Unmanned aerial vehicle near field communication matching processing method and system in public network communication area

Abstract

The application provides a public network-free communication area unmanned aerial vehicle near field communication matching processing method and system, relates to the technical field of intelligent control, and the method comprises the following steps: generating a session key and a device authority topology graph according to a quantum random number; based on the device authority topology graph and a channel state tensor, the communication capability of the devices in the unmanned aerial vehicle cluster is matched, a matched device pair set and a communication parameter mapping relationship thereof are generated; according to the session key, the device authority topology graph and the matched device pair set, reconnaissance data, control instructions and state information are jointly encoded to generate an encoded symbol stream; and according to the communication parameter mapping relationship of the matched device pair set, the encoded symbol stream is directionally transmitted and scheduled, and the near field communication matching processing between unmanned aerial vehicles is completed. The application improves the cooperative operation efficiency and task success rate of the unmanned aerial vehicle rescue cluster in the complex environment of a public network-free mountain area.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control, in particular to a method and system for matching near-field communication of unmanned aerial vehicles in a public network communication area. BACKGROUND

[0002] The rescue command center dispatches a mixed cluster composed of multi-rotor unmanned aerial vehicles, including 1 command relay vehicle, 3 three-dimensional modeling reconnaissance vehicles, and 2 material delivery vehicles. The cluster needs to quickly investigate a 5 square kilometer earthquake epicenter area with broken terrain and multiple landslides and landslides. The task requires reconnaissance vehicles to transmit high-resolution orthographic images and thermal imaging data in real time to identify disaster buildings and life heat sources, and delivery vehicles need to accurately deliver first aid kits according to instructions, while all unmanned aerial vehicles need to share location and state information. There is no public network signal in the entire area, and the electromagnetic wave propagation environment is extremely complex due to mountains, valleys, cliffs, and dense forests.

[0003] The traditional communication scheme uses a simplified free space model and cannot cope with the "non-line-of-sight", "multipath deep fading", and "terrain shielding" effects unique to mountainous areas. For example, when the reconnaissance vehicle flies into the back slope or valley, the signal may drop sharply or even be interrupted due to mountain blocking, resulting in the loss of critical disaster image data. The existing technology lacks real-time and active sensing and modeling capabilities for the three-dimensional space channel characteristics of the rescue site, and cannot predict and avoid communication blind areas.

[0004] The emergency task is phased (such as initial wide-area reconnaissance, mid-term key search and rescue, and later material delivery), and the business traffic and priority of each unmanned aerial vehicle changes dramatically at different stages. The traditional static or semi-static frequency spectrum allocation method cannot dynamically and globally optimize the adjustment according to real-time task requirements and channel changes. For example, when a suspected life sign is found and all reconnaissance vehicles need to focus on investigation, the existing scheme cannot quickly and optimally reorganize the channel and time slot to ensure high-speed and low-latency transmission of multi-vehicle cooperative data flow in the area. SUMMARY

[0005] The present application provides a method and system for matching near-field communication of unmanned aerial vehicles in a public network communication area, improving the cooperative operation efficiency and task success rate of unmanned aerial vehicle rescue clusters in complex mountainous environments without public networks.

[0006] To solve the above technical problems, the technical solutions of the present application are as follows:

[0007] In a first aspect, a method for matching near-field communication of unmanned aerial vehicles in a public network communication area is provided, the method comprising:

[0008] Regulating the near-field electromagnetic wave propagation environment to obtain three-dimensional channel characteristic data; collecting three-dimensional channel characteristic data to generate a channel state tensor;

[0009] construct a quantum annealing optimization model based on the channel state tensor; and solve a spectrum resource allocation problem by simulating a quantum tunneling effect according to the quantum annealing optimization model, to obtain a carrier frequency, a time slot offset, and a power control parameter set of each unmanned aerial vehicle;

[0010] construct an overlapping communication frame structure according to the carrier frequency, the time slot offset, and the power control parameter set; and generate a dynamic frame scheduling scheme according to the overlapping communication frame structure;

[0011] extract a channel reciprocity feature sequence based on the dynamic frame scheduling scheme; generate a quantum random number according to the channel reciprocity feature sequence; generate a session key and a device permission topology graph according to the quantum random number; and perform communication capability matching on devices in the unmanned aerial vehicle cluster based on the device permission topology graph and the channel state tensor, to generate a matched device pair set and a communication parameter mapping relationship thereof;

[0012] jointly encode reconnaissance data, control instructions, and state information according to the session key, the device permission topology graph, and the matched device pair set, to generate an encoded symbol stream;

[0013] perform directional transmission scheduling on the encoded symbol stream according to the communication parameter mapping relationship of the matched device pair set, to complete near-field communication matching processing between unmanned aerial vehicles.

[0014] In a second aspect, a near-field communication matching processing system for unmanned aerial vehicles in a non-public network communication area includes:

[0015] a regulation and control module configured to regulate and control a near-field electromagnetic wave propagation environment to obtain three-dimensional channel feature data, and collect the three-dimensional channel feature data to generate a channel state tensor;

[0016] an optimization module configured to construct a quantum annealing optimization model based on the channel state tensor, and solve a spectrum resource allocation problem by simulating a quantum tunneling effect according to the quantum annealing optimization model, to obtain a carrier frequency, a time slot offset, and a power control parameter set of each unmanned aerial vehicle;

[0017] a construction module configured to construct an overlapping communication frame structure according to the carrier frequency, the time slot offset, and the power control parameter set, and generate a dynamic frame scheduling scheme according to the overlapping communication frame structure;

[0018] a matching module configured to extract a channel reciprocity feature sequence based on the dynamic frame scheduling scheme, generate a quantum random number according to the channel reciprocity feature sequence, generate a session key and a device permission topology graph according to the quantum random number, and perform communication capability matching on devices in the unmanned aerial vehicle cluster based on the device permission topology graph and the channel state tensor, to generate a matched device pair set and a communication parameter mapping relationship thereof;

[0019] The processing module is used for jointly encoding the reconnaissance data, the control instruction and the state information according to the session key, the device permission topology graph and the matched device pair set, generating an encoded symbol stream; and performing directional transmission scheduling on the encoded symbol stream according to the communication parameter mapping relationship of the matched device pair set, completing the inter-unmanned aerial vehicle near field communication matching processing.

[0020] The above scheme of the present application at least includes the following beneficial effects:

[0021] Through active perception and intelligent regulation of the environment, quantum heuristic global resource optimization, dynamic frame scheduling, physical layer endogenous security generation, task-aware communication matching and cross-layer joint encoding transmission, the core problems of unreliable communication, low efficiency, weak security and poor self-adaptability of unmanned aerial vehicle clusters in the public network communication area in a complex near field environment are systematically solved, and the cooperative operation efficiency and task reliability of the cluster in emergency rescue and other scenarios are improved. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 is a flowchart of the unmanned aerial vehicle near field communication matching processing method in the public network communication area provided by the embodiment of the present application;

[0023] Figure 2 is a schematic diagram of the unmanned aerial vehicle near field communication matching processing system in the public network communication area provided by the embodiment of the present application. DETAILED DESCRIPTION

[0024] Exemplary embodiments of the present disclosure will be described in greater detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be accurately conveyed to those skilled in the art.

[0025] As Figure 1 shown, the embodiment of the present application proposes a near field communication matching processing method for unmanned aerial vehicles in a public network communication area, which comprises the following steps:

[0026] Step 1: Regulate the near field electromagnetic wave propagation environment to obtain three-dimensional channel characteristic data; collect the three-dimensional channel characteristic data to generate a channel state tensor;

[0027] Step 2: Based on the channel state tensor, a quantum annealing optimization model is constructed; according to the quantum annealing optimization model, the frequency spectrum resource allocation problem is solved by simulating the quantum tunneling effect to obtain the carrier frequency, time slot offset and power control parameter set of each unmanned aerial vehicle;

[0028] Step 3, constructing an overlapping communication frame structure according to the carrier frequency, time slot offset and power control parameter set, and generating a dynamic frame scheduling scheme according to the overlapping communication frame structure;

[0029] Step 4, extracting a channel reciprocity feature sequence based on the dynamic frame scheduling scheme, generating a quantum random number according to the channel reciprocity feature sequence, generating a session key and a device authority topology graph according to the quantum random number, and matching the communication capabilities of the devices in the unmanned aerial vehicle cluster based on the device authority topology graph and the channel state tensor to generate a matched device pair set and a communication parameter mapping relationship thereof;

[0030] Step 5, jointly encoding the reconnaissance data, control instructions and state information according to the session key, the device authority topology graph and the matched device pair set to generate an encoded symbol stream;

[0031] Step 6, performing directional transmission scheduling on the encoded symbol stream according to the communication parameter mapping relationship of the matched device pair set to complete the near-field communication matching processing between unmanned aerial vehicles.

[0032] In the embodiments of the present application, by actively regulating the near-field electromagnetic environment and constructing a three-dimensional channel feature tensor, the system can accurately perceive and adapt to complex communication environments; combined with a quantum annealing optimization model for resource allocation, the global or approximately global optimal solution can be found, thereby establishing stable and reliable communication links for the unmanned aerial vehicle cluster in the near-field region with strong interference and significant multipath effect, greatly improving the robustness of communication. The design of dynamic frame scheduling and overlapping communication frame structure makes the communication signal have randomness and dynamics in the time-frequency domain, which is difficult to be continuously tracked and interfered by external devices; combined with quantum random numbers generated based on channel reciprocity for encryption, physical layer security is realized, which significantly enhances the anti-interference and low probability interception / detection capability of communication; the quantum annealing optimization model can efficiently solve the complex spectrum, time slot and power joint allocation problem, avoiding the problem that traditional algorithms are prone to local optimization, which maximizes the utilization of limited spectrum resources, and the overlapping communication frame structure further improves the spectrum efficiency, supporting efficient concurrent near-field communication in dense unmanned aerial vehicle clusters. Through communication capability matching based on device authority topology graph and channel state, appropriate communication parameters can be allocated to each pair of communication devices (matching device pair set), which effectively reduces the transmission power of the unmanned aerial vehicle and prolongs the overall endurance time of the cluster while ensuring communication quality. Quantum random numbers are generated based on channel reciprocity, and session keys are generated based on the quantum random numbers, realizing one-time pad or high security level key distribution. This process is completely based on the characteristics of the wireless channel itself and does not rely on pre-shared keys or complex public key infrastructure, building a lightweight, endogenous security system that effectively prevents eavesdropping and man-in-the-middle attacks. The entire processing flow (perception, optimization, scheduling, matching, encryption, transmission) is highly automated and does not rely on a central node or public network infrastructure. The generated device authority topology graph can dynamically reflect the networking relationship and capability of the cluster, and even if part of the nodes fail or dynamically join, the system can quickly re-match and schedule, with strong self-organizing and anti-destroying survival ability.

[0033] In a preferred embodiment of the present application, step 1, the near-field electromagnetic wave propagation environment is regulated to obtain three-dimensional channel feature data; three-dimensional channel feature data is collected to generate a channel state tensor, including:

[0034] Step 1.1, by deploying a reconfigurable intelligent surface array to dynamically regulate the near-field electromagnetic wave propagation path in the public network-free communication area, a multi-dimensional spatial beamforming is formed; based on the multi-dimensional spatial beamforming, the beam pointing parameters of the unmanned aerial vehicle onboard phased array antenna are configured, specifically including:

[0035] According to the topographic survey data (including collapse points, landslide bodies, and valley distribution) of a 5-square-kilometer area in the earthquake zone, within the coverage range of the command relay machine, 3-5 key commanding points (such as highlands that have not completely collapsed, and stable cliff platforms) are selected to deploy RIS arrays. Each RIS array is composed of no less than 100 electromagnetic regulation units, which support continuous adjustment of the amplitude and phase of incident electromagnetic waves. After deployment, the command relay machine sends initialization instructions to all RIS arrays, completes time synchronization (based on Beidou timing), and configures basic parameters (the working frequency band is matched with the unmanned aerial vehicle communication frequency band, and the initial reflection mode is set to omnidirectional scattering).

[0036] The command relay machine receives the initial position information of each unmanned aerial vehicle in real time (obtained through the on-board GPS / IMU), and identifies the line-of-sight blocking areas (such as back slopes and valleys) in the current communication link based on the three-dimensional terrain model. For these areas, the command relay machine sends regulation instructions to the RIS arrays in the corresponding areas, adjusts the reflection phase of each unit in the array, and constructs an indirect communication path between the unmanned aerial vehicle, the RIS, and the command relay machine. At the same time, the RIS arrays feed back their real-time reflection state to the command relay machine, which calculates a multi-dimensional spatial beamforming scheme based on the principle of multi-path superposition. This scheme includes directional beam parameters for the command relay machine, 3 reconnaissance aircraft, and 2 delivery aircraft, ensuring that each unmanned aerial vehicle can obtain independent beam coverage.

[0037] The command relay machine sends the generated multi-dimensional spatial beamforming parameters (including beam pointing angle, beam width, and gain configuration) to the on-board phased array antenna control system of each unmanned aerial vehicle. The antenna system adjusts the phase difference between the array elements according to the received parameters, accurately points the antenna beam to the corresponding RIS array or command relay machine (when line-of-sight is available). For example, for reconnaissance aircraft flying into a valley, the antenna beam will be configured to point to the RIS array at the entrance of the valley, ensuring non-line-of-sight communication through RIS reflection; while the delivery aircraft in the open area directly points the beam to the command relay machine, reducing transmission loss. After the configuration is completed, the unmanned aerial vehicle feeds back the antenna state to the command relay machine, forming a closed-loop confirmation.

