Unmanned aerial vehicle millimeter wave ultra-long distance emergency communication method and system

Abstract

The invention discloses an unmanned aerial vehicle millimeter wave ultra-long distance emergency communication method and system. According to the method, prospective channel information is obtained through coupling a rainfall attenuation and atmospheric waveguide opportunity dual-prediction model; a cross-domain collaborative optimization model with the communication frequency point sequence and the flight path sequence as joint decision variables is constructed, and a collaborative control instruction set is solved; and finally, controlling the unmanned aerial vehicle to synchronously execute frequency point switching and track adjustment. According to the method, a traditional mode of isolated optimization of communication and flight control is broken through, the optimal propagation environment is formed by actively and cooperatively changing the spatial position and communication parameters, rainfall attenuation can be effectively avoided, atmospheric waveguides and other opportunities can be reliably utilized, and the reliability, throughput and coverage range of long-distance millimeter wave communication of the unmanned aerial vehicle are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the cross-technical field of wireless communication and autonomous system collaborative control. Specifically, it relates to a millimeter-wave ultra-long-range emergency communication method and system for unmanned aerial vehicles (UAVs) based on weather forecasting and physical propagation models. It is suitable for emergency communication scenarios that operate in the millimeter-wave frequency band, which is susceptible to rainfall attenuation and atmospheric waveguide effects, and whose communication distance significantly exceeds the coverage range of conventional line-of-sight links. Background Technology

[0002] In natural disasters (such as floods, earthquakes, and snowstorms) or public safety incidents, ground communication infrastructure is often damaged or disrupted. Drones carrying communication base stations can quickly take off to form temporary aerial networks, a crucial means of restoring communication in disaster areas. Millimeter-wave bands (such as 24-86 GHz), with their ultra-wide bandwidth, can support high-speed emergency services such as high-definition video transmission from disaster areas and the aggregation of large amounts of sensor data, achieving ultra-long-distance coverage. However, millimeter-wave signals are extremely sensitive to precipitation (including rain, snow, and hail) during propagation in space; their intensity is severely attenuated by the absorption and scattering of precipitation particles, and the attenuation intensifies with increasing frequency. In emergency communication scenarios with extremely high reliability requirements, overcoming the instability or even interruption of millimeter-wave links caused by rain and snow is a critical technical bottleneck that must be overcome.

[0003] Traditional solutions have evolved along two isolated paths, neither of which can meet the high reliability requirements of emergency communications in severe weather conditions:

[0004] The first approach is communication link adaptation, such as adaptive modulation and coding (AMC) or switching based on real-time channel state information. For example, Chinese patent CN201910702222.0A proposes adjusting the modulation and coding scheme based on predicted rainfall. Essentially, this type of method optimizes communication signal parameters at a fixed spatial point on the UAV to passively combat time-varying channel fading, belonging to an "on-site resistance" strategy. However, its performance improvement is limited when encountering deep fading caused by heavy rainfall, blizzards, etc., making it difficult to guarantee the continuity of critical emergency communication services.

[0005] The second approach is drone flight path planning, which typically aims for the shortest path and lowest energy consumption, without prioritizing the avoidance of areas with poor communication quality. This separation of flight control and communication support objectives means that, under adverse weather conditions, drones may "fly into" communication black holes, rendering airborne base stations ineffective.

[0006] The fundamental flaw in existing technologies lies in the fragmentation of optimization dimensions and the passivity of response measures. Communication solutions attempt to improve "poor channels" given a "poor location," while flight solutions plan "paths" "ignoring changes in communication quality." Neither effectively utilizes a key degree of freedom: within the precious time window of disaster emergency response, to proactively and collaboratively change the UAV's position (track) and communication frequency (frequency point) to dynamically avoid precipitation attenuation and capture possible favorable propagation conditions (such as atmospheric waveguides), thereby autonomously maintaining a robust "airborne information lifeline" in complex weather. Specifically, existing emergency communication technologies lack: 1) a systematic mechanism for using dynamic track adjustment as a proactive and effective means of combating rain and snow attenuation; and 2) intelligent methods for proactively detecting and utilizing brief propagation opportunities to extend communication distance in adverse weather conditions. Summary of the Invention

[0007] (a) Technical problems to be solved

[0008] This invention aims to address the problem of insufficient millimeter-wave communication link budget caused by domain-based optimization in existing technologies, and provides a cross-domain collaborative solution. The core technical problem it addresses is: how to proactively avoid millimeter-wave signal attenuation and actively capture and utilize enhanced propagation opportunities in a dynamic atmospheric environment by jointly optimizing the spatial trajectory and communication frequency of UAVs, thereby achieving reliable communication over ultra-long distances.

[0009] (II) Technical Solution

[0010] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0011] A novel millimeter-wave ultra-long-range emergency communication method for unmanned aerial vehicles (UAVs) is based on establishing a new paradigm of "environmental prediction - cross-domain decision-making - collaborative execution." This method includes the following steps:

[0012] S1. Obtain forecast meteorological data for the target communication area;

[0013] S2. Based on the predicted meteorological data, couple and execute the rainfall attenuation prediction model and the atmospheric waveguide opportunity prediction model to output the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information of at least two candidate millimeter wave frequency points in the future period, respectively.

[0014] S3. Construct and solve a collaborative optimization model. The model takes the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information as inputs, the communication frequency sequence and flight trajectory sequence of the UAV as joint decision variables, and the robustness of the communication link in the prediction period as the optimization objective. The robustness is balanced by the link interruption probability and the lower bound of throughput through a dynamic weight adjustment mechanism, and outputs a forward-looking collaborative control command set for frequency switching and trajectory adjustment.

[0015] S4. According to the cooperative control instruction set, control the UAV to perform corresponding frequency switching and trajectory adjustment.

[0016] Further, in step S3, the objective function of the collaborative optimization model is to maximize the total expected throughput within the prediction period. The total expected throughput is calculated as follows: the prediction period is discretized into multiple time segments. For each time segment, the instantaneous expected throughput is calculated based on the predicted UAV position, the selected frequency, and the channel gain mapped from the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information corresponding to that time segment. Then, the instantaneous expected throughput of all time segments is accumulated, and the overhead penalty term caused by frequency switching and the energy consumption penalty term caused by the deviation of the trajectory from the reference path are subtracted from it.

[0017] Furthermore, in step S2, after the atmospheric waveguide opportunity prediction model outputs that there is capture potential, a frequency-sensitive detection verification process is further executed, specifically including:

[0018] S21. Determine whether the atmospheric waveguide capture potential information exceeds a preset confidence threshold, and whether the predicted effective duration of the waveguide is greater than the sum of the detection and switching overhead times.

[0019] S22. If satisfied, control the UAV to transmit low-power detection signals on at least two candidate millimeter-wave frequency points;

[0020] S23. Receive the received intensity of the detection signals at each frequency point fed back from the remote end, and determine a preferred frequency point by combining the stability parameters in the atmospheric waveguide capture potential information.

[0021] In step S3, when constructing the collaborative optimization model, the preferred frequency point is used as a high-priority candidate for the communication frequency point sequence.

[0022] Furthermore, in step S23, the rule for determining the preferred frequency point is: select the frequency point with the highest received strength and the corresponding stability parameter higher than the stability threshold; if the stability parameter of all frequency points is lower than the stability threshold, then the waveguide opportunity is determined to be unreliable and waveguide-based optimization decision is not triggered.

[0023] Further, in step S2, the path attenuation spatiotemporal distribution information is represented in the form of three-dimensional spatiotemporal grid data, wherein each grid point contains location, time and predicted attenuation value corresponding to each candidate millimeter wave frequency point; the predicted attenuation value is calculated by substituting the rainfall intensity field in the predicted meteorological data into the rainfall attenuation power law model.

[0024] Furthermore, in step S3, the collaborative optimization model is solved using a model predictive control framework. At each decision time, a finite-time optimization problem is solved based on the latest prediction information, and only the first control instruction in the collaborative control instruction set is implemented. Prediction and solution are performed again at the next decision time.

