term window can be decreased to enhance the monitoring ability for slow-onset risks.

[0034] When it is determined that the current terminal is in a high-risk state through the above-mentioned integral judgment mechanism of the risk score and it is confirmed that the trigger condition is established, the emergency short message construction process will be immediately started, the minimum structured communication data content will be generated, and a high-priority emergency sending request will be initiated through the Beidou short message channel.

[0035] Conduct emergency short message structure construction and sending: In terms of message content design, the system adopts the principle of field minimization, only retaining the core fields to ensure that the total data length is within the Beidou short message communication capacity (such as not exceeding 120 bytes). The field structure is sorted according to the practical priority of emergency response, including but not limited to the following elements: Terminal identity identification field: Used to uniquely identify the device or person initiating the emergency communication, which can be the device ID or registration code; Location information field: Includes longitude, latitude, and optional altitude information to provide a basis for geographical positioning; Risk type field: Output by the system risk determination module, using preset codes to represent types such as falling, instability, stillness, and abnormal air pressure; Risk level field: Characterizes the risk score level in the current state, divided into several risk level labels (such as low, medium, high, extremely high); Timestamp field: Records the sending time to ensure that the receiving party has a clear perception of the occurrence sequence of events; Status summary field: Used to encode several sensor anomaly marks, such as whether there is magnetic interference, attitude flip, stillness timeout, etc., to support initial analysis at the remote end; In terms of the sending strategy of emergency short messages, it can be set to have the highest immediate sending priority without queuing for the task scheduling window. At the same time, to ensure the reachability of emergency information, the system will enter the retransmission guarantee process after the first sending. The specific method is as follows: If the receiving confirmation or response feedback from the scheduling end is not received within the set time, the system will repeat sending the same short message at the set time interval and record the time point and sending status of each sending behavior. The number of retransmissions is set to a finite number, such as three or five times. After exceeding this number, the terminal enters the standby listening state, waiting for other links to recover or manual access intervention; In addition, to improve the information processing efficiency of the receiving end, a unified emergency identification header can be added to the structure of all emergency short messages, facilitating the Beidou receiving system or the scheduling platform to quickly parse, classify, store, and preferentially display them. For multiple consecutive risk events, the most recent risk score trigger point will be used as the benchmark to prevent the accumulation of duplicate messages or information redundancy caused by repeated fluctuations of the risk score in a short period.

[0036] It should be noted that the threshold information related to this embodiment is pre-set by professionals and will not be explained in detail here. There are some cases where the English letters of some parameters are the same in this embodiment, but different meanings are explained during use, and they will not be explained one by one here.

[0037] Through multi-source environment perception and link state fusion modeling, the present invention realizes real-time quantitative evaluation of the reachability of the communication environment, can dynamically judge the stability and effectiveness of the public network link, and provides data support for the switching of communication channels. Secondly, the scheduling scoring model constructed based on the link state and task characteristics enables the terminal to preferentially transmit tasks with high urgency and high adaptability under limited link conditions, avoiding low-value tasks from occupying key communication resources and improving the utilization efficiency of communication resources. In addition, by introducing a multi-time scale risk integration mechanism, the recognition accuracy of sudden states such as falling, static, and sudden air pressure changes is improved, and a structured emergency short message can be automatically generated without human intervention, ensuring the rapid upload of SOS information and location feedback in case of public network interruption or personnel loss of contact, and enhancing the stability, security, and engineering adaptability of two-way communication in a network-free scenario.

[0038] The above formulas are all dimensionless and take their numerical values for calculation. The formula is obtained by collecting a large amount of data for software simulation to obtain a formula closest to the actual situation. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0039] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product.

[0040] Those of ordinary skill in the art can realize that the modules and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. A professional technician can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0041] In addition, the functional modules in each embodiment of this application can be integrated into a processing module, or each module can exist physically alone, or two or more modules can be integrated into one module.

[0042] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application, and all should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

[0043] Finally, the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A two-way short message communication method for scenarios without network, characterized in that: It includes the following steps: Based on the multi-source sensor data integrated by the terminal, construct a perturbation tensor structure, generate a communication reachability score through non-linear mapping, and combine a sliding window to detect sudden changes in the communication environment; Construct a communication state vector according to the 4G link activity score factor, and execute an asymmetric hysteresis switching strategy according to the combined judgment result of the communication reachability score and the 4G link activity score to complete the primary and backup channel switching between the public network communication channel and the Beidou short message channel; Extract the attribute vector of the task to be scheduled, construct a non-linear priority scoring function, generate a task scheduling subset under communication resource constraints, and perform structure mapping and format compression processing according to the main channel type to form a transmission task packet adaptable to the link; Continuously monitor the risk factors and perform risk scoring, combine multi-time scale integration windows to determine a high-confidence abnormal state for the risk trend, and when the high-confidence abnormal state is confirmed, generate an emergency short message with the smallest structure and give priority to sending it through the Beidou link.