[0038] Step 1.2, according to the beam pointing parameters, the unmanned aerial vehicle performs multi-point scanning measurement in three-dimensional space, obtains channel impulse response data under different spatial positions, different time slices, and different frequency dimensions; and performs time-frequency analysis and processing on the channel impulse response data to extract three-dimensional channel feature data, specifically including:

[0039] The command relay machine combines the task requirements of the earthquake area (focus on covering the collapse area and the concentrated area of damaged buildings) with the terrain data to plan differentiated three-dimensional scanning paths for the three three-dimensional modeling reconnaissance machines. Each path divides the earthquake area into 50m x 50m grid cells, and sets three measurement points of different altitudes (50m, 80m, and 120m, respectively, to adapt to different building heights and terrain undulations) in each grid cell to ensure coverage of the entire 5 square kilometer area. At the same time, each measurement point is assigned a time slice (100ms for each point, with a 50ms time interval between adjacent points) and frequency resources to avoid signal interference when multiple machines are measuring.

[0040] Each reconnaissance machine flies according to the planned path, and when it reaches each measurement point, it fixes the antenna attitude according to the beam pointing parameters configured in step 1.1 and starts data collection. During the collection process, the command relay machine and the RIS array send known pilot signals (using Zadoff-Chu sequences, which have good autocorrelation) to the unmanned aerial vehicle, which synchronously receives the pilot signals and the reflected signals, records the channel impulse response (CIR) data at different spatial positions (marked by the onboard GPS / IMU in real time, with an accuracy of 0.1m), different time slices (synchronized by the Beidou timing timestamp), and different frequency dimensions (collected in order according to the pre-set sub-channels). At the same time, the unmanned aerial vehicle transmits the original CIR data to the command relay machine in real time to ensure that the data is not lost.

[0041] The command relay machine pre-processes the received original CIR data, first eliminating environmental noise through band-pass filtering (filtering frequency band matching pilot signal frequency band), and then removing the direct wave interference in the multipath signal. Subsequently, time-frequency analysis is performed: in the time dimension, the delay spread characteristics of the CIR data are extracted through a sliding window (window length 20ms) to reflect the time distribution of the multipath signal; in the frequency dimension, the CIR data of each time slice is subjected to Fourier transform to obtain the frequency domain response, and the Doppler frequency offset (reflecting the influence of the unmanned aerial vehicle movement on the signal) and the frequency selective fading characteristics are extracted; in the spatial dimension, the signal amplitude variation within the beam coverage range is analyzed in combination with the position information of different measurement points to extract the spatial attenuation characteristics. Finally, three-dimensional channel characteristic data containing delay spread, Doppler frequency offset, frequency fading coefficient, and spatial attenuation coefficient are extracted from the time, frequency, and space dimensions.

[0042] In step 1.3, based on the three-dimensional channel characteristic data, multi-dimensional information including spatial position information, timestamp information, and frequency domain response information is collected by distributed sensing nodes; and the multi-dimensional information is time and space aligned and calibrated to obtain calibrated multi-dimensional information, specifically including:

[0043] In the command relay machine, each unmanned aerial vehicle and the deployed RIS array, respectively integrate distributed sensing nodes to form a multi-source data acquisition network. Each sensing node synchronously acquires three types of core information: first, spatial position information, the positions of the command relay machine and the unmanned aerial vehicles are located by GPS / IMU combination positioning (update frequency 10 Hz), and the position of the RIS array is a fixed coordinate (calibrated when deployed), while recording the attitude angles (pitch, roll, heading) of each device; second, timestamp information, all sensing nodes are synchronized based on the Beidou timing system, ensuring that the timestamp accuracy of the acquired data reaches 1 ms; third, frequency domain response information, in addition to the frequency fading coefficients extracted in step 1.2, supplementary acquisition of auxiliary parameters such as signal-to-noise ratio (SNR) and received signal strength indicator (RSSI) of each frequency subchannel. The acquired multi-dimensional information is associated with the three-dimensional channel characteristic data of step 1.2 to form a data set supplemented by channel characteristics-position-time-frequency.

[0044] Temporal and spatial deviation detection and calibration: First, time deviation calibration, the command relay machine extracts the timestamps of all data, and for data with time offset (such as some unmanned aerial vehicles with delayed timestamps due to communication delay), linear interpolation method is used to correct the timestamps to ensure that all data of the same measurement event are time-synchronized. Subsequently, spatial deviation calibration is performed, combining the three-dimensional terrain model and the attitude angle data of each device to correct the GPS positioning error of the unmanned aerial vehicle. For example, when the unmanned aerial vehicle is in a valley, the GPS signal may be affected by multipath and produce deviation, at this time, through the matching of attitude angle and terrain data, the positioning coordinates are adjusted to the actual flight position. For the reflection signal path of the RIS array, according to its fixed coordinates and the position of the unmanned aerial vehicle, the signal propagation distance is recalculated to correct the path loss deviation in the spatial attenuation coefficient.

[0045] Conduct consistency test on the calibrated multi-dimensional information, by comparing the data collected by different devices in the same time and same area (such as SNR of the same channel collected by the command relay machine and the reconnaissance machine), eliminate abnormal data with deviation exceeding the threshold (such as SNR deviation greater than 3 dB). At the same time, combined with the phased requirements of the task in the earthquake area, the key data matching the current task are retained, for example, in the initial wide-area reconnaissance stage, the spatial attenuation and frequency response data of the wide coverage area are mainly retained, and in the middle stage of search and rescue, the channel characteristic data of the life heat source area are strengthened to ensure the accuracy of the calibrated data set.

[0046] Step 1.4, the calibrated multi-dimensional information is tensorized according to the spatial dimension, time dimension and frequency dimension to construct a three-dimensional channel characteristic matrix; the three-dimensional channel characteristic matrix is normalized and dimensionally compressed to generate a channel state tensor representing the channel spatial correlation, which includes:

[0047] determine the three-dimensional structure dimensions of the tensor, that is, the spatial dimension, the time dimension and the frequency dimension, wherein: the spatial dimension is indexed by the spatial position combination of the UAV-RIS-command relay, including the UAV position coordinates, the RIS array number, the spatial attenuation coefficient and the like; the time dimension is indexed by the calibrated time stamp, including the delay spread, the Doppler frequency offset and the like which change with time; the frequency dimension is indexed by the frequency subchannel number, including the frequency fading coefficient, the SNR, the RSSI and the like. Fill the calibrated multi-dimensional information according to the above three-dimensional structure to construct an initial three-dimensional channel feature matrix, and each element in the matrix corresponds to a complete channel feature set at a specific spatial position-specific time-specific frequency. On this basis, the normalized three-dimensional matrix is subjected to tensor decomposition operation, and the PARAFAC (Parallel Factor) decomposition algorithm is adopted, which can uniquely decompose the three-dimensional tensor (denoted as X ∈ R (I×J×K) , J is the number of time slices, and K is the number of frequency subchannels) into the product form of three independent low-dimensional factor matrices and a core tensor, and the specific implementation is as follows:

[0048] First, initialize three factor matrices A ∈ R (I×F) , B ∈ R (J×F) , C ∈ R (K×F) , wherein F is the number of common factors, that is, the target dimension after compression, which corresponds to the feature projection matrix of the spatial, time and frequency dimensions respectively.

[0049] Second, three factor matrices are iteratively optimized by the alternating least squares method (ALS), each time fixing two matrices, solving the third matrix based on minimizing the error (using mean square error MSE as the objective function) between the original tensor and the reconstructed tensor, and avoiding overfitting through a regularization term in the iteration process to ensure that the decomposition result can reflect the real channel characteristics.

[0050] Third, the optimal number of common factors F is determined through cross-validation, and the data set is divided into a training set (70%) and a test set (30%) during verification. The minimum reconstruction error of the test set is used as the criterion to determine the value of F, and the value of F is usually 1 / 5-1 / 3 of the original dimension (for example, the spatial dimension is compressed from 500 position combinations to 100 common factors). After decomposition, the three low-dimensional factor matrices A, B and C extract the channel correlation characteristics of the spatial dimension, the dynamic change characteristics of the time dimension and the frequency domain response characteristics of the frequency dimension, and discard the redundant information in the original data.

[0051] Finally, the three low-dimensional factor matrices are multiplied with the optimized core tensor to reorganize a three-dimensional channel state tensor, which not only retains the three-dimensional correlation of space-time-frequency, but also improves the data processing efficiency through dimension compression. The generated channel state tensor can be transmitted to the communication scheduling system of the unmanned aerial vehicle cluster in real time. Based on the spatial correlation characteristics reflected by the tensor, the scheduling system can quickly identify the channel quality of different areas, allocate optimal spectrum resources for multi-machine cooperative communication in the key search and rescue area, reduce the data transmission delay, and ensure the stable transmission of disaster image and life detection data.

[0052] To eliminate the dimensional differences of different dimensional parameters (such as the unit of spatial attenuation coefficient is dB, and the unit of Doppler frequency offset is Hz), the initial three-dimensional channel feature matrix is normalized. The Z-Score standardization method is used to calculate the mean and standard deviation of the parameters in each dimension, and the original parameters are converted into standardized data with a mean of 0 and a standard deviation of 1. Considering that the initial matrix has a high dimension (the spatial dimension contains hundreds of position combinations, the time dimension contains thousands of time slices, and the frequency dimension contains hundreds of sub-channels), dimension compression is needed to improve the efficiency of subsequent processing. The tensor decomposition algorithm (such as PARAFAC decomposition) is used to decompose the normalized three-dimensional matrix, and low-dimensional factor matrices reflecting the core features of the channel are extracted. During the decomposition process, the spatial correlation of the channel is retained as the target, and the optimal compression dimension is determined through cross-validation (usually compressed to 1 / 5-1 / 3 of the original dimension), ensuring that the compressed data can still accurately represent the channel correlation characteristics between different spatial positions. Finally, the compressed low-dimensional factor matrices are recombined to generate a channel state tensor representing the spatial correlation of the channel, which can be fed back to the communication scheduling system of the unmanned aerial vehicle cluster to provide data support for dynamic spectrum allocation and communication link optimization.

[0053] In a preferred embodiment of the present application, step 2, based on the channel state tensor, a quantum annealing optimization model is constructed; according to the quantum annealing optimization model, the frequency spectrum resource allocation problem is solved by simulating the quantum tunneling effect to obtain the carrier frequency, time slot offset and power control parameter set of each unmanned aerial vehicle, including:

[0054] Step 2.1, based on the spatial correlation and time-frequency characteristics in the channel state tensor, a target function for spectrum resource allocation is constructed, including: step 2.11, separating the spatial dimension component, time dimension component and frequency dimension component from the channel state tensor; based on the spatial dimension component, calculating the spatial isolation matrix between unmanned aerial vehicle nodes, based on the time dimension component, extracting the channel coherence time sequence, and based on the frequency dimension component, obtaining the frequency domain flatness index, specifically including:

[0055] Based on the channel state tensor generated in step 1.4 (characterizing the channel space-time-frequency three-dimensional correlation characteristics), three independent dimension components are separated by the factor matrix after tensor decomposition, and the core parameters constructed by the support target function are calculated respectively:

[0056] Dimension component separation, that is, extracting the spatial dimension factor matrix A, the time dimension factor matrix B, and the frequency dimension factor matrix C obtained by step 1.4 decomposition from the channel state tensor. The spatial dimension factor matrix A contains the correlation characteristics of all unmanned aerial vehicle-RIS-command relay machine spatial position combinations, the time dimension factor matrix B contains the channel dynamic change characteristics of different time slices, and the frequency dimension factor matrix C contains the response characteristics of each frequency subchannel. Through matrix transposition and feature alignment processing, the indexes of the three components are ensured to correspond one by one with the unmanned aerial vehicle number, timestamp, and subchannel number. Spatial isolation degree matrix calculation, that is, based on the unmanned aerial vehicle position coordinates in the spatial dimension factor matrix A, the RIS array deployment position, and the spatial attenuation coefficient, the spatial isolation degree between any two unmanned aerial vehicles (including command relay machines) is calculated by using the logarithmic distance path loss model combined with the beam orthogonality algorithm. Specifically, for each pair of unmanned aerial vehicles (i, j), first calculate the total attenuation value of signal propagation according to the three-dimensional coordinates of the two unmanned aerial vehicles and the reflection path of the intermediate RIS array, then calculate the orthogonality coefficient between the beams (the higher the orthogonality, the smaller the mutual interference) combined with the beam pointing angles of the two unmanned aerial vehicle onboard phased array antennas, and finally weight and sum the total attenuation value and the orthogonality coefficient with a weight of 0.6:0.4 to obtain the spatial isolation degree value of the unmanned aerial vehicle pair (i, j). Finally, a 6x6-dimensional spatial isolation degree matrix is constructed (6 is the total number of unmanned aerial vehicle clusters, including 1 command relay machine, 3 reconnaissance machines, and 2 delivery machines). Channel coherence time sequence extraction, that is, extracting the Doppler frequency offset and delay spread characteristic parameters of each unmanned aerial vehicle from the time dimension factor matrix B, calculating the channel coherence time (i.e., the maximum time interval for which the channel remains stable) of each unmanned aerial vehicle at each time slice; arrange the coherence times of all time slices in chronological order to form the channel coherence time sequence of each unmanned aerial vehicle, and eliminate abnormal fluctuations through moving average filtering (window length of 5 time slices) to ensure the smoothness of the sequence.

[0057] Frequency domain flatness index acquisition, that is, for the frequency fading coefficients of each frequency subchannel in the frequency dimension factor matrix C, calculate the variance of the fading coefficients of each subchannel and the adjacent 5 subchannels (the smaller the variance, the flatter the frequency domain); normalize the variance value (the normalization range is 0-1, the smaller the variance, the closer the flatness index to 1) to obtain the frequency domain flatness index of each frequency subchannel; arrange the flatness indexes of all subchannels in frequency order to form the frequency domain flatness sequence.