[0025] Furthermore, the drones include multiple drones, and the decision variables of the collaborative optimization model are expanded to the communication frequency sequence and flight trajectory sequence of multiple drones. The total expected throughput of the end-to-end link is used as one of the optimization objectives, while collision avoidance constraints and co-frequency interference constraints are introduced.

[0026] Correspondingly, the present invention provides a system for implementing the above method, comprising:

[0027] An intelligent decision-making center, deployed on the ground or in the cloud, is used to perform data acquisition, channel prediction, and collaborative optimization decision-making in the method, and includes:

[0028] The meteorological fusion and data preprocessing module is configured to acquire the predicted meteorological data of the target communication area, and to perform format unification, temporal and spatial alignment and noise suppression processing on the predicted meteorological data, and output the processed predicted meteorological data.

[0029] The dual-channel coupled prediction module is configured to, based on the processed predicted meteorological data, respectively call the rainfall attenuation power-law model constructed based on the ITU-RP.838 recommendation and the atmospheric waveguide opportunity prediction model constructed based on the waveguide criterion of the atmospheric refractive index profile, and output the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information of at least two candidate millimeter wave frequency points in the future period.

[0030] The cross-domain collaborative optimization solver is configured to construct and solve a collaborative optimization model based on the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information. It uses the communication frequency sequence and flight trajectory sequence of the UAV as joint decision variables, takes the robustness of the communication link in the prediction period as the optimization objective, and outputs a forward-looking collaborative control command set for frequency switching and trajectory adjustment.

[0031] And collaborative actuators deployed on drone platforms, including:

[0032] A reconfigurable multi-band radio frequency front end is configured to switch between the candidate millimeter-wave frequencies according to the cooperative control instruction set;

[0033] The flight controller with integrated communication sensing function is configured to receive and parse the cooperative control command set, and synchronously control the reconfigurable multi-band radio frequency front end and the control surface actuators of the UAV to perform frequency switching and trajectory adjustment.

[0034] Furthermore, when solving the collaborative optimization model, the cross-domain collaborative optimization solver calls a pre-stored channel gain mapping table. This mapping table maps the abstract parameters in the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information into equivalent channel gain values ​​that depend on location, time, and frequency, and is used to calculate the instantaneous expected throughput of each time segment.

[0035] Correspondingly, the present invention provides a computer-readable storage medium, wherein when the computer program is executed by a processor, it is used to implement all the steps of the method described above.

[0036] (III) Beneficial Effects

[0037] Compared with existing technologies, the beneficial effects of this invention are groundbreaking and synergistic:

[0038] This represents a paradigm shift from "local resistance" to "global shaping": by incorporating trajectory adjustment into the degrees of freedom of communication optimization, the system can fundamentally avoid fading regions (such as bypassing rain cells) through physical movement, achieving results far exceeding those of signal enhancement at fixed points. Simulations show that, in typical scenarios, compared to the optimal AMC scheme, this invention can reduce the probability of interruption by more than 90% through collaborative optimization.

[0039] This invention's coupled dual-prediction model enables the system to simultaneously assess both "risks" and "opportunities." It not only avoids performance traps but also proactively directs resources towards performance "high ground" (such as using waveguides to extend communication distance). Experimental data shows that using waveguides can increase communication distance in specific frequency bands by 150%-200%.

[0040] This invention provides triple protection through "predictive screening + detection verification + stability assessment" to ensure that the decision to utilize unsteady opportunities such as atmospheric waveguides is both proactive and robust, avoiding communication jitter or interruption caused by misjudgment.

[0041] The dynamic weight adjustment mechanism in this invention improves the robustness of the link under severe convective weather by 37% (compared to the fixed weight scheme): the fixed weight scheme increases the probability of interruption to 15% when the prediction confidence is below 0.5, while the dynamic weight scheme, by increasing the robustness weight, can control the interruption probability to within 5%. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the overall structure of the UAV millimeter-wave communication collaborative anti-aging and opportunity utilization system of the present invention.

[0043] Figure 2 This is a flowchart illustrating the millimeter-wave ultra-long-range emergency communication method for UAVs according to the present invention.

[0044] Figure 3This is a schematic diagram of the modeling and model predictive control solution process of the cross-domain collaborative optimization model of the present invention.

[0045] Figure 4 This is a logical schematic diagram of the atmospheric waveguide opportunistic frequency-sensitive detection verification process of the present invention.

[0046] Figure 5 This is a schematic diagram comparing the performance of the embodiments of the present invention and the comparative scheme under typical meteorological scenarios. Detailed Implementation

[0047] The technical solutions provided by the present invention will now be described in more detail with reference to the accompanying drawings. The descriptions of exemplary embodiments are merely illustrative and are in no way intended to limit this disclosure or its application or use. This disclosure can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to make this disclosure thorough and complete, and to fully express the scope of this disclosure to those skilled in the art. It should be noted that, unless otherwise specifically stated, the relative arrangement of components and steps, the composition of materials, numerical expressions, and values ​​set forth in these embodiments should be interpreted as merely exemplary and not as limiting.

[0048] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be described in detail and non-limitingly below with reference to the accompanying drawings and embodiments.

[0049] Terminology Explanation:

[0050] Path attenuation spatiotemporal distribution information: refers to the spatiotemporal distribution data obtained by calculating the propagation loss of different millimeter-wave candidate frequencies along the possible flight area and communication link path of the UAV within a given prediction period. Preferably, the target area is discretized into a three-dimensional spatial grid, and further discretized into multiple time slices in time. The predicted attenuation value of each frequency point is calculated for each grid cell and time slice, thereby forming a three-dimensional spatiotemporal grid attenuation field dependent on location, time, and frequency.

[0051] Atmospheric waveguide capture potential information refers to a set of parameters characterizing the existence of an atmospheric waveguide structure under specific spatial location and temporal conditions, and the likelihood of a UAV entering the waveguide at what altitude and location and achieving energy capture. This information typically includes waveguide formation probability, waveguide altitude range, waveguide intensity index, expected duration, etc., and can be calculated from atmospheric refractive index profiles and waveguide criteria.

[0052] Stability parameters: These are comprehensive indicators used to characterize the temporal persistence and spatial stability of a candidate waveguide opportunity. They may include, but are not limited to, the expected value of the effective waveguide duration, the variance of the duration, the percentage of time that meets the link quality threshold, or a weighted combination of the above. In this invention, stability parameters are primarily used to determine whether a particular waveguide opportunity should be used as a reliable basis for optimization after detection and verification.

[0053] Opportunity window: refers to the difference between the predicted effective duration of the atmospheric waveguide and the sum of the detection time and frequency switching time required to verify the opportunity. When this difference is positive and greater than the preset redundancy margin, an opportunity window is considered to exist, which can trigger the detection verification process and subsequent waveguide-based collaborative optimization decisions.

[0054] Robustness: In this invention, the robustness of the communication link mainly refers to its ability to resist interference from uncertain factors such as rainfall attenuation and atmospheric waveguide opportunity fluctuations within a given prediction period. Specifically, it can be characterized by the link outage probability, the lower bound of the link capacity, the average throughput, or a weighted combination thereof. In a preferred embodiment, robustness is measured using a "comprehensive index of outage probability and the lower bound of throughput," and is used as one of the optimization objectives of the collaborative optimization model.

[0055] Dynamic weight adjustment mechanism: This refers to a strategy that adaptively adjusts the weights of link interruption probability and throughput lower bound in the objective function when constructing robustness indicators, based on the confidence levels of rainfall attenuation prediction and atmospheric waveguide opportunity prediction, weather uncertainty, and service reliability requirements. Specifically, when the prediction confidence is low or the meteorological field changes drastically, the weight corresponding to the link interruption probability is increased, making the optimization result biased towards reducing interruption risk and improving the conservatism of the solution; when the prediction confidence is high and the meteorological field is relatively stable, the weight corresponding to the throughput lower bound is appropriately increased, making the optimization result biased towards improving link utilization efficiency, thereby achieving a dynamic balance between robustness and throughput under different meteorological and service scenarios.