2. The two-way short message communication method for a network-free scenario according to claim 1, characterized in that: Based on the multi-source sensor data integrated by the terminal, construct a perturbation tensor structure. The specific steps include: Collect on-site environmental information through the sensor module integrated by the terminal. The sensors used include a barometric pressure sensor, a three-axis accelerometer, a geomagnetic sensor, a gyroscope, a light sensor, an infrared or ultrasonic ranging sensor; Perform mapping processing of the same scale on the sensing data output by each type of sensor to obtain an environment standardization feature vector; When the terminal moves, rotates rapidly in attitude, or senses abnormal magnetic disturbances in the spatial direction, use the acceleration modulus as the perturbation factor, and at the same time use the gradient change of the geomagnetic vector to represent the magnetic space perturbation intensity, and combine the attitude angle change to form a three-dimensional cross mapping to construct a three-dimensional perturbation tensor structure for capturing the combined perturbation trend; In the three-dimensional perturbation tensor structure, each element value is determined by the non-linear combination of three types of perturbation factors. The acceleration factor is placed in the denominator as the perturbation excitation term, and the exponential function is used to enhance the mutation behavior. The attitude angle change and the magnetic space perturbation intensity term are placed in the numerator as the weight control term.

3. The two-way short message communication method for a network-free scenario according to claim 2, characterized in that: And generate a communication reachability score through non-linear mapping, and combine a sliding window to detect sudden changes in the communication environment. The specific steps include: After completing the construction of the environment standardization feature vector and the perturbation structure tensor, fuse and map these two types of data into a communication link reachability index space; For each sensor eigenvalue, introduce a corresponding response adjustment factor and perform mapping in the form of a hyperbolic tangent function in the saturation interval; Extract the perturbation extreme value in the perturbation structure tensor and use it as the global perturbation term. Combine it with the sensor eigenvalue term to construct the total input value of the scoring function; The perturbation extreme value is input as the perturbation sensitivity factor and dominates the low value interval of the scoring function; Input the combined result into a non-linear boundary-preserving function, compress and map the linear superposition result into the interval [0, 1], and output the communication reachability score factor; Use a time-sequential sliding window mechanism to maintain a scoring record window of a fixed length inside the terminal. Each time a new communication reachability score is received, calculate the difference with the communication reachability score result at the previous moment in the window. If the current score drops by more than the set drop threshold, generate a communication mutation flag bit, which is regarded as a communication environment mutation behavior.

4. The two-way short message communication method for the offline scenario according to claim 3, wherein: Construct a communication state vector according to the 4G link activity scoring factor. According to the combined judgment result of the communication reachability score and the 4G link activity score, execute an asymmetric hysteresis switching strategy. The specific steps are as follows: Use the communication reachability score as the environmental factor for channel switching, and introduce the 4G link activity scoring factor as the communication state vector to evaluate signal stability and position perturbation trend; The 4G link activity score is used to represent the dynamic communication state of the current public network link. The signal stability of the 4G link activity score is obtained by continuously recording the change of the 4G signal strength received by the terminal within a given monitoring period and calculating the decline rate of the received power per unit time. According to the continuous position point sequence of the terminal positioning module, calculate the spatial displacement amplitude within the same period to obtain the position perturbation trend of the terminal in the physical space, and use a weighted combination structure to construct the 4G link activity scoring function; The two indicators of the communication reachability score and the 4G link activity score jointly constitute the link state vector. The communication reachability score reflects the physical feasibility of the environment for the link, while the 4G link activity score reflects the signal behavior stability of the network itself; Set a segmented hysteresis type switching strategy according to the link state vector.

5. The two-way short message communication method for a scenario without network according to claim 4, wherein: Complete the primary and backup channel switching between the public network communication channel and the Beidou short message channel, including the following steps: When the communication reachability score exceeds 0.8, it is determined to be in a high reachability state, and the current 4G main link communication is maintained; when the communication reachability score is lower than 0.5, it is determined to be in a low reachability state, and the channel switching strategy is triggered to switch to the Beidou short message link; when the communication reachability score is less than or equal to 0.8 and greater than or equal to 0.5, it is in the switching hysteresis state, and no immediate switching is performed, and it enters the monitoring buffer period; The 4G link activity score is used as an auxiliary judgment basis. If the 4G link activity score shows a negative growth trend, trigger the channel switching strategy and enter the standby channel in advance; After switching to the standby channel, adaptively adjust the structure of the current communication task; Construct a communication task structure mapping mechanism, and classify the current tasks to be sent into type-one tasks, type-two tasks, and type-three tasks according to their priorities; When switching to the Beidou link, extract the necessary fields in the type-one tasks to form a structure. The necessary fields include the terminal number, longitude and latitude position, alarm type identifier, timestamp, and scene feature encoding to avoid exceeding the short message capacity limit.