[0058] Step 2.12. Calculate normalized spatial orthogonality coefficients between each pair of UAVs according to the spatial isolation matrix; determine the normalized time window length according to the channel coherence time sequence; and divide the normalized subcarrier aggregation area according to the frequency domain flatness index, specifically including:

[0059] Based on the spatial isolation matrix, channel coherence time sequence, and frequency domain flatness index obtained in step 2.11, further calculate the normalized parameters and divide the aggregation area to provide the basis for index mapping:

[0060] The normalized spatial orthogonality coefficient calculation is to determine the maximum and minimum values of all elements in the spatial isolation matrix (i.e., the spatial isolation values between pairs of UAVs). A linear normalization method is used to map each isolation value to the [0, 1] interval, obtaining the normalized spatial orthogonality coefficient. The closer the coefficient is to 1, the better the spatial isolation effect between the two UAVs, and the lower the possibility of mutual signal interference. The normalized time window length determination is to extract the maximum value of the mean of each UAV channel coherence time sequence (denoted as Tmax), and normalize the coherence time mean of each UAV based on Tmax (the normalized range is [0.5, 1], ensuring that the time window length is not less than the minimum task scheduling unit). Combined with the task period of the UAV cluster (e.g., 10s for the wide-area reconnaissance stage and 5s for the key search and rescue stage), the normalized coherence time mean is multiplied by the task period to obtain the normalized time window length of each UAV (i.e., the minimum unit length of time domain scheduling). The normalized subcarrier aggregation area division is to traverse the frequency domain flatness sequence and select subchannels based on whether the flatness index is greater than a preset threshold (e.g., 0.8). Subchannels that continuously meet the threshold condition are divided into an aggregation unit. The number of subchannels in each aggregation unit is counted and normalized according to the proportion of the number of subchannels in the aggregation unit to the total number of subchannels, obtaining the normalized subcarrier aggregation area (each area corresponds to a group of aggregable subchannels, and the normalized value reflects the amount of subchannel resources in the area).

[0061] Step 2.13. Map the normalized spatial orthogonality coefficient to the spatial reuse gain index, the normalized time window length to the time domain scheduling efficiency index, and the normalized subcarrier aggregation area to the frequency spectrum utilization rate index, specifically including:

[0062] Map the three normalized parameters obtained in step 2.12 to the core optimization indexes of the objective function, ensuring that the indexes are directly related to the core needs of spectrum resource allocation:

[0063] The spatial reuse gain index mapping is to establish a positive correlation mapping relationship between the normalized spatial orthogonality coefficient and the spatial reuse gain. When the coefficient is 1, the spatial reuse gain is set to the maximum value (such as 1.5). When the coefficient is 0, the spatial reuse gain is set to the minimum value (such as 0.5). The intermediate value is calculated by linear interpolation. Finally, the spatial reuse gain index of each unmanned aerial vehicle after being paired with other unmanned aerial vehicles is obtained, which reflects the spatial reuse efficiency when multiple unmanned aerial vehicles use the frequency spectrum resource at the same time. The time domain scheduling efficiency index mapping is to map the normalized time window length to the time domain scheduling efficiency index. The closer the window length is to the task period, the closer the scheduling efficiency index is to 1 (indicating less time domain resource waste). The smaller the window length, the lower the scheduling efficiency index (more switching times are required, reducing efficiency). In the mapping process, the task priority (the reconnaissance machine has a higher priority than the delivery machine, and the scheduling efficiency index weight coefficient of the reconnaissance machine is additionally increased by 20%) is combined to ensure the time domain resource guarantee of the key task. The spectrum utilization rate index mapping is to map the normalized value of the normalized subcarrier aggregation region to the spectrum utilization rate index. The more the number of subchannels in the aggregation region and the higher the flatness, the closer the spectrum utilization rate index is to 1 (after aggregation, the guard bandwidth occupation can be reduced, and the transmission rate per unit spectrum can be improved). The spectrum utilization rate index of each aggregation region is calculated to provide a basis for subsequent spectrum allocation.

[0064] Step 2.14, multiply the spatial reuse gain index, the time domain scheduling efficiency index and the spectrum utilization rate index by the corresponding preset weight coefficient, and then sum them up to construct a basic objective function. In the basic objective function, set the interference suppression constraint term and the power consumption constraint term to form the objective function of the spectrum resource allocation, which specifically includes:

[0065] Based on the three core indexes mapped in step 2.13, the complete spectrum resource allocation objective function is constructed in combination with the constraint term:

[0066] The basic objective function is constructed, that is, the weight coefficients are preset according to the task demand of the unmanned aerial vehicle cluster. The weight of the spatial reuse gain index is set to 0.4 (adapted to the multi-machine cooperative communication demand), the weight of the time domain scheduling efficiency index is set to 0.3 (adapted to the task phase change), and the weight of the spectrum utilization rate index is set to 0.3 (adapted to the efficient use of limited frequency spectrum resources). Multiply the three indexes by the corresponding weight coefficients and then sum them up to obtain the basic objective function. The core objective is to maximize the weighted sum of the three, achieving optimal multi-dimensional resource utilization efficiency.

[0067] The constraint term is set, that is, the interference suppression constraint term and the power consumption constraint term are added to the basic objective function:

[0068] The interference suppression constraint term, i.e., based on the spatial isolation matrix, sets the interference power threshold between any two UAVs (such as not more than -80dBm), and by setting the penalty coefficient (when the interference exceeds the threshold, the penalty coefficient increases linearly according to the difference between the interference value and the threshold), ensures that the objective function avoids serious interference in the optimization process; the power consumption constraint term, i.e., combining the battery capacity of the UAV (such as the endurance of the reconnaissance aircraft is 4 hours, and the endurance of the delivery aircraft is 3 hours) and the communication power demand, sets the maximum communication power threshold of each UAV (such as 20dBm), and when the power consumption exceeds the threshold, through a negative correction factor, limits the resource allocation scheme with high power consumption.

[0069] Add the basic objective function to the two constraint terms to form the final spectrum resource allocation objective function, which not only ensures the maximum utilization efficiency of multi-dimensional resources, but also effectively suppresses interference and controls power consumption.

[0070] The penalty coefficient is linearly increased according to the severity of the interference exceeding the threshold, and is dynamically adjusted according to the spatial isolation of the UAV pair, ensuring that only slight interference is moderately punished and serious interference is forced to be avoided. The specific value range and conditions are as follows:

[0071] The overall value range of the penalty coefficient is [0.5, 5.0] (dimensionless, matching the weight of each index in the objective function (0.3-0.4) to avoid excessive or insufficient punishment), and the linear growth slope is 0.25 / dBm (i.e., for every 1dBm increase in interference difference, the penalty coefficient increases by 0.25).

[0072] For example, taking the interference power threshold between any two UAVs (-80dBm) as the reference, define the interference difference △P = actual interference power - interference power threshold (△P > 0 indicates exceeding the threshold, and △P ≤ 0 indicates not exceeding), and combine the spatial isolation value in the spatial isolation matrix (denoted as S, with a value range of [0, 1], S = 1 indicating complete isolation, and S = 0 indicating no isolation), to determine the penalty coefficient in different scenarios:

[0073] Scenario 1: The penalty coefficient is 0.5 (minimum penalty, only as a placeholder for the basic constraint term, and does not affect the core optimization of the objective function) when the interference threshold is not exceeded (△P ≤ 0). Applicable conditions: actual interference power ≤ -80dBm, regardless of the spatial isolation, it is determined as acceptable interference, only the minimum penalty is applied.

[0074] Scenario 2: The penalty coefficient is 0.5 + 0.25 × △P (linearly increasing, range: (0.5, 3.0]) when slightly exceeding the threshold (0 < △P ≤ 10dBm). Applicable conditions: actual interference power is between (-80dBm, -70dBm], and spatial isolation S ≥ 0.6 (high beam orthogonality of the UAV pair, with less interference impact).

[0075] Scenario 3: Moderate exceed threshold (10 < AP < 20 dBm) penalty coefficient value is 3.0 + 0.25 x (AP - 10) (linearly increasing, range: (3.0, 5.0]). Applicable conditions: actual interference power is in (-70 dBm, -60 dBm], or spatial isolation 0.3 < S < 0.6 (drone-to-beam orthogonality is moderate, interference has obvious impact).

[0076] Scenario 4: Serious exceed threshold (AP > 20 dBm) penalty coefficient is forced to be 5.0 (maximum penalty, makes the objective function value of the resource allocation scheme significantly reduced, almost discarded by the optimization model). Applicable conditions: actual interference power > -60 dBm, or spatial isolation S < 0.3 (drone-to-beam orthogonality is low, interference has serious impact on communication link). Example: AP = 25 dBm (actual interference power -55 dBm), S = 0.2, penalty coefficient = 5.0; even if AP = 18 dBm (actual interference power -62 dBm) but S = 0.2, still take 5.0 according to this scenario.

[0077] The negative correction factor is linearly increased according to the amplitude of power exceeding the threshold, and adapts to the difference in endurance of the drone (the delivery machine has shorter endurance and is more sensitive to power), which suppresses high-power schemes by reducing the objective function value. The specific value range and conditions are as follows:

[0078] The negative correction factor value range is [-0.8, 0] (dimensionless, matching the weight of each index in the objective function (0.3~0.4), negative value means reducing the score of the objective function), the linear growth slope is -0.1 / dBm (i.e. for every 1 dBm increase in power exceeding the threshold, the absolute value of the correction factor increases by 0.1, the reduction amplitude of the objective function score increases). The reconnaissance machine (endurance 4 hours), the correction factor value range is [-0.6, 0] (higher endurance redundancy, punishment is moderate); the delivery machine (endurance 3 hours), the correction factor value range is [-0.8, 0] (lower endurance redundancy, punishment is more strict). For example, taking the maximum communication power threshold of each drone (20 dBm) as the benchmark, define the power difference APtx = actual communication power - maximum communication power threshold (APtx > 0 indicates exceeding the threshold, APtx < 0 indicates not exceeding), and determine the negative correction factor according to the type of drone:

[0079] No exceeding power threshold (△Ptx≤0), that is, the correction factor is 0 (no negative correction, no impact on the target function score); applicable condition, that is, the actual communication power is less than or equal to 20dBm, regardless of the type of unmanned aerial vehicle, it is determined that the power consumption is reasonable. Slightly exceeding the power threshold (0<△Ptx≤6dBm), that is, the reconnaissance machine correction factor is 0-0.1x△Ptx (linearly increasing, range: (0,-0.6]); applicable condition, that is, the actual communication power is between (20dBm, 26dBm], and the endurance pressure is small. Delivery machine, that is, the correction factor is 0-0.13x△Ptx (linearly increasing, range: (0,-0.78], approximately -0.8); applicable condition: the actual communication power is between (20dBm, 26dBm], the endurance pressure is large, and the punishment is more severe.

[0080] Serious exceeding of the power threshold (△Ptx>6dBm); reconnaissance machine, that is, the correction factor is forced to be -0.6 (maximum negative correction, significantly reducing the target function score); applicable condition, actual communication power> 26dBm, endurance pressure increases dramatically, and is strictly limited. Delivery machine, that is, the correction factor is forced to be -0.8 (maximum negative correction, almost discarding the scheme); applicable condition: actual communication power> 26dBm, endurance cannot support long-term tasks, and high-power allocation is strictly prohibited.

[0081] According to the communication demand and topology structure of the unmanned aerial vehicle cluster, a constraint condition set of spectrum resource allocation is determined, specifically including: combining the communication demand and topology structure of the unmanned aerial vehicle cluster (star topology with command relay machine as the core and RIS assistance), the following constraint condition set is determined:

[0082] Bandwidth constraint, i.e. according to the business needs of different UAVs, set the minimum bandwidth threshold, 3 reconnaissance aircraft need to transmit high-resolution images and thermal imaging data, the minimum bandwidth is set to 20MHz / aircraft; 2 aircraft need to transmit control instructions and position information, the minimum bandwidth is set to 5MHz / aircraft; Command relay aircraft need to aggregate all data and issue instructions, the minimum bandwidth is set to 50MHz. Time delay constraint, i.e. the image data return time delay of reconnaissance aircraft is not more than 500ms, the instruction receiving time delay of delivery aircraft is not more than 100ms, the communication time delay of command relay aircraft and RIS array is not more than 50ms, to ensure the real-time of emergency rescue. Power constraint, i.e. the communication power of each UAV does not exceed the preset threshold (such as 20dBm), and the total power consumption does not exceed 30% of the battery capacity per hour (to avoid affecting the flight endurance). Interference constraint, i.e. the co-frequency interference power between any two UAVs does not exceed-80dBm, and the interference power between UAV and RIS array does not exceed-90dBm, to ensure the stability of communication link. Topology constraint, i.e. the spectrum resource allocation of all UAVs needs to be centered on the command relay aircraft to ensure that the command relay aircraft can receive all UAV data and issue scheduling instructions, and the spectrum resource configuration of RIS array needs to be synchronized with the UAVs in the coverage area.

[0083] Step 2.2, convert the objective function and constraint condition set into Ising model expression form, construct the Hamiltonian of the quantum annealing optimization model; according to the ground state energy distribution characteristics of the Hamiltonian, initialize the spin state of the quantum bit and the coupling strength parameter, which specifically includes:

[0084] This step constructs the Hamiltonian and initializes the quantum annealing related parameters according to the objective function and constraint condition set obtained in step 2.1, which provides model support for subsequent optimization solution, and the specific implementation process is as follows:

[0085] Ising model conversion of objective function and constraints: Ising model is to represent binary decision variables of resource allocation (such as 1 represents allocating a certain spectrum resource, -1 represents not allocating) by spin states (+1 or -1) of quantum bits. First, the decision variables in the spectrum resource allocation problem (carrier frequency selection, time slot allocation, power level selection) are mapped to the spin states of quantum bits, for example, 50 frequency sub-channels (800MHz-2.4GHz divided by 1MHz interval) available correspond to 50 quantum bits, and the spin state of each quantum bit represents whether the sub-channel is allocated to a certain unmanned aerial vehicle; similarly, 10 time slots (100ms each) correspond to 10 quantum bits, and 3 power levels (10dBm, 15dBm, 20dBm) correspond to 2 quantum bits (represented by binary coding). Then, the objective function constructed in step 2.1 is converted into the energy term of the Ising model, and the constraint condition set is converted into the penalty term (when the constraint condition is not satisfied, the penalty term will increase the system energy, prompting the optimization process to avoid such solutions), and finally the objective expression of the Ising model is formed.