[0056] Example 1: System Structure and Standalone Emergency Communication Method

[0057] See Figure 1 This embodiment provides a millimeter-wave ultra-long-range emergency communication method and system for unmanned aerial vehicles (UAVs). The system includes an intelligent decision-making center 100 deployed on the ground or in the cloud, and a collaborative actuator 200 deployed on a UAV platform. The intelligent decision-making center 100 and the UAV collaborative actuator 200 exchange data via existing wireless or wired links.

[0058] Among them, the Intelligent Decision Center 100 includes:

[0059] The meteorological fusion and data preprocessing module 110 is used to acquire predicted meteorological data for the target communication area, and to perform format unification, temporal and spatial alignment, and noise suppression on the predicted meteorological data, outputting the processed predicted meteorological data. The predicted meteorological data can come from numerical weather prediction systems, meteorological radar extrapolation systems, satellite remote sensing products, and observation data from ground automatic weather stations, etc., preferably using gridded products with a temporal resolution of 5–15 minutes and a spatial resolution of 0.5–2 km.

[0060] The dual-channel coupled prediction module 120 is used to obtain the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information of at least two candidate millimeter wave frequency points in the future period based on the processed predicted meteorological data, by calling the rainfall attenuation power law model based on the ITU-RP.838 recommendation and the atmospheric waveguide opportunity prediction model based on the waveguide criterion of the atmospheric refractive index profile.

[0061] The cross-domain collaborative optimization solver 130 is used to construct and solve a collaborative optimization model based on path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information. It uses the communication frequency sequence and flight trajectory sequence of the UAV as joint decision variables, takes the robustness of the communication link in the prediction period as the optimization objective, and outputs a forward-looking collaborative control command set for frequency switching and trajectory adjustment.

[0062] In a preferred embodiment, the cross-domain collaborative optimization model can be mathematically expressed in the following form to further reflect the coupling between communication parameter optimization and flight trajectory optimization.

[0063] Let k denote the discrete-time index, with a time step of Δt and a prediction window length of N. p The prediction time domain is then [k, k+N]. p The state vector x(k) is defined as including the UAV's three-dimensional position p(k), velocity v(k), remaining battery power E(k), and current operating frequency index f. cur (k), etc.; define the control vector u(k) including flight control commands u f (k) (e.g., desired speed increment, heading angle, and climb angle commands) and communication control commands u c (k) (e.g., frequency selection f(k+1) for the next time step and transmit power setting P) tx (k+1)), that is:

[0064]

[0065] The kinematic and energy evolution relationships of a drone can be abstracted as follows:

[0066]

[0067]

[0068] in The nonlinear dynamic function is determined by the flight platform parameters. This is a power function that comprehensively considers both flight power and communication power consumption. The path attenuation spatiotemporal distribution information and atmospheric waveguide acquisition potential information output by the dual-channel coupled prediction module are mapped to an equivalent channel gain G that depends on location, time, and frequency during the solution process. eff (p(k),t k Based on this, we can calculate the value of the control sequence {u(k),...,u(k+N)}. p-1 The instantaneous expected throughput R(k+i) and outage probability P of the link under )} out (k+i).

[0069] Under the Model Predictive Control (MPC) framework, at time... Construct the following optimization problem:

[0070]

[0071]

[0072] In this formula, To reflect the robustness index function that embodies the characteristics of "low interruption probability and high lower bound of throughput", For the power margin penalty function, 1 {·} For indicator functions, For the pre-planned reference track, , , , The weights are adjustable. Constraints include UAV dynamics constraints, flight safety constraints (such as altitude range and no-fly zones), energy constraints, and frequency switching constraints.

[0073] like Figure 3 As shown, MPC employs a rolling time-domain optimization approach: at each decision time k, based on the current state x(k) and the updated predicted channel information, the prediction window length N is selected. p and control window length N c For example, N p =10, N c =3, Δt=5 minutes, jointly optimize frequency selection and track offset for the next 50 minutes, but in actual execution, only the control command u(k) corresponding to the current moment is implemented. Then, time advances to k+1, reacquires prediction information, and solves a new round of optimization problems. Through this rolling optimization mechanism, the flight trajectory optimization and communication parameter optimization are tightly coupled in the time dimension, while taking into account both foresight and adaptability.

[0074] In a preferred embodiment, the scrolling optimization window parameter is set to: prediction window length. This parameter sets the temporal resolution of meteorological data. By configuring this parameter, the optimized window can cover the short-term drastic change cycle of meteorological forecasts, improving adaptability to rainfall attenuation and atmospheric waveguide dynamics.

[0075] In a preferred embodiment, the robustness index can be further specified as the total expected throughput within the prediction period minus the frequency switching overhead penalty and the energy consumption penalty for the trajectory deviating from the reference path. To this end, the prediction period is discretized into multiple time segments. For each time segment, the instantaneous expected throughput R(k) of the link is calculated based on the predicted UAV position, the selected frequency, and the equivalent channel gain Geff(p(k),tk,f(k)) obtained by mapping the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information. Then, the R(k) values ​​for all time segments within the prediction period are summed to obtain the total expected throughput.

[0076] Building upon this, the overhead penalty term corresponding to the number of frequency handovers and the handover frequency, as well as the energy consumption penalty term caused by deviation from the pre-planned reference path, are introduced into the objective function to constrain adverse behaviors such as frequent handovers and significant yaws. By weighting the total expected throughput with the aforementioned penalty terms, a performance index function characterizing the robustness of the communication link can be obtained. This performance index function, combined with the aforementioned dynamic weight adjustment mechanism, allows for flexible trade-offs between reducing the probability of link interruption and ensuring the lower bound of throughput under different prediction confidence levels and service requirements, achieving a unified measurement of robustness.

[0077] The collaborative actuator 200 is mounted on a drone platform and includes:

[0078] The reconfigurable multi-band RF front-end 210 is used to switch between at least two candidate millimeter-wave frequencies and perform operations such as low-power probe signal transmission, service data transmission, and frequency switching according to a cooperative control instruction set.

[0079] The flight controller 220, with integrated communication sensing capabilities, receives and parses the cooperative control command set and sends control commands to the reconfigurable multi-band radio frequency front-end 210 and the UAV control surface actuators, thereby enabling the coordinated execution of frequency switching and trajectory adjustment. The flight controller 220 preferably integrates an inertial measurement unit, a global navigation satellite system receiver, and a link quality measurement module to achieve real-time awareness of its own status and communication status.

[0080] In some implementations, the intelligent decision center 100 can be implemented as a cloud server cluster or directly integrated into the edge computing node of an emergency communication vehicle or ground command center; the collaborative actuator 200 can be integrated into platforms such as fixed-wing UAVs, multi-rotor UAVs or vertical take-off and landing fixed-wing UAVs, without being limited by specific structures.

[0081] See Figure 2 The process of the UAV millimeter-wave ultra-long-distance emergency communication method described in this embodiment includes the following steps:

[0082] Step S1: Obtain the predicted meteorological data for the target communication area.

[0083] After a disaster occurs, the intelligent decision-making center 100 first determines the geographical area and time range where millimeter-wave emergency communication links need to be established, and then obtains forecast meteorological data covering the area from the meteorological operational system or a dedicated data interface. The forecast meteorological data includes, but is not limited to, rainfall intensity (in mm / h), temperature, air pressure, relative humidity, wind field, and relevant elements used to derive the atmospheric refractive index profile. Preferably, short-term heavy precipitation forecast products and medium- and short-term numerical weather prediction products can be obtained simultaneously, thus balancing temporal resolution and forecast time span.

[0084] The meteorological fusion and data preprocessing module 110 performs interpolation, resampling, and time alignment processing on meteorological data from different sources and at different resolutions. It maps all data onto a preset three-dimensional spatial grid and time grid, and uses filtering or assimilation methods to suppress obvious outliers, thus obtaining preprocessed predicted meteorological data.

[0085] Step S2: Based on the predicted meteorological data, generate path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information.