6. The two-way short message communication method for the offline scenario according to claim 5, wherein: Extract the attribute vector for the task to be scheduled and construct a non-linear priority scoring function, including the following steps: Abstract the task structure into an attribute vector, which includes the task generation time, maximum delay time, data volume size, task urgency, and link reachability score; After the task attribute vector is constructed, perform task scheduling sorting and construct a non-linear priority scoring function; The priority scoring function is used to quantify the scheduling value of each task under the current communication state and is calculated by jointly modeling task attributes and network states; Extract the urgency, generation time, maximum allowable delay time, and communication reachability score of each task to be scheduled; Use a non-linear function to construct a time modulation term. Taking the difference between the task generation time and the current system time as the input, multiplying it by the adjustment parameter and using it as the variable of the exponential function to generate the time effect score of the task; Evaluate the transmitability of the task under the link conditions, multiply the maximum allowable delay time of the task by the link reachability score, and input the result into a non-linear response function to map it to a transmitability score. The transmitability score reflects whether the task can be transmitted currently; Structurally combine the time effect score and the transmission feasibility score, and output a normalized score value as the priority score through a boundary compression function, which is restricted between 0 and 1.

7. A two-way short message communication method for a network-free scenario according to claim 6, characterized in that: Generate a task scheduling subset under communication resource constraints, and perform structural mapping and format compression processing according to the main channel type to form a transmission task packet adaptable to the link, including the following steps: Set the available transmission capacity threshold for the current scheduling period according to the main communication channel type. The available transmission capacity threshold is the upper bound of the scheduling window, which reflects the total amount of data that can be transmitted per unit time under the current link conditions. The available transmission capacity threshold under a 4G link can be given according to the real-time bandwidth evaluation result, and for the Beidou short message link, it is the maximum length of a fixed message; Multiply the bandwidth capacity identifier of the current communication channel by the length of the scheduling period to obtain the allocable communication budget as the transmission capacity; The bandwidth capacity identifier is the estimated 4G network rate or the maximum load of the Beidou short message, and the length of the scheduling period is the maximum transmission time available for the current communication scheduling; Sort the calculated task priority score results in descending order to form a task queue to be scheduled; sequentially select the task with the highest score value from the head of the queue, accumulate the data volume until the cumulative total exceeds the current window capacity upper limit, then stop task selection, and form the subset of the current scheduling period as the task scheduling subset; Identify the current main communication channel type to form a transmission task packet adaptable to the link. If it is a 4G public network link, no compression mapping processing is required; If the current main channel is the Beidou short message link, schedule tasks for structural compression and field screening processing to form a transmission task packet. The field screening processing includes field screening and reconstruction, field coding compression, field alignment, and byte control.

8. A two-way short message communication method for a scenario without network according to claim 7, characterized in that: Continuously monitor risk factors and perform risk scoring, and combine multi-time scale integration windows to determine high-confidence abnormal states for risk trends, including the following steps: Construct anomaly detection factors for different risk scenarios, and construct and process a non-linear combined risk scoring function for the anomaly detection factors. The risk scoring function is used to evaluate the potential abnormal state level of the terminal in the current environment; First perform a normalization operation on all anomaly detection factors, and use the hyperbolic tangent function for non-linear amplification. Combine the output values of all factors after non-linear amplification processing, and achieve aggregation through exponential mapping to obtain the risk score; Based on time, three time windows are set, and the three time windows correspond to time window lengths of seconds, tens of seconds, and minutes respectively; Within each time window, the risk scores are cumulatively processed over the entire time period covered by the window, and a risk threshold is set for each time window to detect continuous risk behaviors; If the cumulative risk scores in at least two of the three time windows exceed their respective set risk thresholds simultaneously, it is determined that the current terminal is in a high-confidence abnormal state.

9. A two-way short message communication method for a network-free scenario according to claim 8, characterized in that: After the high-confidence abnormal state is confirmed, an emergency short message with the smallest structure is generated and preferentially sent through the Beidou link, including the following steps: When the trigger condition for the high-confidence abnormal state is established, the emergency short message construction process is immediately started, the smallest structured communication data content is generated, and a high-priority emergency sending request is initiated through the Beidou short message channel.

Images