[0086] Hamiltonian construction of quantum annealing optimization model: Hamiltonian is the core of quantum annealing model, which is used to describe the energy state of the system, composed of transverse magnetic field term and longitudinal magnetic field term:

[0087] Longitudinal magnetic field term, i.e. based on the objective expression of Ising model, reflects the energy distribution corresponding to the objective function and constraints, whose coefficients are determined by the weight coefficients of the objective function and the penalty coefficients of the constraints; for example, the weight coefficient of the spatial multiplexing gain index is directly mapped to the coupling strength of the corresponding quantum bit in the longitudinal magnetic field term. Transverse magnetic field term, i.e. used to simulate quantum tunneling effect, helps the system to jump out of local optimal solution, and its initial strength is set according to the strictness of the constraints (the stricter the constraints, the greater the initial transverse magnetic field strength); add the longitudinal magnetic field term and the transverse magnetic field term to form the complete Hamiltonian of the quantum annealing optimization model, and the ground state energy of the Hamiltonian corresponds to the optimal solution of spectrum resource allocation.

[0088] Quantum bit parameter initialization:

[0089] The spin state initialization, i.e., based on historical spectrum allocation data of the UAV cluster (if there is no historical data, random initialization is adopted), assigns an initial spin state (+1 or -1) to each quantum bit, for example, the quantum bit corresponding to the core spectrum resource (such as 1.8-1.9 GHz) of the command relay machine is initialized to +1 (preferential allocation), and the quantum bit corresponding to the spectrum resource susceptible to interference is initialized to -1 (temporary allocation); the coupling strength parameter initialization, i.e., according to the ground state energy distribution characteristics of the Hamiltonian, the coupling strength between quantum bits is calculated, the quantum bits corresponding to the resources with high correlation degree (such as sub-channels in the same aggregation area) are set to positive coupling strength (promote cooperative allocation), and the coupling strength of the quantum bits corresponding to the resources with low correlation degree is set to negative value (avoid conflict allocation); the absolute value range of the coupling strength is set to [0.1, 1.0], which ensures the stability of the quantum annealing process.

[0090] Step 2.3, by simulating the quantum tunneling effect, the transverse magnetic field strength is dynamically adjusted in the quantum annealing process, and when the energy converges to a stable threshold, the final spin state configuration of the quantum bit is recorded, which specifically includes:

[0091] This step is based on the quantum annealing optimization model constructed in step 2.2, and the parameters are dynamically adjusted by simulating the quantum tunneling effect until the system energy converges to obtain the optimal spin state configuration, and the specific implementation process is as follows:

[0092] The quantum annealing process initialization, i.e., setting the number of iterations of quantum annealing (such as 1000 times), the energy convergence threshold (such as the energy change amount of continuous 50 iterations is less than 10^-6), the initial temperature (such as 100K, simulating the thermal equilibrium state of quantum system); the quantum bit spin state and coupling strength parameter initialized in step 2.2 are input into the quantum annealing simulator, and the optimization process is started.

[0093] Dynamic adjustment of transverse magnetic field strength, i.e., the core of quantum annealing is to realize the transition from quantum state to classical state through the attenuation of transverse magnetic field strength:

[0094] In the initial stage (first 300 iterations), the transverse magnetic field strength is kept at a maximum value (e.g. 10.0), and the strong transverse magnetic field causes the spin state of the quantum bit to frequently flip, simulating quantum tunneling effect, helping to explore the entire solution space and avoid falling into a local optimal solution; in the middle stage (301-800 iterations), the transverse magnetic field strength is gradually reduced according to an exponential decay law (decay coefficient is 0.99 / iteration), at this time the quantum tunneling effect is weakened, and the system begins to converge to a state with lower energy; in the later stage (801-1000 iterations), the transverse magnetic field strength is reduced to a minimum value (e.g. 0.1), the system is basically in a classical state, the spin state tends to be stable, and only fine-tuning is performed near the local optimal solution. Energy convergence monitoring and judgment: during each iteration, the energy change curve is recorded according to the current energy value of the Hamiltonian; when the energy change of 50 consecutive iterations is less than the preset convergence threshold, it is determined that the system energy converges to a stable state, and the corresponding energy value is the approximate ground state energy of the Hamiltonian; if the number of iterations reaches the upper limit and still does not converge, the decay coefficient of the transverse magnetic field strength is adjusted appropriately (e.g. changed to 0.98 / iteration), and the annealing process is restarted until the energy converges. When the energy converges to a stable threshold, the quantum annealing process is stopped, and the final spin state configuration of all quantum bits (the spin state of each quantum bit is +1 or -1) is recorded. The configuration corresponds to the optimal or suboptimal solution of the spectrum resource allocation problem.

[0095] Step 2.4, according to the final spin state configuration of the quantum bit, the spectrum resource allocation scheme corresponding to each unmanned aerial vehicle is decoded; the spectrum resource allocation scheme is mapped to a specific carrier frequency allocation table, a time slot offset configuration table and a power control parameter set, forming a complete spectrum resource scheduling instruction set, specifically including:

[0096] This step forms a specific resource allocation scheme and scheduling instruction set by decoding and mapping the final spin state configuration of the quantum bit obtained in step 2.3, providing executable communication scheduling basis for the unmanned aerial vehicle cluster, and the specific implementation process is as follows:

[0097] Quantum bit spin state decoding: according to the mapping relationship between the quantum bit spin state and the resource allocation decision variable established in step 2.2, the final spin state configuration is decoded:

[0098] Carrier frequency decoding, i.e. decoding the quantum bit spin state +1 corresponding to the frequency subchannel as allocating the subchannel, and -1 as not allocating the subchannel; combined with the normalized subcarrier aggregation area divided in step 2.12, the subchannels decoded as allocated in the same aggregation area are integrated to form the carrier frequency set of each unmanned aerial vehicle (such as reconnaissance machine 1 allocating 20 subchannels of 1.2GHz-1.22GHz, meeting the 20MHz bandwidth requirement). Time slot offset decoding, i.e. decoding the quantum bit spin state +1 corresponding to the time slot as occupying the time slot, and -1 as not occupying the time slot; according to the normalized time window length obtained in step 2.12, a continuous time slot block is allocated to each unmanned aerial vehicle (such as delivery machine 1 allocating time slots 2-3, corresponding to a time length of 200ms, meeting the low latency requirement), and the time slot offset of each unmanned aerial vehicle is calculated (i.e. the time difference between the first occupied time slot and the system starting time slot, such as an offset of 100ms).

[0099] Power control parameter decoding, i.e. decoding the combination of quantum bit spin states corresponding to the power level as a specific power value (such as spin state “+1, +1” decoding as 20dBm, “+1, -1” decoding as 15dBm, and “-1, +1” decoding as 10dBm); combined with the power constraint term in step 2.14, ensure that the decoded power value does not exceed the pre-set threshold, and adjust the power value according to the spatial isolation matrix (such as when the spatial isolation between unmanned aerial vehicles is low, the power is appropriately reduced to suppress interference).

[0100] Resource allocation scheme mapping and table generation:

[0101] Carrier frequency allocation table generation, i.e. classified by unmanned aerial vehicle number (command relay machine, reconnaissance machines 1-3, delivery machines 1-2), record the carrier frequency range, subchannel number and corresponding RIS array number of each unmanned aerial vehicle (such as the carrier frequency 1.5GHz-1.52GHz of reconnaissance machine 2, corresponding to RIS array 2), the table clearly marks the priority of frequency resource use (the reconnaissance machine in the key search and rescue area has the highest priority); time slot offset configuration table generation, i.e. recording the time slot block number, time slot offset, time slot length and scheduling period of each unmanned aerial vehicle (such as the time slot block 0-4 of the command relay machine, the offset 0ms, the length 500ms, and the scheduling period 1s), to ensure that the time slots of each unmanned aerial vehicle do not overlap, avoiding time domain conflict; power control parameter set generation, i.e. recording the transmission power value, power adjustment step (such as 5dB / step), maximum power threshold and power optimization period (such as 10s / time) of each unmanned aerial vehicle, supporting dynamic adjustment of power according to channel state.

[0102] The complete scheduling instruction set is formed, that is, the above three tables are integrated, classified according to the types of unmanned aerial vehicles and task stages, and a complete frequency spectrum resource scheduling instruction set is generated, the instruction set including frequency allocation instructions, time slot configuration instructions, power control instructions, each instruction being attached with an execution timestamp (based on Beidou timing, accuracy 1 ms) and verification feedback requirements (the unmanned aerial vehicle needs to feed back the execution result to the command relay machine after execution); finally, the scheduling instruction set is issued to each unmanned aerial vehicle and RIS array through the stable communication link established in step 1, ensuring that all devices execute the frequency spectrum resource allocation scheme synchronously, and realizing efficient scheduling of multi-machine cooperative communication.

[0103] In a preferred embodiment of the present application, step 3, according to the carrier frequency, time slot offset and power control parameter set, the overlapping communication frame structure is constructed; according to the overlapping communication frame structure, a dynamic frame scheduling scheme is generated, including:

[0104] Step 3.1, based on the carrier frequency allocation table, an independent subcarrier group is allocated to each unmanned aerial vehicle to obtain a subcarrier group allocation result; based on the time slot offset configuration table, a frame start time offset is set for each unmanned aerial vehicle; based on the power control parameter set, a corresponding transmission power level is configured for each unmanned aerial vehicle, specifically including:

[0105] This step takes the carrier frequency allocation table, time slot offset configuration table and power control parameter set output in step 2.4 as input, completes the basic resource anchoring of unmanned aerial vehicle communication, and provides subcarrier-time-power three-dimensional parameter support for subsequent frame structure construction, and the specific implementation process is as follows:

[0106] Subcarrier group allocation, that is, extracting the bandwidth requirement (20MHz / aircraft for reconnaissance machine, 5MHz / aircraft for delivery machine, 50MHz for command relay machine), carrier frequency range and subchannel flatness index of each unmanned aerial vehicle in the carrier frequency allocation table in step 2.4, taking frequency domain flatness ≥0.8, subcarrier spacing 15kHz and protection bandwidth 500kHz as the division principle, to ensure that the frequency domain response in each subcarrier group is stable and the interference in the group is reduced.

[0107] Differential subcarrier group allocation, namely command relay machine: allocate 333 subcarriers (including guard bandwidth) of 1.8GHz-1.85GHz frequency band, split according to control channel subcarrier group and data aggregation subcarrier group, of which control channel group accounts for 10% (33 subcarriers, used for issuing scheduling instructions), and data aggregation group accounts for 90% (300 subcarriers, used for receiving data of each unmanned aerial vehicle); reconnaissance machine (3), namely each is allocated a continuous 20MHz frequency band (such as reconnaissance machine 1 is allocated 1.2GHz-1.22GHz), corresponding to 1333 subcarriers, split according to image data subcarrier group + thermal imaging data subcarrier group, the proportion is 7:3 (image data needs higher bandwidth); delivery machine (2), namely each is allocated a 5MHz frequency band (such as delivery machine 1 is allocated 1.5GHz-1.505GHz), corresponding to 333 subcarriers, a single group can meet the control instruction and position feedback requirements.

[0108] Assign a unique identifier to each subcarrier group (such as reconnaissance machine 2-image group-002), record the subcarrier number range, center frequency, bandwidth and corresponding RIS array coverage range in the group, simulate the adjacent frequency interference between subcarrier groups through the spectrum analyzer, ensure that the isolation between groups is ≥30dB, and form the final subcarrier group allocation result table.

[0109] Frame start time offset setting, namely the time slot block information of step 2.4 time slot offset configuration table, determines the core parameters of the communication frame, the frame period is set to 10ms (adapted to the channel coherence time, reduces the influence of channel change on frame transmission), each frame contains 10 time slots (each time slot is 1ms), the frame header occupies 0.5ms, and the frame tail is reserved 0.2ms as a guard interval to avoid frame crosstalk. Differential calculation of offset, taking the frame start time of the command relay machine as the system reference time (t=0), calculating the frame start offset according to the time slot offset requirements of each unmanned aerial vehicle: reconnaissance machine, because high-resolution data needs to be transmitted, the frame start offset is set to 1ms, 3ms and 5ms staggered (such as reconnaissance machine 1 offset 1ms, reconnaissance machine 2 offset 3ms), to ensure that the data frames are partially overlapped in time domain but the core data area does not conflict; delivery machine, the frame start offset is set to 2ms and 4ms, staggered with the reconnaissance machine, while reserving 0.3ms of redundant offset to cope with the time synchronization error caused by the change of unmanned aerial vehicle flight attitude; all offsets are calibrated based on the Beidou timing system, and the synchronization accuracy is controlled within 10μs.

[0110] Offset configuration and feedback, write the frame start time offset into the communication module register of each unmanned aerial vehicle, after configuration is completed, the unmanned aerial vehicle feeds back the synchronization state to the command relay machine through the control channel, if the synchronization error exceeds 50μs, then the offset is adjusted again until the requirements are met.

[0111] The transmit power level configuration, i.e. extracting the power threshold (20 dBm) in the power control parameter set in step 2.4, adjusting the step size (5 dB / step), combined with the real-time signal-to-noise ratio (SNR) fed back by the channel state tensor in step 1.4, establishes an "SNR-power level" mapping table: when SNR≥20 dB, it is level 1 (10 dBm, low power energy saving); when 10 dB≤SNR<20 dB, it is level 2 (15 dBm, balance rate and power consumption); when SNR<10 dB, it is level 3 (20 dBm, high power to ensure communication).

[0112] The UAV type differentiation configuration, i.e. the command relay machine: fixed power level 2 (15 dBm), ensuring coverage of all UAVs and stable power consumption; the reconnaissance machine: matching the level according to the mapping table during normal reconnaissance, and forced to level 3 when suspected signs of life are found, to ensure data transmission reliability; the delivery machine: mainly level 1 or 2, reduced to level 1 when approaching the delivery point (distance to target≤100 m) to avoid high power interference with ground receiving equipment.

[0113] The power level parameter is written through the control interface of the UAV on-board power amplifier, and after configuration, the power calibration process is started. The command relay machine receives the pilot signal of the UAV, measures the deviation between the actual received power and the theoretical value, and if the deviation exceeds 2 dB, it is corrected through a power adjustment instruction until the deviation≤1 dB, forming the final transmit power level configuration result.