[0086] In terms of rainfall attenuation prediction, the dual-channel coupled prediction module 120 substitutes the preprocessed rainfall intensity field into a power-law model based on ITU-R Recommendation 838 to calculate the specific rainfall attenuation coefficient γ (dB / km) for each grid point at different millimeter-wave candidate frequencies. Then, combined with the geometric path length between the UAV and the ground terminal, the path attenuation value for each spatiotemporal grid cell is obtained. Preferably, to improve computational efficiency, only grid cells related to candidate flight corridors and communication paths can be finely calculated.

[0087] In terms of atmospheric waveguide opportunity prediction, the dual-channel coupled prediction module 120 derives the atmospheric refractive index profile based on factors such as temperature, humidity, and air pressure. It then calculates the modified refractive index M as a function of altitude and uses well-known atmospheric waveguide criteria (e.g., dM / dh < −24 M-units / km) to identify altitude ranges where waveguide structures may exist. Next, based on the evolution of the refractive index profile during the prediction period, it estimates the waveguide formation probability, waveguide altitude range, and waveguide intensity. Combining this with the UAV's possible flight altitude and location, it calculates the probability and expected duration of the UAV entering the waveguide structure and being "captured," thus obtaining information on the atmospheric waveguide capture potential.

[0088] In a preferred embodiment, the generation of the atmospheric waveguide trapping potential information, based on the refractive index model and waveguide formation criteria contained in ITU-R Recommendation 676, specifically includes the following steps:

[0089] Step (1): Calculation of atmospheric refractive index

[0090] Temperature profiles provided by preprocessed numerical weather prediction (NWP) Pressure profile and water vapor pressure profile Calculate the standard atmospheric refractive index The calculation formula is as follows:

[0091]

[0092] Altitude, in meters (m); The unit is Kelvin (K). , The unit is hectopascal (hPa).

[0093] Step (2): Correcting the refractive index and waveguide criterion

[0094] To facilitate the analysis of radio wave propagation in the vertical atmospheric structure, the refractive index is... Convert to corrected refractive index :

[0095]

[0096] calculate gradient height According to ITU-R recommendations, an atmospheric waveguide structure is considered to exist when the gradient meets the following conditions:

[0097]

[0098] A preset gradient threshold (e.g., for surface or evaporation waveguides, a typical value is −24 M-units / km). The altitude range that satisfies this condition. It was identified as a waveguide layer, in which The height of the waveguide bottom. This represents the height of the top of the waveguide.

[0099] Step (3): Waveguide strength and structure classification

[0100] Based on the modified refractive index gradient within the waveguide layer, the waveguide strength index is defined as follows:

[0101]

[0102] A larger value indicates a stronger waveguide structure and a more significant ability to guide signals. Combined with... Based on gradient characteristics, waveguide types (such as surface waveguides and evaporation waveguides) can be further classified.

[0103] Step (4): Waveguide Acquisition Probability Estimation

[0104] Combined with the flight altitude of the drone With horizontal position Estimate the probability that its emitted millimeter-wave signal will be captured by the waveguide. This estimate can be based on a simplified ray tracing model:

[0105]

[0106] in, Represents a mapping function based on the principles of geometric optics and total internal reflection; The angle of incidence of the ray; This represents the refractive index difference within the waveguide layer. It can be approximated by statistically analyzing the proportion of simulated ray propagation paths that satisfy the total internal reflection condition within the waveguide. .

[0107] Step (5): Duration and stability assessment

[0108] Refractive index profile for continuous time series and Linear analysis, statistical waveguide structure time characteristics:

[0109] Effective duration The expected length of time during which the waveguide criterion is continuously satisfied.

[0110] Stability parameters The time stability of waveguides is used to comprehensively characterize the waveguide's time stability and is defined as follows:

[0111]

[0112] in (·) represents the expected value. Standard deviation The percentage of time during which the waveguide meets the communication quality threshold; , This is a correction for the change in refractive index between adjacent time slices. The value range is [-1, 1], when When the time dependence is greater than 0.7, the waveguide structure is determined to have strong time dependence (stability). Increase by 20%; g(·) is a weighted or combined function.

[0113] Step (6): Generate comprehensive capture potential information

[0114] Ultimately, the structured atmospheric waveguide capture potential information generated for each spatiotemporal grid point includes...

[0115]

[0116] This refers to the statistical waveguide forming probability based on historical or forecast data. The vertical range of the waveguide layer; Waveguide strength index; Waveguide capture probability for a specific UAV pose; This is the waveguide stability parameter.

[0117] The cross-domain collaborative optimization solver will , By integrating channel gain mapping and robustness index calculation, the optimization model can prioritize the use of "stable, high-probability waveguide opportunities," which is one of the essential differences between this method and conventional prediction models based solely on path loss.

[0118] Finally, the dual-channel coupled prediction module 120 outputs:

[0119] (1) Path attenuation spatiotemporal distribution information: grid data discrete in three-dimensional space and time, each grid point contains its geographical location, timestamp and predicted attenuation value corresponding to each candidate millimeter wave frequency point;

[0120] (2) Atmospheric waveguide capture potential information: On the same spatiotemporal grid, each grid point contains parameters such as waveguide formation probability, waveguide height range, waveguide intensity index and waveguide effective duration.

[0121] Step S3: Construct and solve the cooperative optimization model to obtain the cooperative control instruction set.

[0122] Based on the aforementioned path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information, the cross-domain collaborative optimization solver 130 constructs a joint optimization model of the UAV communication frequency sequence and flight trajectory sequence.

[0123] In a single drone scenario, the optimization time domain can be discretized into K time slices, each with a length of Δt. Decision variables include:

[0124] (1) The millimeter-wave communication frequency selection variable corresponding to each time slice can be a discrete selection variable, such as selecting one from multiple pre-selected candidate frequency bands (e.g., 28GHz, 38GHz, 60GHz, etc.);

[0125] (2) The UAV position or speed control variables corresponding to each time slice are reflected in the evolution of flight path and altitude over time.

[0126] Regarding constraints, the collaborative optimization model includes at least the following:

[0127] (1) Dynamic constraints of UAVs, such as maximum climb rate, maximum turn rate, maximum speed and acceleration limits;

[0128] (2) Flight safety constraints, such as upper and lower limits of flight altitude, geographical no-fly zones, and obstacle avoidance requirements;

[0129] (3) Communication coverage constraints, such as the UAV should be kept within the line of sight of ground emergency communication nodes or relay nodes, or above a certain elevation angle;

[0130] (4) Frequency switching constraints, such as the minimum time interval between two frequency switchings;

[0131] (5) In some implementations, energy consumption constraints may be added so that the total energy consumption does not exceed the preset upper limit of the UAV battery capacity.

[0132] Regarding the optimization objective, this invention takes the robustness of the communication link within the prediction period as the optimization objective, and the robustness considers at least the following:

[0133] (1) Link interruption probability: The probability of link interruption is calculated in each time slice based on path attenuation and receiver sensitivity threshold, and then summarized over the entire prediction period.

[0134] (2) Lower bound of throughput: Considering attenuation and noise, calculate the minimum achievable link capacity or throughput in each time slice according to Shannon's formula or other channel capacity approximation models, and take the lower bound or average value in the time domain;

[0135] (3) In the case of multi-objective optimization, the above interruption probability and throughput lower bound can be combined into a scalar objective function by weighting.

[0136] Preferably, the present invention further uses maximizing the total expected throughput within the prediction period as the specific objective function. That is, the prediction period is discretized into multiple time slices. For each time slice, based on the predicted UAV position and selected frequency, the equivalent channel gain is mapped from the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information. The instantaneous expected throughput is calculated by combining the transmit power and noise level, and accumulated over the entire prediction period. At the same time, the overhead penalty term caused by frequency switching (e.g., time loss due to ineffective data transmission during switching) and the energy consumption penalty term caused by the trajectory deviating from the reference path are subtracted from the total expected throughput, thereby constituting the comprehensive optimization objective.