[0114] In step 3.2, according to the subcarrier group allocation result and the frame start time offset, a time-frequency two-dimensional overlapping frame structure is constructed; in the time-frequency two-dimensional overlapping frame structure, a control channel region, a data channel region and a pilot channel region are set; according to the transmit power level, power allocation proportions are configured for different regions, specifically including:

[0115] This step takes the subcarrier group allocation result of step 3.1, the frame start time offset, and the transmit power level as input, constructs a frame structure with overlapping characteristics and region division, and completes power adaptation for each region. The specific implementation process is as follows:

[0116] The time-frequency two-dimensional overlapping frame structure is constructed, i.e. taking time-frequency as the two-dimensional coordinate axis, the time axis is divided into 10 time slots (numbered 0-9) with 1 ms as the frame period, and the frame start offset of each UAV is labeled; the frequency axis is divided according to the subcarrier group, and the center frequency and bandwidth range of each UAV subcarrier group are labeled (such as the image group center frequency 1.3 GHz and bandwidth 14 MHz of reconnaissance machine 2).

[0117] The overlapping characteristic implementation logic is that the time domain overlap is implemented by using the frame start offset difference of each unmanned aerial vehicle, and the frequency multiplexing is implemented by using the frequency domain isolation of the subcarrier group. For example, the frame (t=1 ms-11 ms) of the reconnaissance machine 1 and the frame (t=2 ms-12 ms) of the delivery machine 1 overlap in the time domain in the period of t=2 ms-11 ms, but the subcarrier groups are located in the frequency bands of 1.2 GHz and 1.5 GHz respectively, and there is no frequency domain conflict. The frame (t=0-10 ms) of the command relay machine overlaps with the frames of all unmanned aerial vehicles, and the signal is distinguished by using the exclusive subcarrier group.

[0118] The frame structure visualization and checking is that a time-frequency two-dimensional overlapping frame structure heat map is generated, the horizontal axis is time (0-12 ms), the vertical axis is frequency (0.8 GHz-2.4 GHz), the frame coverage area of each unmanned aerial vehicle is marked by different colors, and it is checked whether there is a simultaneous frequency conflict (the same time-frequency resource is occupied by multiple machines) in the overlapping area. If there is, return to step 3.1 to adjust the offset or the subcarrier group until there is no simultaneous frequency conflict.

[0119] The intra-frame channel region division is that the regions are divided according to the principle of control priority, data priority and pilot assistance, and the total proportion satisfies the frame period of 10 ms. That is, the control channel region accounts for 5% (0.5 ms) of the frame period, is located at the frame head position, and is used to transmit key control information such as frame synchronization signal, scheduling instruction and state feedback. The data channel region accounts for 85% (8.5 ms) of the frame period, is located in the middle of the frame, and is subdivided into sub-regions (such as image sub-region and thermal imaging sub-region of the reconnaissance machine) according to the service type. The pilot channel region accounts for 10% (1 ms) of the frame period, is distributed in the data channel region at an interval of inserting a pilot every 0.85 ms, and is used for real-time channel estimation.

[0120] The subcarrier allocation of each region is that the control channel region is allocated 10% of the subcarriers in the subcarrier group (for example, 133 of the 1333 subcarriers of the reconnaissance machine 1 are used for the control channel), and QPSK modulation (strong anti-interference capability) is used. The data channel region is allocated 80% of the subcarriers in the subcarrier group, the reconnaissance machine uses 64QAM modulation (high bandwidth), and the delivery machine uses 16QAM modulation (balance between rate and reliability). The pilot channel region is allocated 10% of the subcarriers in the subcarrier group, and a fixed pilot sequence (Zadoff-Chu sequence) is used to ensure the accuracy of channel estimation.

[0121] The region boundary identification setting is that a synchronization identification sequence (such as a 16-bit fixed binary code 1100110011001100) is inserted at the start and end positions of each region, which is used for the unmanned aerial vehicle communication module to quickly identify the region boundary and avoid confusion between data and control signals.

[0122] The power level-based regional power distribution is based on the transmission power level of step 3.1 as the total power reference, and the proportion is divided according to the principles of high priority of control channel, medium priority of pilot channel and on-demand allocation of data channel, while combining the spatial isolation adjustment of the channel state tensor. The unmanned aerial vehicle with low spatial isolation (easy to be interfered) improves the proportion of control channel power.

[0123] The differentiated power allocation scheme is as follows: power level 1 (10 dBm): control channel accounts for 40% (4 dBm), pilot channel accounts for 30% (3 dBm), and data channel accounts for 30% (3 dBm), which is suitable for low-bandwidth and low-interference scenarios such as delivery machines; power level 2 (15 dBm): control channel accounts for 30% (4.5 dBm), pilot channel accounts for 20% (3 dBm), and data channel accounts for 50% (7.5 dBm), which is suitable for stable communication scenarios such as command relay machines; power level 3 (20 dBm): control channel accounts for 35% (7 dBm), pilot channel accounts for 25% (5 dBm), and data channel accounts for 40% (8 dBm), which is suitable for high-interference scenarios such as reconnaissance machines.

[0124] The received power and signal-to-noise ratio of each region are calculated by the link budget tool to ensure that the SNR of the control channel is greater than or equal to 15 dB (to ensure reliable reception of commands), the SNR of the data channel is greater than or equal to 10 dB (to ensure data transmission rate), and the SNR of the pilot channel is greater than or equal to 20 dB (to ensure channel estimation accuracy). If not, adjust the proportion until it meets the standard.

[0125] Step 3.3, based on the time-frequency two-dimensional overlapping frame structure, analyze the interference mode and channel occupation between frames to obtain the channel occupation and interference mode; according to the channel occupation and interference mode, determine the frame scheduling conflict detection rule and priority determination criterion, specifically including:

[0126] With the time-frequency two-dimensional overlapping frame structure of step 3.2 as input, analyze the interference and channel occupation state between frames, establish conflict detection and priority rules, and provide basis for subsequent scheduling decisions. The specific implementation process is as follows:

[0127] The analysis of the interference mode and channel occupation between frames is as follows: using the command relay machine, collect interference data within 100 consecutive frame periods, combine the channel state tensor of step 1.4, and divide it into three categories according to the source and nature of the interference: same frequency interference, i.e. interference between different unmanned aerial vehicle frame overlapping regions within the same subcarrier group (such as conflict caused by subcarrier group configuration error); adjacent frequency interference, i.e. interference between adjacent subcarrier groups (such as crosstalk between the 1.22 GHz subcarrier group of reconnaissance machine 1 and the 1.23 GHz subcarrier group of reconnaissance machine 2); frame overlapping interference, i.e. time domain overlapping but frequency domain isolated interference between frames (such as overlapping frames of reconnaissance machine 1 and delivery machine 1, interference caused by power overflow).

[0128] Interference mode feature extraction, that is, three types of interference are extracted respectively. The same frequency interference is interference power, conflict time, and involved UAV pair. The adjacent frequency interference is interference power, frequency offset value, and subcarrier spacing. The frame overlap interference is interference power, overlap time, and power level difference. Through clustering algorithm (such as K-Means), the interference mode is classified to form a typical mode library such as same frequency strong interference (> -70 dBm), adjacent frequency medium interference (-80 dBm ~ -70 dBm), and frame overlap weak interference (< -80 dBm).

[0129] Channel occupation quantitative analysis, that is, based on the time-frequency two-dimensional overlapping frame structure heat map, the occupation rate of each time-frequency resource block (1ms x 1MHz) is counted, and is classified according to high occupation (> 80%), medium occupation (50%-80%), and low occupation (< 50%). At the same time, the channel utilization rate of each UAV (actual data transmission amount of data channel / maximum capacity of channel) is calculated to form a channel occupation situation table, and the high occupation period (such as t = 3ms-7ms, multi-UAV frame overlap) and the high utilization rate UAV (such as reconnaissance machine 2, continuous transmission of image data) are marked.

[0130] Frame scheduling conflict detection rule establishment, that is, combined with interference mode and channel occupation analysis, three types of core conflicts are defined: time-frequency conflict, that is, the same time-frequency resource block is occupied by multiple UAVs at the same time (the core reason of same frequency interference); power conflict, that is, the transmission power of a certain area exceeds the channel capacity, causing the interference of adjacent frequency UAVs to exceed the standard (the core reason of adjacent frequency interference); synchronization conflict, that is, the frame start time offset error of UAVs exceeds the protection interval, causing frame header overlap (the core reason of frame overlap interference).

[0131] Differential detection rule making: time-frequency conflict detection, that is, the occupation state of each time-frequency resource block is monitored in real time. If a resource block is occupied by 2 or more UAVs at the same time, a conflict alarm is triggered, and the detection accuracy is 10us x 100kHz; power conflict detection, that is, the transmission power of each UAV is collected in real time. If the power of a certain area exceeds 10% of the upper limit of the power allocation of the area, or the interference power of the adjacent frequency area exceeds -80dBm, a conflict alarm is triggered; synchronization conflict detection, that is, through the receiving time difference of the frame header synchronization identification sequence, the deviation of the actual frame start time of the UAV from the preset offset is calculated. If the deviation exceeds 0.2ms (the maximum value of the protection interval), a conflict alarm is triggered.

[0132] Conflict level division, that is, according to the influence degree of conflict on communication, the conflict is divided into three levels: first level (emergency), that is, time-frequency conflict and involves key data transmission of reconnaissance machine; second level (important), that is, power conflict or synchronization conflict, which affects data transmission rate; third level (general), that is, frame overlap weak interference, which does not affect core business. Different levels correspond to different processing priorities.

[0133] The priority determination criterion is established, that is, the three-dimensional determination system of the priority of the type of the unmanned aerial vehicle, the priority of the business urgency and the priority of the channel quality is established in combination with the rescue task characteristics of the unmanned aerial vehicle cluster, and the weights of each dimension are 0.5, 0.3 and 0.2 respectively (the priority of the type of the unmanned aerial vehicle is the core). Each dimension priority is subdivided, that is, the priority of the type of the unmanned aerial vehicle (weight 0.5) is that the command relay machine (1st, highest) > three-dimensional modeling reconnaissance machine (2nd) > material delivery machine (3rd, lowest); the priority of the business urgency (weight 0.3) is that the transmission of life heat source detection data (1st) > the transmission of landslide area image (2nd) > the transmission of routine investigation data (3rd) > the feedback of the state of the delivery machine (4th) > the transmission of non-urgent instructions (5th). The priority of the channel quality (weight 0.2) is that the channel quality is good (SNR ≥ 20 dB, 1st) > the channel quality is good (10 dB ≤ SNR < 20 dB, 2nd) > the channel quality is poor (SNR < 10 dB, 3rd), and the priority of the unmanned aerial vehicle with poor channel quality is appropriately improved to avoid communication interruption. The weighted summation method is used to calculate the real-time priority score of each unmanned aerial vehicle (score = type score × 0.5 + business score × 0.3 + channel score × 0.2, and the score difference of each level is 2 points), and the lower the priority score is, the higher the priority is, so the priority of the reconnaissance machine is higher than that of the delivery machine.

[0134] Step 3.4, according to the frame scheduling conflict detection rule and the priority determination criterion, a frame scheduling decision tree is constructed; based on the frame scheduling decision tree, combined with real-time channel state feedback information, an adaptive dynamic frame scheduling scheme is generated, the dynamic frame scheduling scheme includes a frame scheduling time sequence table, a power adjustment instruction and a conflict avoidance strategy, specifically including:

[0135] With the conflict detection rule (including conflict type, level division) and the priority determination criterion (three-dimensional weighted system) of step 3.3 as input, a hierarchical decision tree model is constructed, the key of which is to ensure the accuracy and uniqueness of the decision output through multi-dimensional decision nodes and differentiated scheduling strategies, specifically including:

[0136] A four-level progressive decision node design is adopted, the division of each node is directly related to the technical output of the previous step, ensuring the coherence and constraint effectiveness of the decision logic, and the specific levels are as follows:

[0137] The first level (root node) is the conflict existence determination node, the real-time frame scheduling conflict monitoring result is input, and two branches of existing conflict / no conflict are output; this node directly calls the conflict detection rule of step 3.3, taking the time-frequency resource block occupation state, the power threshold and the synchronization deviation as the monitoring indexes, to realize the rapid preliminary judgment of the conflict.

[0138] The second level, i.e. the conflict level decision node, is effective only for the branch with conflict, and inputs the interference power or the service type involved in the conflict, and outputs three branches of level one (emergency) / level two (important) / level three (general); the level division directly uses the conflict level standard of step 3.3, wherein the level one conflict specifically refers to the time-frequency conflict involving the transmission of life source detection data of the unmanned aerial vehicle, which is the highest priority processing scenario.

[0139] The third level, i.e. the priority score decision node, inputs the real-time priority score of the unmanned aerial vehicle involved in the conflict (calculated according to the three-dimensional weighting method of step 3.3, the score range is 1.0-5.0, and the lower the score, the higher the priority), and outputs two branches of score≤2.0 (high priority) / score>2.0 (low priority); the node is the core embodiment of the service priority guarantee technical target, and ensures that the communication resources of the high-priority unmanned aerial vehicle (such as a reconnaissance aircraft) are preferentially satisfied.

[0140] The fourth level, i.e. the channel quality decision node, inputs the real-time signal-to-noise ratio (SNR) of the high-priority unmanned aerial vehicle, and outputs three branches of SNR≥20dB (excellent) / 10dB≤SNR<20dB (good) / SNR<10dB (poor); the node provides a basis for channel adaptation, and avoids communication interruption due to fixed scheduling in a poor channel scenario.

[0141] The scheduling strategy of the decision tree leaf node is defined, each terminal (leaf node) of each decision path corresponds to a unique scheduling strategy, and the executability of the decision output is ensured, and the strategy is as follows (covering typical scenarios):

[0142] Path 1 (conflict exists, level one, score≤2.0, and any channel quality): the high-priority unmanned aerial vehicle (such as a reconnaissance aircraft transmitting life detection data) maintains the current frame parameters (subcarrier group, frame offset), and the low-priority unmanned aerial vehicle (such as a delivery aircraft) immediately executes the combined strategy of frame offset, 1ms adjustment and power level reduction by 1 level, to ensure uninterrupted communication of high-priority services.

[0143] Path 2 (conflict exists, level two, score>2.0, and SNR<10dB): all conflict-involved unmanned aerial vehicles synchronously execute the strategy of shifting the subcarrier group to the protection bandwidth by 1MHz and extending the frame protection interval to 0.3ms, which avoids the conflict and improves the transmission reliability in a poor channel scenario.

[0144] Path 3 (no conflict, no level, no score, and any channel quality): the current frame parameters are maintained, and only the data channel modulation mode is dynamically adjusted according to the channel quality (excellent, adjusted to 64QAM; good, adjusted to 16QAM; poor, adjusted to QPSK), to achieve the balance between resource waste and stable communication.

[0145] To ensure the practicability of the decision model, historical conflict data sets and cross-validation methods are used to optimize the decision tree: 100 groups of measured conflict data in the earthquake zone (including different terrains and business types) are selected as the training set, and 20 groups of data are selected as the test set. The success rate of conflict resolution is greater than or equal to 95%, and the delay of scheduling is less than or equal to 1ms. The decision logic is optimized by adjusting the node decision threshold (such as the priority score threshold). The final decision tree model has a unique mapping relationship between input and output.

[0146] Each UAV uploads three-dimensional feedback data to the command relay machine through the control channel after completing one frame of data transmission (i.e., every 10ms, synchronized with the frame period). The dimensions are selected to correspond to the needs of the decision tree nodes one by one, avoiding redundancy:

[0147] Scheduling related dimensions: frame transmission status (success / failure / retransmission times), current business type (such as life detection-scouting machine 1); channel quality dimensions: real-time SNR, bit error rate (BER), channel impulse response; device state dimensions: UAV remaining power, flight attitude (affecting antenna beam pointing, indirectly related to channel quality).

[0148] The command relay machine processes the feedback data in two steps: cleaning, removing abnormal data with BER>10%, and calibrating SNR (correcting measurement bias caused by terrain shielding) through the channel state tensor in step 1.4; standardization: converting multi-dimensional data into structured data of UAV ID-real-time priority score-channel quality level-business state, directly adapting the input interface of the decision tree, ensuring that the data transmission delay to the decision tree is less than or equal to 0.5ms.

[0149] Based on the output results of the decision tree and the standardized feedback data, a complete scheme including frame scheduling timing table, power adjustment instructions and conflict avoidance strategies is generated. Each component has clear technical features and is related to each other, specifically:

[0150] Using time axis and UAV ID as two-dimensional indexes, the core parameters of each UAV in the next 100 frame periods (1s, considering real-time and planning) are clearly defined. Key technical fields in the table include:

[0151] Basic parameters: frame start time (accurate to 10μs), frame end time, subcarrier group number (related to the allocation results in step 3.1).

[0152] Priority identification: high-priority business is marked as emergency support, and 5% of spare time-frequency resources are reserved (such as 1.9GHz-1.905GHz subcarrier group).

[0153] Modulation method: according to the channel quality, the modulation type of the data channel is marked (such as "scout 2-64QAM").

[0154] Power adjustment instructions (energy and communication balance): based on decision tree strategy and device state dimension data generation, instruction format uses machine recognizable code and human readable description dual mode, core technical features include:

[0155] Instruction parameters: target power level (1-3 levels, associated with step 3.1), power ratio of each channel area (control / data / pilot, associated with step 3.2), execution timestamp (accurate to frame start time);

[0156] If the remaining power of the UAV is ≤20%, the instruction is forced to add a power level upper limit of 2 levels; if SNR <10dB, the instruction marks the emergency permission to temporarily break through the power threshold of 10% (≤22dBm). For potential conflicts identified by the decision tree (discovered through time sequence table rehearsal, such as t=50ms when reconnaissance machine 3 and delivery machine 2 will overlap frames), develop a hierarchical avoidance strategy, with the following technical priorities:

[0157] First-level avoidance (priority): time domain avoidance, i.e. adjust the frame start offset of low-priority UAVs (e.g. delivery machine 2 from 4ms to 6ms), with no frequency domain resource waste; second-level avoidance (backup), frequency domain avoidance, i.e. shift the subcarrier group of high-priority UAVs to the guard bandwidth (e.g. reconnaissance machine 3 shifts 1MHz), to avoid interference; third-level avoidance (emergency), power avoidance, i.e. reduce the power level of low-priority UAVs by 1 level, only used when time and frequency domain avoidance is not feasible.

[0158] Build a communication scenario in the earthquake zone (including terrain shielding and multipath fading model), input the scheduling scheme to verify the core indicators: conflict resolution success rate ≥95%; business transmission rate (reconnaissance machine ≥100Mbps, delivery machine ≥10Mbps); scheduling delay ≤1ms; if the indicators are not met, return to the decision tree adjustment node strategy and generate a new scheme to ensure the practicality of the output scheme. The verified scheme is sent through the encrypted control channel of the command relay machine (using AES-128 encryption), and each UAV receives and completes parameter configuration within 1 frame period (10ms); to avoid synchronization errors, use the Beidou timing and frame header synchronization identifier dual synchronization mechanism to ensure that the frame parameter adjustment of all UAVs takes effect at the same time (time deviation ≤10μs); after the UAV executes the scheme, it immediately returns the execution result code (such as 00-success, 01-power overrun) through the feedback channel; the command relay machine has an emergency scheduling module, which triggers the backup strategy (such as enabling the reserved backup subcarrier group) immediately if it receives a failure code such as 01, and generates a new adjustment instruction within 0.5ms to realize the fast closed loop of fault-repair, ensuring the continuity of the communication in the earthquake rescue area.

[0159] In a preferred embodiment of the present application, step 4, based on the dynamic frame scheduling scheme, extracts the channel reciprocity feature sequence; generates quantum random numbers according to the channel reciprocity feature sequence; generates session keys and device authority topology according to the quantum random numbers, including:

[0160] Step 4.1, based on the frame scheduling timing table in the dynamic frame scheduling scheme, configures a bidirectional channel sounding time slot; in the bidirectional channel sounding time slot, controls the reciprocity channel measurement between the UAV pairs to obtain uplink channel response data and downlink channel response data, specifically including:

[0161] With the frame scheduling timing table of step 3.4 as input, through time slot accurate configuration and bidirectional channel synchronous measurement, uplink / downlink channel data supporting reciprocity analysis are obtained, and the specific implementation process is as follows:

[0162] Based on the frame start offset, time slot occupancy range and business priority of each UAV in the frame scheduling timing table, the sounding time slot is configured, that is, the time slot segment of the core business (such as reconnaissance machine life sounding data transmission) is prohibited to insert the sounding time slot, and the frame guard interval (0.2ms), low traffic time slot (such as delivery machine state feedback gap) or reserved time slot (safety configuration time slot marked in the timing table) is preferentially selected to ensure that sounding and communication business do not conflict with each other.

[0163] The core parameters of the sounding time slot are uniformly set, that is, the single time slot length is 0.2ms (matching the channel coherence time to avoid measurement failure caused by channel mutation), the time slot interval is 10ms (consistent with the frame period to ensure periodic monitoring), and each sounding period contains 2 consecutive sub-slots (sub-slot 1: uplink sounding, sub-slot 2: downlink sounding, interval 0.05ms for transceiver switching); in the frequency domain, 5% of the subcarriers in each UAV subcarrier group are allocated to the sounding time slot (such as 67 out of 1333 subcarriers of reconnaissance machine 1 for sounding), using orthogonal frequency division multiplexing (OFDM) mode, subcarrier spacing 15kHz, to avoid frequency domain interference with data / control channels.

[0164] Combined with the frame scheduling timing table, a sounding time slot configuration table is generated for each UAV, marking the sounding time slot start time (accurate to 10us), subcarrier number, and transceiver switching time, etc. For example, the sounding time slot of reconnaissance machine 1 is t=1.8ms-2.0ms (sub-slot 1: 1.8ms-1.9ms, sub-slot 2: 1.95ms-2.05ms), and the subcarrier number is 1-67; the configuration table is issued to each UAV through an encrypted control channel, and the UAV writes the parameters into the communication module register after receiving, completing the time slot configuration initialization.

[0165] Based on the star topology of the UAV cluster (command relay machine as the core), determine the measurement link type: point-to-point link between the command relay machine and each UAV (a total of 6, including 1 command machine, 3 reconnaissance machines, and 2 delivery machines); link between UAVs of the same type (such as reconnaissance machine 1 and reconnaissance machine 2, a total of 3), ensuring coverage of all communication scenarios; measure in units of UAV pairs (such as command machine-reconnaissance machine 1 pair), and execute the measurement in order according to priority (core link is prior to cooperative link).

[0166] Synchronization is achieved using a unified triggering mechanism for the command machine. The command relay machine sends a synchronization trigger signal (using a 16-bit binary synchronization code 1010101011001100) to the UAV pairs participating in the measurement 10 μs before the start of the detection time slot. After receiving the synchronization signal, the UAV pair sends a pilot signal to the command machine in sub-slot 1 (uplink), and the command machine sends a pilot signal to the UAV in sub-slot 2 (downlink). The pilot sequence uses the Zadoff-Chu sequence (length 256) with good orthogonality to ensure that the uplink and downlink pilots do not interfere with each other.

[0167] Uplink and downlink channel response data collection: uplink data collection, the command machine receives the pilot signal sent by the UAV in sub-slot 1, calculates the uplink channel response data based on the least squares algorithm, including channel amplitude, phase, delay spread, etc. Parameters, synchronously record the collection timestamp and UAV position information; downlink data collection: the UAV receives the pilot signal sent by the command machine in sub-slot 2, calculates the downlink channel response data using the least squares algorithm, and also records the timestamp and position information; the collected data is stored in a temporary buffer area, and the data is uploaded to the command machine once every 10 ms.

[0168] After the command machine receives the uplink and downlink data, it first verifies the consistency of the timestamp (the timestamp deviation of the uplink and downlink data should be ≤5 μs) and the stability of the position (the position of the UAV during the measurement should change ≤0.5 m). If not, mark it as invalid data and trigger a re-measurement of the next detection period; valid data is associated with the corresponding UAV pair identifier and detection time slot information to form a structured data set of UAV pair-uplink and downlink channel response-time and space parameters.

[0169] Step 4.2. Perform reciprocity verification processing on the uplink and downlink channel response data, and calculate the channel reciprocity error index; when the channel reciprocity error index is less than the preset threshold, extract the channel amplitude fluctuation sequence and phase jitter sequence, and combine to generate the channel reciprocity feature sequence, which specifically includes:

[0170] Firstly, the uplink channel response data and the downlink channel response data are preprocessed to ensure the consistency of the time and space reference of the two types of data. Based on the timestamps and location information recorded in step 4.1, the uplink and downlink data within the same drone pair and the same detection time slot are accurately aligned. Abnormal data with a timestamp deviation exceeding 5μs or a change in the drone's position during the measurement period exceeding 0.5m are excluded to avoid distortion of the reciprocity judgment caused by device movement or synchronization errors. For the valid data after alignment, the core response parameters of the uplink and downlink channels are extracted, including channel amplitude, phase, delay spread, and frequency fading coefficient, to form paired parameter sets.

[0171] Subsequently, a channel reciprocity verification process is performed to verify the uplink-downlink symmetry of the wireless channel within a short time (within 10ms of the detection time slot). This is achieved by calculating two types of reciprocity error indicators: 1) amplitude reciprocity error, which is the ratio of the absolute difference between the uplink and downlink channel amplitude parameters to their average value; 2) phase reciprocity error, which is the ratio of the difference between the uplink and downlink channel phase parameters (after taking modulo 2π) to π. The two types of error indicators are weighted and summed with a weight of 0.6:0.4 to obtain a comprehensive channel reciprocity error indicator. This indicator directly reflects the symmetry degree of the uplink and downlink channels, and the smaller the value, the better the reciprocity.

[0172] The determination of the pre-set reciprocity error threshold needs to consider the characteristics of the complex channel environment in the earthquake area. The threshold is set to 0.15 (i.e., 15%), which can tolerate slight reciprocity deviations caused by mountainous multipath effects and effectively exclude severe non-reciprocity data caused by terrain shielding and sudden interference. When the calculated channel reciprocity error indicator is less than 0.15, it is determined that the current channel meets the reciprocity requirements, and the effective feature sequence is extracted. If the error indicator exceeds the threshold, the data is marked as invalid, triggering the re-measurement process of the next detection period.

[0173] In the feature sequence extraction stage, for the uplink and downlink channel response data that meet the reciprocity requirements, the channel amplitude fluctuation sequence and the phase jitter sequence are extracted in time sequence (in units of detection time slots). The amplitude fluctuation sequence is calculated by the difference between the channel amplitude parameters in consecutive detection periods, reflecting the random variation characteristics of the channel amplitude. The phase jitter sequence is calculated by the difference between the channel phase parameters in consecutive detection periods (after taking modulo 2π), reflecting the random jitter characteristics of the channel phase. The two sequences are merged point by point in timestamp order to form a channel reciprocity feature sequence that combines amplitude and phase random characteristics. This sequence is generated based entirely on the physical layer characteristics of the wireless channel and has natural randomness and unpredictability.

[0174] Step 4.3, generating an original random bit stream according to the channel reciprocity feature sequence; post-processing the original random bit stream to generate a quantum random number meeting the quantum randomness standard, specifically comprising:

[0175] First, the original random bit stream is generated, and the channel reciprocity feature sequence in the analog domain is converted into a binary bit stream in the digital domain. For the amplitude fluctuation sequence, an adaptive hierarchical quantization strategy is adopted, and the amplitude fluctuation value is divided into 2^n equal probability intervals (n≥1, n=1 in this embodiment, i.e. binary quantization) according to the statistical distribution characteristics of the sequence (determined based on the sequence data of the first 100 detection periods), each interval corresponds to 1 binary number (0 or 1), for example, the fluctuation value less than the mean value is mapped to 0, and the value greater than or equal to the mean value is mapped to 1; for the phase jitter sequence, a similar binary quantization method is adopted, and the phase jitter value (0~2π) is divided into two equal probability intervals, which are mapped to 0 and 1 respectively. The quantized binary numbers of the two sequences are alternately spliced in chronological order to form an original random bit stream. In the splicing process, interleaving processing is used to avoid the influence of the correlation of a single sequence, and the initial randomness of the bit stream is ensured.