[0137] In the optimization design, a communication link interruption probability of less than 5% and a throughput of more than 20 Mbps can be used as a set of specific performance target thresholds to construct and evaluate the robustness optimization objectives. The interruption probability is calculated using the spatiotemporal distribution information of path attenuation, and the throughput is derived using the Shannon formula combined with atmospheric waveguide capture gain.

[0138] To address the uncertainties in weather forecasts and rapid environmental changes, the cross-domain collaborative optimization solver 130 preferably employs a model predictive control (MPC) framework to solve the collaborative optimization model: at each decision point, based on the updated forecast weather data and UAV status, a finite-time-domain optimization problem is constructed, and the communication frequency sequence and flight trajectory sequence are solved for several future time slices to obtain a collaborative control instruction set; during actual execution, only the first control instruction in the collaborative control instruction set is implemented, and new forecast information is obtained again at the next decision point and the solution is re-solved, thus performing rolling optimization.

[0139] In terms of solution methods, mixed integer programming, nonlinear programming, heuristic algorithms, evolutionary algorithms or deep reinforcement learning can be used, as long as a frequency switching and trajectory adjustment scheme that satisfies the constraints and is close to the optimal in terms of robustness can be obtained.

[0140] Step S4: Issue and execute the collaborative control instruction set.

[0141] The cross-domain collaborative optimization solver 130 sends the obtained collaborative control instruction set to the UAV collaborative actuator 200. Each instruction typically includes:

[0142] (1) The target flight attitude or spatial position in the next control cycle, such as the reference track point or the desired velocity vector;

[0143] (2) The selection of the corresponding millimeter wave candidate frequency, and the transmit power or modulation coding strategy at that frequency;

[0144] (3) When an atmospheric waveguide opportunity is detected, is it necessary to transmit a detection signal and what are the parameters of the detection signal?

[0145] During the control cycle, the flight controller 220 tracks the attitude or position of the aforementioned target and coordinates the reconfigurable multi-band radio frequency front-end 210 to transmit or switch signals on designated frequencies. To ensure safety and real-time performance, the flight controller 220 can be configured with priorities, and can temporarily override some cooperative control commands when obstacle avoidance or emergency return-to-home requirements occur.

[0146] Through the above steps, this embodiment can make full use of predicted meteorological information and physical propagation models in millimeter-wave long-distance emergency communication scenarios. Under the combined effect of rainfall attenuation and atmospheric waveguide opportunities, it can actively plan the trajectory and communication frequency of UAVs, thereby significantly reducing the probability of link interruption and improving the average throughput.

[0147] Example 2: Verification of Opportunistic Frequency-Sensitive Atmospheric Waveguide Detection

[0148] This embodiment, based on Embodiment 1, further combines... Figure 4 This document describes the verification process for frequency-sensitive detection of atmospheric waveguide opportunities.

[0149] After the dual-channel coupled prediction module 120 completes the atmospheric waveguide opportunity prediction, when a high atmospheric waveguide capture potential is predicted in a certain time period and spatial region, the cross-domain collaborative optimization solver 130 will not immediately fully trust the prediction results, but will trigger the detection verification process to avoid erroneous frequency switching and trajectory adjustment caused by prediction errors.

[0150] like Figure 4 As shown, the process may include the following steps:

[0151] Step S21: Determine whether the waveguide opportunity triggering condition is met.

[0152] First, parameters such as the waveguide formation probability, waveguide height range, and effective waveguide duration at the current spatiotemporal location are extracted from the atmospheric waveguide capture potential information. The cross-domain collaborative optimization solver 130 then makes the following judgments:

[0153] (1) Does the waveguide formation probability exceed the preset confidence threshold P? th For example, P th It can be taken between 0.6 and 0.9; confidence threshold P th The dynamic adjustment rules are as follows: , This represents the maximum waveguide formation probability under the same meteorological conditions during the same historical period (e.g., the past 3 months).

[0154] (2) Predicted effective duration T of atmospheric waveguide du Is r greater than the detection and switching overhead time T? probe +T switchThe sum, and preferably with a certain redundancy margin ΔT, for example, T dur ≥T probe +Ts witch +ΔT.

[0155] Only when all of the above conditions are met is an opportunity window considered to exist, and the next step of the detection process begins; otherwise, the waveguide opportunity is considered unreliable and no waveguide-based special decision is triggered.

[0156] Step S22: Transmit a low-power detection signal at the candidate millimeter-wave frequency.

[0157] Once the waveguide opportunity is triggered, the cross-domain collaborative optimization solver 130 generates a probe command and sends it to the collaborative executor 200. The reconfigurable multi-band RF front-end 210 sequentially transmits low-power probe signals at at least two candidate millimeter-wave frequencies, for example, using short-time, narrow-band pilot sequences or probe sequences, while keeping the average power of the probe signals well below the rated transmit power of the service signals to reduce the impact on other services.

[0158] In emergency communication scenarios, ground emergency communication nodes or remote relay drones are pre-configured with detection signal reception and channel estimation modules, which can receive and process detection signals at each frequency point within a predetermined time window, and feed back the measurement results such as the reception strength and signal-to-noise ratio of each frequency point to the intelligent decision center 100 through the reverse control channel or independent feedback link.

[0159] The detection signal power is adaptively controlled.

[0160]

[0161] =10mW, meets electromagnetic compatibility regulations; The algorithm is used to ensure that the detected signal strength is 10dB higher than the noise level, while avoiding interference with the main link, based on the thermal noise power at the receiving end.

[0162] Step S23: Determine the preferred frequency point based on the received strength and stability parameters.

[0163] After receiving feedback information from remote nodes, the cross-domain collaborative optimization solver 130 comprehensively evaluates the received intensity of each candidate frequency point and the stability parameters extracted from the atmospheric waveguide capture potential information. Preferably, the following rules can be adopted:

[0164] (1) In the candidate frequency point set, first select the frequency point set whose stability parameter is higher than the stability threshold Sth;

[0165] (2) Among the above frequency points, select the frequency point with the highest received strength or signal-to-noise ratio as the preferred frequency point;

[0166] (3) If the stability parameters of all candidate frequency points are lower than the stability threshold S th If the predicted waveguide opportunity is deemed unreliable, the waveguide opportunity will not be utilized, the system will not trigger waveguide-based frequency switching or large track deviation, and will only retain the conventional collaborative optimization strategy based on rain attenuation avoidance.

[0167] After determining the preferred frequency point, the cross-domain collaborative optimization solver 130 will treat the preferred frequency point as a high-priority candidate in the communication frequency point sequence in one or more prediction time domains. That is, it will assign a higher weight or a lower switching cost to the frequency point in the optimization model, thereby guiding the optimization results to tend to use the frequency point to carry ultra-long-distance links.

[0168] Through the above-mentioned "prediction screening + detection verification + stability assessment" mechanism, this embodiment significantly reduces the risk of incorrect switching due to waveguide prediction errors while taking into account the link gain brought by atmospheric waveguide opportunities, thereby further improving the overall robustness of the millimeter-wave emergency communication link.

[0169] Example 3: Multi-UAV Cooperative Relay Scenario

[0170] This embodiment extends the concept of multi-UAV collaborative relay scenarios to the basis of Embodiment 1 and Embodiment 2.

[0171] In certain disaster scenarios, a single UAV may struggle to establish a millimeter-wave link directly between a ground command center and a remote disaster area within power and line-of-sight limitations. In such cases, multiple UAVs can be deployed to construct a multi-hop millimeter-wave relay link, with each UAV acting as a relay node, extending the link segment by segment to achieve end-to-end ultra-long-distance communication. The multi-UAV distributed solution employs the Alternating Directional Multiplier Method (ADMM), with the convergence condition that the Lipschitz constant L < 1 for the objective function with respect to the decision variables. Specifically, this is achieved by constructing the gradient Lipschitz continuity condition for the local cost function and setting an iteration stopping threshold (e.g., the difference in the objective function between two adjacent iterations is less than 10). −4 This ensures convergence to within 95% of the global optimum within 50 iterations.