[0176] Subsequently, the original random bit stream is post-processed in multiple stages to eliminate potential statistical bias and correlation, and to ensure that the final random number meets the quantum randomness standard (such as the NIST SP800-22 random number test standard). In the first stage, the bias is removed by using the exclusive or equalization method. The original bit stream is grouped by two consecutive bits, if the two groups of bits are the same (00 or 11), they are discarded, if they are different (01 or 10), the previous bit is kept as the valid bit. Through this processing, the 0 / 1 imbalance bias (such as quantization bias caused by channel slow fading) that may exist in the bit stream is eliminated; in the second stage, the correlation is eliminated by using the sliding window de-duplication method, a sliding window with a length of 8 is set, and the bit sequence in the window is de-duplicated, if there are 4 or more consecutive repeated bits, the repeated part is removed, and the non-repeated bits are kept, to eliminate the short-term correlation that may exist in the sequence; in the third stage, the compression processing is performed, and the de-biased and de-correlated bit stream is compressed by using Huffman coding, to keep the core random features and remove redundant information, and to ensure the compactness of the random number.

[0177] Finally, the randomness verification is performed. The post-processed bit stream is input into a preset randomness verification system, which includes uniformity verification, independence verification, run-length verification and other core verification indexes (covering key verification items in NIST SP800-22 standard). If the bit stream passes all verification indexes (the pass rate of each verification is greater than or equal to 99%), it is determined to meet the quantum randomness standard, and is output as the final quantum random number; if the verification fails, the de-biasing processing stage is returned to optimize the parameters (such as adjusting the sliding window length), until the quantum random number meeting the requirements is generated. The generation process is completely based on the physical layer reciprocity characteristics of the wireless channel, without relying on external random sources or complex algorithms, has endogenous security characteristics, and the generation rate is synchronized with the detection period (10 ms / group of random numbers), which can meet the security requirements of real-time communication of the UAV cluster.

[0178] Step 4.4, based on the quantum random number, generating a session key; according to the topology structure and task role of the UAV cluster, combining the session key to generate a device permission topology map, which defines the communication permission level and data access range between devices, specifically including:

[0179] Based on the quantum random number generated in step 4.3, a lightweight and high-security key system and dynamic permission topology are constructed to support the secure cooperative communication of the UAV cluster, and the specific implementation process is as follows:

[0180] The session key generation stage takes quantum random number as the core, and designs key parameters combined with the communication security requirements of the UAV cluster. According to the transmission security requirements of the emergency rescue scene, the length of the session key is set to 256 bits (satisfying the AES-256 encryption standard), and the continuous 256 bits of the quantum random number are intercepted as the basic key material. If the length of the quantum random number is insufficient, the quantum random number is generated multiple times and spliced to supplement it, ensuring that the key length meets the requirements. To enhance the dynamic nature and anti-cracking ability of the key, a one-time periodic key update mechanism is adopted, and the key validity period is synchronized with the scheduling period of the dynamic frame scheduling scheme (1 s in this embodiment). After each scheduling period ends, the session key is regenerated based on the new quantum random number, avoiding the security risks caused by long-term use of the same key. After the key is generated, it is distributed through the stable communication link established in step 1 (beamforming link based on RIS array). In the distribution process, quantum random number encryption transmission is adopted (i.e., the session key of the current period is encrypted using the session key of the previous period), ensuring that the key is not eavesdropped or tampered with during transmission. After receiving the key, each UAV verifies the integrity of the key through the key check code (generated based on the first 32 bits of the quantum random number), and stores it in the secure cache area after the verification is passed, for subsequent data encryption and decryption.

[0181] The device permission topology graph generation stage needs to combine the topology structure of the unmanned aerial vehicle cluster, the task role and the session key association to construct a hierarchical and dynamic permission system. First, the core topology relationship of the unmanned aerial vehicle cluster is determined (taking the command relay machine as the center node, 3 reconnaissance machines and 2 delivery machines as edge nodes, and the RIS array as an auxiliary communication node), the communication permission level is divided based on the topology relationship, and three levels of permissions are provided: the first level of permission (core permission) is the point-to-point communication permission between the command relay machine and each unmanned aerial vehicle, which supports bidirectional transmission of all types of data (reconnaissance data, control instructions, state information, etc.); the second level of permission (cooperative permission) is the communication permission between unmanned aerial vehicles of the same type (such as between reconnaissance machines), which only supports the transmission of cooperative operation related data (such as position synchronization information and regional coverage complementary information); and the third level of permission (basic permission) is the communication permission between the delivery machine and the reconnaissance machine, which only supports one-way transmission of necessary information (such as target position information transmitted from the reconnaissance machine to the delivery machine).

[0182] Based on the session key, a unique access identifier is assigned to each permission level, the first level of permission is associated with a global session key (shared by all devices), the second level of permission is associated with a session key dedicated to devices of the same type (such as a reconnaissance machine dedicated key), and the third level of permission is associated with a one-way access key (such as a key from a reconnaissance machine to a delivery machine), ensuring that data access of different permission levels is strictly controlled. At the same time, the data access range of each device is clearly defined in the topology graph, for example, the command relay machine can access the transmission data of all devices, the reconnaissance machine can only access the related data of itself and the cooperative unmanned aerial vehicle, and the delivery machine can only access the target data of the command relay machine and the corresponding reconnaissance machine, thereby avoiding data leakage.

[0183] The device permission topology graph has dynamic updating capability. When the topology structure of the unmanned aerial vehicle cluster changes (such as when a reconnaissance machine exits the cluster due to insufficient endurance, or a new rescue unmanned aerial vehicle joins), or when the task phase is switched (such as from wide-area reconnaissance to key search and rescue), the session key and permission identifier are regenerated based on the new quantum random number, and the permission level and access range in the topology graph are updated synchronously. The updated topology graph is issued to all devices through the command relay machine, and each device adjusts its communication permission configuration according to the topology graph, ensuring that the cluster can still maintain safe and orderly communication order in a dynamic changing scenario. The topology graph not only realizes fine control of communication permissions, but also ensures the security of the permission system through dynamic association of quantum random numbers.

[0184] In a preferred embodiment of the present application, based on the device permission topology graph and the channel state tensor, the communication capability of the devices in the unmanned aerial vehicle cluster is matched, and a matched device pair set and its communication parameter mapping relationship are generated, including:

[0185] Step 4.5, extract the communication permission level, data access range, and task role identification of each UAV device from the device permission topology map; extract the channel quality index, spatial correlation coefficient, and time delay spread parameter between each pair of UAVs from the channel state tensor, specifically including: extracting device communication management parameters from the device permission topology map, the communication permission level needs to clarify the permission level of each device (first-level core permission, second-level cooperative permission, and third-level basic permission), for example, the command relay machine has first-level permission with all UAVs, the reconnaissance machine has second-level permission with other reconnaissance machines, and the reconnaissance machine has third-level permission with the delivery machine. Meanwhile, record the communication direction corresponding to the permission (one-way / two-way); the data access range needs to clarify the type of data allowed to be transmitted by each device, for example, the command relay machine can access all reconnaissance data, control instructions, and status information, the reconnaissance machine can only access the position synchronization data of cooperative UAVs and its own reconnaissance data, and the delivery machine can only receive the delivery instructions of the command relay machine and the target position data of the reconnaissance machine; the task role identification needs to mark the core function of each device, for example, the command relay machine is responsible for data aggregation and instruction issuance; the reconnaissance machine 1 is responsible for three-dimensional modeling and life detection; and the delivery machine 2 is responsible for precise delivery of first-aid kits, ensuring that subsequent matching meets task requirements.

[0186] Then extract the channel adaptation parameters from the channel state tensor. The channel quality index needs to be based on the time-frequency-space three-dimensional features of the tensor to filter the core parameters reflecting the stability of the link, including real-time signal-to-noise ratio, bit error rate, signal strength, and frequency fading coefficient. The signal-to-noise ratio and bit error rate directly determine the communication reliability, and the frequency fading coefficient reflects the degree of influence of multipath effect; the spatial correlation coefficient needs to be based on the spatial dimension component of the tensor to calculate the spatial correlation degree between any two UAVs. The higher the coefficient, the higher the similarity of the channel environment of the two UAVs, and the more suitable for cooperative communication or spectrum multiplexing; the time delay spread parameter needs to be based on the time dimension component of the tensor to obtain the signal transmission time delay and time delay fluctuation range between each pair of UAVs. This parameter directly affects the matching priority of real-time requirement high services (such as life detection data transmission).

[0187] All extracted parameters need to be structured and associated, with device pairs as the basic index (such as command relay machine-reconnaissance machine 1, reconnaissance machine 1-reconnaissance machine 2, etc.), constructing a multi-dimensional parameter set of device pair-communication permission-task role-channel quality-spatial correlation-time delay spread, ensuring that the management requirements and channel characteristics of each device pair are one-to-one corresponding.

[0188] Step 4.6, according to the communication authority level and the task role identification, construct a device communication demand matrix; according to the channel quality index, the spatial correlation coefficient and the time delay spread parameter, construct a channel matching degree evaluation matrix; the device communication demand matrix and the channel matching degree evaluation matrix are weighted and fused to generate a comprehensive matching score matrix, which specifically includes: first, construct a device communication demand matrix, the core is to convert the device's authority level and task role into quantitative communication demand intensity. The rows and columns of the matrix are the devices (including command relay machines) in the unmanned aerial vehicle cluster, and the elements in the matrix are the communication demand scores (with a value range of 0-10 points) of the corresponding device pairs. The score calculation rule combines the authority level and the task role double dimensions: the device pair with the first level of authority (command machine and each device) and involving the core task (such as the command machine issuing scheduling instructions to the reconnaissance machine, and the reconnaissance machine uploading life detection data to the command machine) has a demand score of 8-10 points; the device pair with the second level of authority (same type device cooperation, such as between reconnaissance machines) and involving cooperative tasks (such as position synchronization and complementary coverage area) has a demand score of 5-7 points; the device pair with the third level of authority (different type device one-way communication, such as reconnaissance machine transmitting target position to delivery machine) and involving basic tasks has a demand score of 3-4 points; the device pair without communication authority or task association (such as between delivery machines) has a demand score of 0 points. After the matrix is constructed, it needs to be dynamically modified according to the current task stage (such as the key search and rescue stage to improve the demand score of the reconnaissance machine and the command machine) to ensure that the demand quantification is consistent with the actual task scenario.

[0189] Secondly, construct a channel matching degree evaluation matrix, the core is to convert the channel quality, spatial correlation, time delay spread and other parameters into quantitative channel adaptation scores (with a value range of 0-10 points). The rows and columns of the matrix are consistent with the device communication demand matrix, and the elements are the channel adaptation scores of the corresponding device pairs. The score calculation rule adopts a weighted sum of multiple parameters: the channel quality weight accounts for 0.5, among which the signal-to-noise ratio and the bit error rate dominate the score, the higher the signal-to-noise ratio and the lower the bit error rate, the higher the score; the spatial correlation coefficient weight accounts for 0.3, the higher the coefficient, the better the signal channel environment cooperation of the device pair, and the higher the adaptation score; the time delay spread parameter weight accounts for 0.2, the smaller the time delay and the narrower the fluctuation range, the higher the score, especially for real-time business, the weight of the time delay parameter can be temporarily increased to 0.3. For example, the signal-to-noise ratio between the command machine and a reconnaissance machine is high, the spatial correlation is strong, and the time delay is small, so the channel adaptation score is 9 points; the channel attenuation between a reconnaissance machine and a delivery machine is serious, and the time delay fluctuation is large, so the channel adaptation score is 3 points.

[0190] Finally, the two matrixes are fused to generate a comprehensive matching score matrix. The setting of the fusion weight needs to consider the task priority and communication reliability. The weight of the device communication demand matrix is set to 0.4 (to ensure that the core task is matched first), and the weight of the channel matching degree evaluation matrix is set to 0.6 (to ensure that the matching result adapts to the channel environment and improves the communication stability). During the fusion calculation, the scores of the corresponding device pairs in the two matrixes are multiplied by their respective weights and then summed to obtain the comprehensive matching score of the device pair (the value range is 0-10 points). For example, the demand score of a certain device pair is 9 points, and the channel adaptation score is 8 points. The comprehensive matching score of the device pair is 9x0.4+8x0.6=8.4 points. After the fusion is completed, the comprehensive matching score matrix is normalized to ensure that the scores of all device pairs are distributed in the range of 0-10 points, which facilitates subsequent pairing screening;

[0191] Step 4.7, based on the comprehensive matching score matrix, pairing the unmanned aerial vehicle devices two by two to obtain a set of successfully matched device pairs, specifically including:

[0192] All device pairs in the comprehensive matching score matrix are sorted by comprehensive score from high to low. The higher the score, the higher the priority of the device pair. For device pairs with the same score, device pairs involving primary permissions (command machines and devices) or core tasks (life detection and precise delivery) are preferred. If there are still parallel situations, device pairs with higher channel adaptation scores are preferred to ensure that core businesses and high-quality channel resources are matched first.

[0193] Subsequently, two-by-two pairing screening is performed. Starting from the highest priority device pair, the pairing feasibility is confirmed in turn. If both devices in the device pair are not occupied by other pairs, and the comprehensive score of the device pair is not lower than the preset pairing threshold (combined with the channel environment in the earthquake area, the threshold is set to 4 points, and the communication reliability of device pairs below this score is difficult to guarantee), the device pair is determined to be successfully matched. If one of the devices is already occupied, the device pair is skipped and the screening continues. If the comprehensive score of a device pair is higher than the threshold but there is a device occupation conflict, and the device pair involves core tasks (such as communication between command machines and reconnaissance machines), the comprehensive score of the paired device is reevaluated. If the score of the paired device is lower than the current device pair, the original pairing can be released to prioritize the matching of core task device pairs.