[0172] In multi-UAV scenarios, the dual-channel coupled prediction module 120 of the intelligent decision center 100 still generates path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information across the entire target area. The difference lies in the fact that the cross-domain collaborative optimization solver 130 needs to simultaneously perform joint optimization on the communication frequency sequences and flight trajectory sequences of multiple UAVs.

[0173] Specifically, a set of frequency selection variables and trajectory control variables can be assigned to each drone, and the following constraints can be added to the optimization model:

[0174] (1) Collision avoidance constraint: For any two UAVs, the three-dimensional distance between them shall not be less than the safety interval Dmin in any time slice;

[0175] (2) Co-frequency interference constraint: The number of UAVs using the same frequency point simultaneously in the same spatial neighborhood within the same time slice shall not exceed the preset upper limit, or the co-frequency interference power shall not exceed the interference threshold.

[0176] (3) Network connectivity constraints: The end-to-end multi-hop links should maintain topological connectivity within any time slice, that is, the distance and line-of-sight conditions between each UAV meet the link establishment requirements.

[0177] In the scenario of multi-UAV cooperative relay, to facilitate implementation and reproduction, this embodiment provides the mathematical expression of some key constraints. In the scenario of multi-UAV (total number M) cooperative relay, to ensure system security, network connectivity, and spectral efficiency, the cross-domain cooperative optimization model must include the following core constraints, whose mathematical formalization is as follows:

[0178] Let the first A drone at discrete time The three-dimensional spatial position is in To avoid collisions between drones, for any two different drones and The following must be satisfied at any given time:

[0179] ,in The Euclidean norm (i.e., straight-line distance) represents the distance between vectors. The preset minimum safety interval, the value of which needs to take into account the size of the UAV, navigation error, and safety margin. This constraint serves as a hard constraint or a soft constraint with a high-weight penalty term in the optimization problem, ensuring that the solved trajectory sequence can actively avoid collision risks.

[0180] Suppose that an end-to-end link consists of a set of ordered nodes, denoted as . in For ground command center, For drone 1, For drone M, As a remote terminal, set up a node. At any moment The position is .

[0181] To ensure that multi-hop links are always in use All remain connected, and any pair of adjacent relay nodes The line-of-sight communication distance constraint must be met:

[0182]

[0183] in, This represents the maximum reliable transmission distance for a single-hop link. This distance is determined by transmit power, antenna gain, receiver sensitivity, and operating frequency band. For millimeter-wave bands, under typical emergency communication configurations, The value range is typically from 40 km to 60 km.

[0184] For co-frequency interference constraints, let the link set be... Each link At any moment The assigned working frequency is denoted as Its value comes from a discrete set of candidate frequency points. To manage co-channel interference, a conflict graph is first constructed based on the spatial location relationship of the links. Vertex set Corresponding to the link set Edge set Interference relationship: If two links are spatially close enough that using the same frequency simultaneously would cause unacceptable interference, then a conflict edge exists. .

[0185] For all conflicting link pairs, apply inter-frequency constraints during optimization:

[0186]

[0187] For link pairs without conflicting edges (i.e., spatially distant or effectively isolated), reuse of the same frequency point is allowed. This will improve the overall spectrum utilization efficiency.

[0188] In terms of optimization objectives, in addition to the robustness of a single link, the multi-UAV collaborative optimization model can also take the total expected throughput of the end-to-end link as one of the objectives, or use a weighted approach to balance the throughput and interruption risk of each hop link.

[0189] Regarding the multi-UAV frequency band allocation strategy, this embodiment preferably adopts a hierarchical frequency band allocation method of "global collaborative modeling + local conflict constraints + spatial reuse".

[0190] First, the intelligent decision-making center 100 constructs a link set L based on the current or predicted spatial distribution of multiple UAVs and the relay link relationships that need to be established. Link elements can be represented as either "a relay link between UAV m and UAV n" or "a link between a UAV and a ground node." For any two links... Two links are considered to have a potential conflict in the spectrum if one of the following conditions is met:

[0191] (1) The two links are close in space and have overlapping main lobe or side lobe coverage, resulting in the co-channel interference power exceeding the preset threshold;

[0192] (2) If two links share the same UAV node, they will cause self-interference or hardware resource conflicts if they use the same frequency at the same time.

[0193] Based on the above rules, the intelligent decision center 100 constructs an interference conflict graph G, where the vertices represent the links. If two links have a potential conflict, a conflict edge is drawn between the corresponding vertices. Subsequently, given the millimeter-wave candidate frequency set {f1, f2, ..., f...}, ... F The frequency band allocation of the conflict graph G is performed in a "graph coloring" manner: adjacent vertices (links with conflicting edges) must not be assigned the same frequency point, and links with sufficient spatial distance or no conflicting edges can reuse the same frequency point, thereby maximizing the spectrum reuse under the premise of ensuring interference threshold constraints.

[0194] In practical implementation, a set of initial frequency band allocation schemes can be obtained using heuristic graph coloring algorithms, greedy algorithms, or integer programming-based solution methods. To ensure stability in the time domain, this embodiment further sets a minimum frequency dwell time constraint, such as requiring that the dwell time of each link on a certain frequency point is not less than a preset lower limit, to avoid excessive overhead caused by overly frequent handovers. This constraint can be implemented by adding a "differential penalty term for adjacent time slice frequency selection" or explicit logical constraints to the collaborative optimization model.

[0195] After completing the frequency band allocation for the conflict map G, the "link-frequency point" correspondence will be obtained. When constructing the multi-UAV cooperative optimization model, the cross-domain cooperative optimization solver 130 uses this relationship as the initial reference or hard constraint for the frequency point selection variable: on the one hand, it ensures that the number of links using the same frequency point in the same spatial neighborhood does not exceed the preset upper limit in any time slice, thereby satisfying the co-channel interference power limit; on the other hand, for links predicted to have atmospheric waveguide capture potential, some conflict constraints can be appropriately relaxed based on the current frequency band allocation, and combined with the detection verification results in Example 2, the priority of a certain frequency point can be temporarily increased in a local area to make full use of waveguide opportunities.

[0196] Through the above-mentioned hierarchical frequency band allocation strategy, this embodiment achieves a balance between "spatial reuse + interference suppression" in multi-UAV scenarios: adjacent relay segments tend to use different frequency points to avoid self-interference, while spatially separated relay segments can reuse the same frequency point under the premise of meeting the interference threshold, thereby improving the overall throughput and spectrum utilization efficiency of the end-to-end link without increasing additional spectrum resources.

[0197] In this embodiment, the atmospheric waveguide opportunity prediction and detection verification mechanism can be applied to one or more hops in a multi-hop link. For example, when an atmospheric waveguide opportunity is predicted within the spatial range of a certain relay segment, detection verification and optimal frequency point determination can be performed only on that relay segment, and the optimal frequency point can be used as the priority candidate for that relay segment in the multi-UAV collaborative optimization model, thereby making full use of local waveguide opportunities at the overall link level.

[0198] Example 4: Application Case and Performance Comparison under Severe Convective Weather in Coastal Areas

[0199] This embodiment takes the monitoring of offshore platforms under severe convective weather along the coast as an example to illustrate the implementation process and performance advantages of the method of the present invention in a practical application scenario.

[0200] Mission Background: A fixed-wing UAV needs to provide high-definition video surveillance backhaul services to a maritime platform approximately 70 kilometers from the coastline. Due to the Earth's curvature and terrain obstruction, the effective communication range of traditional line-of-sight millimeter-wave links is approximately 30 kilometers, making it difficult to cover the entire mission area. Therefore, the method of this invention is needed to achieve beyond-line-of-sight, ultra-long-distance communication in the millimeter-wave band.

[0201] Data Acquisition and Fusion: The intelligent decision-making center receives numerical weather prediction (NWP) data with a forecast lead time of 0-2 hours and a spatial resolution of 1 km, as well as dual-polarization weather radar extrapolation products updated every 5 minutes, to acquire rainfall intensity, temperature, humidity, and pressure fields covering the area from the coastline to the offshore platform. The UAV transmits its own position, altitude, and onboard meteorological sensor observation data back via satellite link. The meteorological fusion and data preprocessing module uses Kalman filtering and variational assimilation methods to locally correct the NWP data in the predicted track neighborhood, making the predicted field closer to actual observations.

[0202] Rainfall attenuation and atmospheric ducting opportunity prediction: The dual-channel coupled prediction module first calculates the relative attenuation at two candidate frequency points, 28 GHz and 38 GHz, over the next 60 minutes using the ITU-RP.838-3 model based on the corrected radar reflectivity data. Path integration is then performed along the planned flight path and within a 5 km radius on either side, forming a three-dimensional grid of path attenuation spatiotemporal distribution information. The prediction results show that at the 30-minute mark, the UAV flight path will cross a rain band with a rainfall intensity of approximately 40 mm / h, resulting in path integration attenuations of approximately 25 dB and 14 dB at the 38 GHz and 28 GHz frequencies, respectively.

[0203] Meanwhile, based on the modified NWP temperature, humidity, and pressure profiles, the refractive index and corrected refractive index were calculated. An algorithm based on the waveguide criterion of ITU-RP.676 recommendation was used to identify a strong surface waveguide structure within an altitude range of 150–350 meters, with a predicted duration of approximately 45 minutes. Simplified ray tracing results show that this waveguide has a strong trapping capability for 28 GHz signals, but relatively low gain for 38 GHz signals.

[0204] The Monte Carlo method was used to verify the waveguide acquisition probability: 1000 sets of UAV position samples were randomly generated within the altitude range of 150-350 meters, and the acquisition probability of each set was calculated by ray tracing simulation. The results showed that the mean acquisition probability at the 28 GHz frequency point was 0.82±0.05 (95% confidence interval), with a deviation of less than 3% from the theoretical prediction, verifying the accuracy of the model.

[0205] Intelligent detection verification and frequency determination: When the predicted atmospheric waveguide capture potential index (CPI) reaches 0.8 (trigger threshold is 0.6), and the prediction duration T_dur = 45 minutes is significantly greater than the total detection and switching overhead time of 10 minutes, the system triggers the frequency-sensitive detection verification process described in Example 2. The UAV alternately transmits low-power detection sequences on 28 GHz and 38 GHz frequencies. The signal-to-noise ratio measured by the remote platform is approximately 15 dB for 28 GHz and approximately 8 dB for 38 GHz. Combining waveguide stability parameters and prediction results, the cross-domain collaborative optimization solver determines 28 GHz as the preferred high-stability frequency for this waveguide opportunity.

[0206] MPC-driven cross-domain collaborative optimization and execution: Based on the above prediction and verification results, the cross-domain collaborative optimization solver initiates the model prediction control process, with Δt = 5 minutes and a prediction window N. p =10 (corresponds to 50 minutes), Control Window N c =3 Rolling optimization was performed. The optimization results suggest that the UAV should appropriately reduce its altitude and shift to the sea side by about 2 to 3 kilometers before approaching the rainband, so that its flight altitude falls within the waveguide altitude range of 150 to 350 meters; at the same time, the communication frequency should be switched to 28 GHz in advance and maintained during crossing the rainband and using the waveguide to reduce frequent switching.

[0207] Comparison Schemes and Performance Results: For comparison purposes, this embodiment sets up the following two comparison schemes:

[0208] Comparison Scheme A: Adjusting flight altitude and frequency only based on rainfall attenuation prediction, without predicting and utilizing atmospheric waveguide opportunities, that is, prioritizing frequencies with lower attenuation when avoiding high-attenuation rainbands, but without waveguide identification and detection verification;

[0209] Comparison Scheme B: Without using any meteorological forecast information, passive frequency switching and trajectory fine-tuning are performed solely based on real-time measured link signal-to-noise ratio (SNR): when the SNR is below a threshold SNR... th When the SNR recovers, the system attempts to switch to a backup frequency and makes minor adjustments to altitude or yaw angle according to a preset priority order. If the SNR recovers, the current frequency and track are maintained. In contrast, scheme B does not distinguish between "ordinary multipath enhancement" and "opportunistic links with obvious waveguide structures," nor does it consider environmental changes in the future.

[0210] To further highlight the technical effects of this invention, the implementation method of Comparative Scheme B in this specification is defined as follows: Comparative Scheme B does not connect to weather forecasting and physical propagation models at all, and only relies on the real-time link quality indicators (such as received signal-to-noise ratio SNR and bit error rate BER) reported by the UAV for "passive" adaptive control. Specifically, when the detected SNR is lower than a preset threshold, the system first adjusts the trajectory within a small range near the current flight altitude (e.g., a lateral offset of ±100 meters or a fine-tuning of altitude of ±50 meters). If the SNR remains lower than a lower threshold within a short period, it attempts to switch to a backup millimeter-wave frequency, with frequency priorities arranged sequentially according to the static configuration. If the SNR cannot be restored to a qualified range after multiple switches, a conservative strategy of speed reduction or bit rate reduction is triggered.

[0211] The key feature of this scheme is that it makes local, reactive decisions based solely on instantaneous or short-term average link quality. Compared to the proactive control mode of "prediction-driven + collaborative optimization + detection verification" in this invention, the comparative scheme B is prone to the following problems: First, it lacks prediction of the impending heavy rain fading area, leading to remedial measures only after fading has occurred, resulting in significant link jitter and interruptions; second, it cannot identify and lock onto opportunistic propagation structures such as atmospheric waveguides, often only occasionally obtaining brief gains after the waveguides have formed, and it is difficult to continuously utilize these gains through coordinated adjustments of trajectory and frequency; third, the lack of global constraints on frequency switching may lead to ineffective or excessively frequent switching, thereby increasing control overhead and system complexity.

[0212] In contrast, this invention introduces large-scale channel prediction based on the ITU-R model and atmospheric waveguide capture potential assessment into a cross-domain collaborative optimization model, and performs unified planning of frequency sequence and flight path under the MPC framework. This enables UAVs to proactively plan detour paths and frequency pre-switching before severe convective weather occurs, and to quickly complete detection verification and optimal frequency determination when waveguide opportunities arise, fundamentally improving link robustness and resource utilization efficiency.

[0213] Simulation results show that during periods of typical heavy rainfall and the coexistence of surface waveguides, the link interruption probability of the proposed solution is reduced by more than 90% compared to the comparative solution A and by more than 70% compared to the comparative solution B. While ensuring the interruption probability remains below a preset threshold (e.g., 10^-3), the average throughput of the proposed solution is increased by approximately 2.5 times compared to the comparative solution A and by approximately 1.6 times compared to the comparative solution B. These results verify that cross-domain collaborative optimization driven by meteorological forecasting and physical propagation models can effectively shape the propagation environment, fully exploit opportunities such as atmospheric waveguides, and simultaneously reduce ineffective frequency switching and excessive flight path detours.

[0214] Computer-readable storage media

[0215] This embodiment illustrates the implementation of the method of the present invention on a computer-readable storage medium.

[0216] In this embodiment, a computer program is stored on a computer-readable storage medium. When executed by a processor (e.g., a cloud server CPU, an edge computing node CPU, or a GPU), the program implements the steps described in Embodiments 1 to 3, including but not limited to:

[0217] (1) Obtain the predicted meteorological data of the target communication area and perform format unification and spatial-temporal alignment;

[0218] (2) The rainfall attenuation power law model based on ITU-RP.838 recommendation and the atmospheric waveguide opportunity prediction model based on the waveguide criterion of atmospheric refractive index profile are called to generate path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information.

[0219] (3) Construct and solve a collaborative optimization model with communication frequency sequence and flight trajectory sequence as joint decision variables and link robustness and total expected throughput as optimization objectives;

[0220] (4) When an atmospheric waveguide opportunity that meets the triggering conditions is detected, the frequency-sensitive detection verification process is executed to determine the preferred frequency.

[0221] (5) The generated collaborative control instruction set is sent to the UAV collaborative actuator, and the flight controller and reconfigurable multi-band radio frequency front end perform frequency switching and trajectory adjustment.

[0222] The aforementioned computer programs can be written in high-level programming languages ​​or assembly languages ​​and stored in read-only memory, flash memory, disks or other non-temporary storage media, and can be called and executed by a processor.

[0223] Performance verification example:

[0224] To verify the effectiveness of the present invention, a simulation test was conducted under a severe convective weather scenario along a coast. In this scenario, a millimeter-wave emergency communication link needs to be established between a maritime target area approximately 80 km from the ground command center. The candidate frequencies were 28 GHz and 38 GHz, the UAV flight altitude range was 300 m to 3000 m, the spatial resolution of the predicted meteorological data was 1 km, and the temporal resolution was 10 minutes.

[0225] Comparison with Option A: Adjusting the drone's flight altitude and frequency solely based on rain attenuation prediction, without utilizing atmospheric waveguide opportunities;

[0226] Comparison Scheme B: Frequency switching and trajectory fine-tuning are performed based on real-time link quality measurements, without using meteorological forecast information;

[0227] The present invention adopts the integrated method of "rainfall attenuation prediction + atmospheric waveguide opportunity prediction + synergistic optimization + detection verification" as described in Examples 1 to 3.

[0228] Simulation results show that during typical periods of heavy rainfall, the link interruption probability of the proposed solution is reduced by more than 90% compared to the comparative solution A and by more than 70% compared to the comparative solution B. Under the premise of ensuring that the interruption probability is kept below a predetermined threshold, the average throughput of the proposed solution is increased by about 2 to 3 times compared to the comparative solution A and by about 1.5 times compared to the comparative solution B. Figure 5 The diagram shows a comparison curve of the interruption probability and average throughput over time, which can intuitively reflect the significant performance advantages of the present invention under complex weather conditions.

[0229] As can be seen from the above embodiments, the present invention achieves high robustness and high throughput of millimeter-wave ultra-long-distance emergency communication links by organically combining meteorological forecasting with physical propagation models, active detection verification mechanisms, and UAV trajectory / frequency point collaborative optimization. The technical effect is significantly better than traditional single-dimensional optimization schemes.

[0230] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that various modifications or equivalent substitutions can be made to the technical solutions of the above embodiments without departing from the spirit and substance of the present invention, and such modifications or substitutions should all fall within the protection scope defined by the claims of the present invention.

Claims

1. A millimeter-wave ultra-long-range emergency communication method for unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: S1. Obtain forecast meteorological data for the target communication area; S2. Based on the predicted meteorological data, couple and execute the rainfall attenuation prediction model and the atmospheric waveguide opportunity prediction model to output the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information of at least two candidate millimeter wave frequency points in the future period, respectively. S3. Construct and solve a collaborative optimization model. The model takes the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information as inputs, the communication frequency sequence and flight trajectory sequence of the UAV as joint decision variables, and the performance index characterizing the robustness of the communication link in the prediction period as the optimization objective. The robustness is balanced between the link interruption probability and the lower bound of the throughput through a dynamic weight adjustment mechanism, and outputs a forward-looking collaborative control command set for frequency switching and trajectory adjustment. S4. According to the cooperative control instruction set, control the UAV to perform corresponding frequency switching and trajectory adjustment.

2. The method according to claim 1, characterized in that, In step S3, the robustness is measured by subtracting the overhead penalty term caused by frequency switching and the energy consumption penalty term caused by the deviation of the trajectory from the reference path from the total expected throughput within the prediction period. The total expected throughput is calculated as follows: the prediction period is discretized into multiple time segments. For each time segment, the instantaneous expected throughput of the time segment is calculated based on the predicted UAV position, the selected frequency, and the channel gain mapped from the path attenuation spatiotemporal distribution information and the atmospheric waveguide capture potential information corresponding to the time segment. Then, the instantaneous expected throughput of all time segments is accumulated.

3. The method according to claim 1 or 2, characterized in that, In step S2, after the atmospheric waveguide opportunity prediction model outputs that there is capture potential, a frequency-sensitive detection verification process is further executed, specifically including: S21. Determine whether the atmospheric waveguide capture potential information exceeds a preset confidence threshold, and whether the predicted effective duration of the waveguide is greater than the sum of the detection and switching overhead times. S22. If satisfied, control the UAV to transmit low-power detection signals on at least two candidate millimeter-wave frequency points; S23. Receive the received intensity of the detection signals at each frequency point fed back from the remote end, and determine a preferred frequency point by combining the stability parameters in the atmospheric waveguide capture potential information. In step S3, when constructing the collaborative optimization model, the preferred frequency point is used as a high-priority candidate for the communication frequency point sequence.

4. The method according to claim 3, characterized in that, In step S23, the rule for determining the preferred frequency point is: select the frequency point with the highest received strength and the corresponding stability parameter is higher than the stability threshold; if the stability parameter of all frequency points is lower than the stability threshold, then the waveguide opportunity is determined to be unreliable and waveguide-based optimization decision is not triggered.

5. The method according to claim 1, characterized in that, In step S2, the path attenuation spatiotemporal distribution information is represented in the form of three-dimensional spatiotemporal grid data, wherein each grid point contains location, time and predicted attenuation value corresponding to each candidate millimeter wave frequency point; the predicted attenuation value is calculated by substituting the rainfall intensity field in the predicted meteorological data into the rainfall attenuation power law model.

6. The method according to claim 1, characterized in that, In step S3, the collaborative optimization model is solved using a model predictive control framework. At each decision time, a finite-time optimization problem is solved based on the latest prediction information, and only the first control instruction in the collaborative control instruction set is implemented. Prediction and solution are performed again at the next decision time.

7. The method according to any one of claims 1 to 6, characterized in that, The drones include multiple drones, and the decision variables of the collaborative optimization model are expanded to the communication frequency sequence and flight trajectory sequence of multiple drones. The total expected throughput of the end-to-end link is used as one of the optimization objectives. At the same time, anti-collision constraints and constraints for limiting the reuse of the same frequency in mutually interfering links are introduced. Preferably, the same frequency reuse constraints are obtained based on the collision graph model.

8. A UAV millimeter-wave communication collaborative anti-aging and opportunity-taking system for implementing the method of claim 7, characterized in that, include: An intelligent decision-making center, deployed on the ground or in the cloud, is used to perform data acquisition, channel prediction, and collaborative optimization decision-making in the method, and includes: The meteorological fusion and data preprocessing module is configured to acquire the predicted meteorological data of the target communication area, and to perform format unification, temporal and spatial alignment and noise suppression processing on the predicted meteorological data, and output the processed predicted meteorological data. The dual-channel coupled prediction module is configured to, based on the processed predicted meteorological data, respectively call the rainfall attenuation power-law model constructed based on the ITU-RP.838 recommendation and the atmospheric waveguide opportunity prediction model constructed based on the waveguide criterion of the atmospheric refractive index profile, and output the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information of at least two candidate millimeter wave frequency points in the future period. The cross-domain collaborative optimization solver is configured to construct and solve a collaborative optimization model based on the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information. It uses the communication frequency sequence and flight trajectory sequence of the UAV as joint decision variables, takes the robustness of the communication link in the prediction period as the optimization objective, and outputs a forward-looking collaborative control command set for frequency switching and trajectory adjustment. And collaborative actuators deployed on drone platforms, including: A reconfigurable multi-band radio frequency front end is configured to switch between the candidate millimeter-wave frequencies according to the cooperative control instruction set; The flight controller with integrated communication sensing function is configured to receive and parse the cooperative control command set, and synchronously control the reconfigurable multi-band radio frequency front end and the control surface actuators of the UAV to perform frequency switching and trajectory adjustment.

9. The system according to claim 8, characterized in that, When solving the collaborative optimization model, the cross-domain collaborative optimization solver calls a pre-stored channel gain mapping table. This mapping table maps the abstract parameters in the path attenuation spatiotemporal distribution information and atmospheric waveguide capture potential information into equivalent channel gain values ​​that depend on location, time, and frequency, and is used to calculate the instantaneous expected throughput for each time segment.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it is used to implement the steps of the method as described in any one of claims 1 to 7.

Images