[0194] The command relay machine needs to be paired with all unmanned aerial vehicles (primary permission forced pairing) to ensure the comprehensiveness of command issuance and data aggregation. The reconnaissance machines need to complete at least one pair of cooperative pairing to ensure complementary regional coverage. The delivery machine needs to be paired with the command machine and the corresponding target area reconnaissance machine to ensure precise reception of the delivery command. After pairing is completed, a set of successfully matched device pairs is formed.

[0195] Step 4.8, for the matched device pair set, the communication parameter mapping is performed, and the corresponding modulation coding scheme, retransmission mechanism parameter and flow control threshold are allocated for each matched device pair; the modulation coding scheme, retransmission mechanism parameter and flow control threshold are bound with the matched device pair, and a communication parameter mapping relationship table containing device identification, matching weight and communication parameter is generated, which specifically includes: communication parameter differentiation allocation is performed, and the core parameters include modulation coding scheme, retransmission mechanism parameter and flow control threshold, and the allocation rule is closely combined with the task type, communication authority and channel quality of the device pair:

[0196] For the device pair (such as command machine-reconnaissance machine) with high comprehensive score (≥7 points), high channel quality (signal-to-noise ratio ≥20 dB) and high bandwidth data transmission (such as high-resolution image, thermal imaging data), a high-order modulation coding scheme (64QAM and Turbo coding) is adopted to improve the transmission rate; for the device pair (such as reconnaissance machine-delivery machine) with medium comprehensive score (4-6 points), general channel quality (10 dB≤signal-to-noise ratio<20 dB) and medium bandwidth data transmission (such as cooperative position information, delivery instruction), a medium-order modulation coding scheme (16QAM and convolutional coding) is adopted to balance the rate and reliability; for the device pair with poor channel quality (signal-to-noise ratio<10 dB) but needing to guarantee core instruction transmission, a low-order modulation coding scheme (QPSK and Hamming coding) is adopted to prioritize communication reliability.

[0197] For the device pair with high real-time requirement (such as the command machine issuing delivery instructions to the delivery machine), a stop-and-wait retransmission mechanism is adopted, the retransmission timeout time is set to 50 ms, and the maximum retransmission number is set to 3 times to avoid excessive retransmission causing time delay accumulation; for the device pair with non-real-time but high data integrity requirement (such as the reconnaissance machine uploading historical reconnaissance data to the command machine), a sliding window retransmission mechanism is adopted, the window size is set to 8, the retransmission timeout time is set to 100 ms, and the maximum retransmission number is set to 5 times to improve data transmission integrity; for the device pair with stable channel quality, the maximum retransmission number can be appropriately reduced to reduce resource occupation.

[0198] For the device pair of command machine and reconnaissance machine, the uplink flow control threshold is set to 100 Mbps (adapted to high-resolution image transmission), and the downlink flow control threshold is set to 20 Mbps (adapted to scheduling instruction transmission); for the cooperative device pair between reconnaissance machines, the flow control threshold is set to 10 Mbps (adapted to position synchronization and other light data); for the device pair of command machine and delivery machine, the downlink flow control threshold is set to 5 Mbps (adapted to delivery instruction), and the uplink flow control threshold is set to 2 Mbps (adapted to delivery state feedback).

[0199] After the parameter allocation is completed, a parameter and device pair binding operation is performed to generate a communication parameter mapping relationship table. The table includes core fields: device pair identification, matching weight (i.e., comprehensive matching score), modulation and coding scheme type, retransmission mechanism parameters (retransmission mode, timeout time, maximum retransmission number), flow control threshold (uplink / downlink), parameter effective time (synchronized with the dynamic frame scheduling scheme). After the mapping relationship table is generated, it is issued to all matching device pairs through the safety channel of the command relay machine. Each device adjusts its own communication module parameters according to the table configuration to ensure the consistency and synergy of the parameters of the paired devices.

[0200] In a preferred embodiment of the present application, step 5, according to the session key, device authority topology graph and matching device pair set, joint encoding of reconnaissance data, control instructions and state information is performed to generate an encoded symbol stream, including:

[0201] Step 5.1, according to the data access range in the device authority topology graph, the reconnaissance data, control instructions and state information are classified and classified to obtain classified data; the classified data is divided into high, medium and low security level data according to the security level, specifically including:

[0202] According to the data access range in the device authority topology graph, the reconnaissance data, control instructions and state information are classified and split; for reconnaissance data, high-resolution orthographic image data, thermal imaging life detection data, and three-dimensional modeling data of collapsed area are split according to the collection subject and purpose; for control instructions, precise positioning instructions, unmanned aerial vehicle flight path adjustment instructions, communication parameter configuration instructions, and cluster coordination scheduling instructions are split according to the instruction priority and execution object; for state information, unmanned aerial vehicle remaining power / attitude information, RIS array reflection state information, and channel quality feedback information are split according to the feedback subject. In the classification process, a unique identification tag is added to each type of data to determine the data generation device, receiving device and transmission direction, ensuring that the data flow and the access range of the authority topology graph are completely matched, and preventing unauthorized data transmission.

[0203] According to the security sensitivity and task priority of the data, the classified data is divided into three security levels. High-density data is defined as the data type that affects the core effectiveness of rescue and the leakage of which will lead to serious consequences, including thermal imaging life detection data, first aid kit accurate delivery instructions, and unmanned aerial vehicle cluster core scheduling instructions. Such data is directly related to life safety and task success, and requires the highest level of security protection. Medium-density data is defined as the data type that supports collaborative work and the leakage of which affects work efficiency, including high-resolution orthographic image data, landslide area three-dimensional modeling data, unmanned aerial vehicle flight path adjustment instructions, and unmanned aerial vehicle attitude / power information. Such data is the basis for cluster collaboration and requires a balance between security and transmission efficiency. Low-density data is defined as the data type that assists in monitoring and has no critical impact on leakage, including RIS array reflection state information, channel quality feedback information, and non-emergency communication parameter query instructions. Such data focuses on transmission efficiency and has relatively low security requirements.

[0204] After the security level division, a security level label is added to each type of data, and is bound with the data identification label to form a structured data list of data identification-generation device-receiving device-security level. The list is synchronized and issued to the command relay machine and all matching device pairs to ensure that the subsequent encryption and coding links can quickly identify the data level, provide the basis for differentiated processing, and at the same time avoid unauthorized devices from accessing high-density data.

[0205] Step 5.2, based on the session key, respectively encrypting high-density data, medium-density data and low-density data to obtain first encrypted data, second encrypted data and third encrypted data, specifically including: extracting the session key and key update mechanism generated in step 4.4, and adapting different encryption strengths and key usage policies according to the data security level. For high-density data, a 256-bit AES encryption algorithm (Advanced Encryption Standard) is used. The session key is updated based on the latest generated quantum random number every time a high-density data frame (such as a single-frame thermal imaging life detection data) is transmitted. The key update frequency is synchronized with the scheduling period of the dynamic frame scheduling scheme (1s / time), ensuring the transmission security of high-sensitive data. For medium-density data, a 128-bit AES encryption algorithm is used, and the key update frequency is 5s / time, balancing security and computational resource consumption, and adapting to the efficient transmission of large-volume data such as high-resolution images. For low-density data, a 64-bit AES encryption algorithm is used, and the key update frequency is 10s / time, maximizing the reduction of the computational load of the unmanned aerial vehicle under the premise of ensuring basic security, and avoiding affecting the flight endurance.

[0206] The encryption execution process is based on a structured data list, and the relay machine is used as the encryption control core to centrally encrypt various data. After receiving the original data uploaded by each UAV, the corresponding encryption algorithm and the current effective session key are called according to the security level label of the data to encrypt the data frame by frame. After encryption, an encryption identifier (including encryption algorithm type, key version number, and encryption timestamp) is added to the encrypted data, and the key version number corresponds to the generation batch of quantum random numbers, so that the receiving end can call the correct key decryption by matching the version number. The first encrypted data (high encryption level), the second encrypted data (medium encryption level), and the third encrypted data (low encryption level) are stored in independent cache areas to avoid confusion of data with different security levels.

[0207] Step 5.3, according to the communication capabilities and task roles of the devices in the matching device pair set, respectively assign forward error correction codes to high-level data, medium-level data, and low-level data to obtain first error correction code, second error correction code, and third error correction code, which specifically includes: extracting the communication capability parameters (including channel quality index, spatial correlation coefficient, and time delay spread parameter) and task role identification of the devices in the matching device pair set, and constructing a data level-communication capability-task role error correction code adaptation decision table. For communication capability evaluation, for device pairs with high channel quality (signal-to-noise ratio ≥ 20 dB, bit error rate < 10 -5 ), high-efficiency error correction codes are preferentially selected; for device pairs with general channel quality (10 dB ≤ signal-to-noise ratio < 20 dB, 10 -5 ≤ bit error rate < 10 -3 ), error correction codes balancing encoding efficiency and error correction capability are selected; for device pairs with poor channel quality (signal-to-noise ratio < 10 dB, bit error rate ≥ 10 -3 ), error correction codes with strong error correction capability are selected.

[0208] The differential error correction code allocation is implemented based on the decision table: for high-density data (first encrypted data), considering the task core nature and data integrity requirement, a high-order forward error correction code with strong error correction capability is allocated, specifically Turbo code (coding rate 1 / 2), which can effectively correct burst errors and random errors in mountainous channel with serious multipath fading and signal blocking, and can ensure lossless transmission of life detection data and accurate delivery instructions; for medium-density data (second encrypted data), considering the transmission efficiency and error correction demand, convolution code (coding rate 3 / 4) is allocated, which has moderate computational complexity and can adapt to the fast transmission of large-volume data such as high-resolution images, while meeting the integrity requirement of collaborative operation data; for low-density data (third encrypted data), taking transmission efficiency as the core, Hamming code (coding rate 7 / 8) with basic error correction capability is allocated, which can quickly process auxiliary monitoring data without occupying too much unmanned aerial vehicle computing resources and communication bandwidth; after the error correction code allocation, the corresponding error correction code parameters (including coding type, coding rate, and code length) are bound for each type of encrypted data, and the communication capability identifier of the matched device pair is associated to ensure that the receiving end can quickly call the corresponding decoding algorithm according to the channel characteristics of the device pair, thereby improving the decoding efficiency.

[0209] Step 5.4, interleave and fuse the first encrypted data, the second encrypted data, and the third encrypted data with the corresponding first error correction code, the second error correction code, and the third error correction code to form a hierarchical encoding data block; perform time sequence weighted sorting on the hierarchical encoding data block and convert it into an encoding symbol stream, which includes data symbols, control symbols, and synchronization symbols, specifically including:

[0210] The first encrypted data and the first error correction code, the second encrypted data and the second error correction code, and the third encrypted data and the third error correction code are processed by interleaving and fusing. The interleaving operation adopts block interleaving, rearranges the continuous encrypted data block according to a preset interleaving matrix, and then combines it with the corresponding error correction code symbol according to the interval mode of data symbols and error correction symbols to form a hierarchical encoding data block. For example, the high-density hierarchical encoding data block is combined according to the mode of 1 thermal imaging data symbol and 1 Turbo error correction symbol, the medium-density data block is combined according to the mode of 2 image data symbols and 1 convolution error correction symbol, and the low-density data block is combined according to the mode of 4 state feedback symbols and 1 Hamming error correction symbol, which ensures that the error correction code can accurately cover the corresponding encrypted data and improve the error recovery capability.

[0211] Then, the three hierarchical coded data blocks are sequentially weighted and sorted, the high-density data block (containing life detection data and accurate delivery instruction) is set to 0.5, the transmission time sequence resource is preferentially allocated, and the core data is preferentially transmitted; the medium-density data block (containing image and modeling data) is set to 0.3, and the high-density data transmission is followed; the low-density data block (containing state feedback and auxiliary information) is set to 0.2, and is in the last position in time sequence. In the sorting process, combined with the time slot configuration of the dynamic frame scheduling scheme, the high-density data block is mapped to the time slot segment with the best channel quality (such as the line-of-sight time slot without obstruction and low interference), and the medium and low-density data blocks are sequentially mapped to the remaining time slots, so that the core data is transmitted under the optimal channel condition.

[0212] Finally, the sorted hierarchical coded data blocks are converted into coded symbol streams, the structure of the symbol stream is designed to adapt to the transmission characteristics of the unmanned aerial vehicle communication module, and three types of core symbols are included: synchronization symbol (accounting for 5%), Zadoff-Chu sequence with good orthogonality is used, which is used for the receiver to realize frame synchronization and time slot alignment, and solves the synchronization deviation problem caused by the time delay fluctuation of the mountainous channel; control symbol (accounting for 10%), including data security level identifier, error correction code parameter, interleaving matrix index and other control information, so that the receiver can quickly analyze the coding rules of the symbol stream; data symbol (accounting for 85%), composed of encrypted data symbol and error correction symbol after interleaving fusion, arranged in sequence according to the hierarchical sorting result. After the symbol stream is generated, the symbol is modulated through the modulation scheme (such as 64QAM, 16QAM, etc.) in the communication parameter mapping relationship table of the matching equipment pair, to form the final baseband signal that can be transmitted.

[0213] In a preferred embodiment of the present application, step 6, according to the communication parameter mapping relationship of the matching equipment pair set, the coded symbol stream is subjected to directional transmission scheduling, comprising:

[0214] Step 6.1, based on the communication parameter mapping relationship table, extracting the modulation and coding scheme, retransmission mechanism parameter and flow control threshold value of each matching equipment pair; configuring symbol mapping rules according to the modulation and coding scheme, setting automatic retransmission request strategy according to the retransmission mechanism parameter, and determining the data buffer size according to the flow control threshold value, specifically including: accurately extracting the core communication parameters of each matching equipment pair from the communication parameter mapping relationship table, including modulation and coding scheme (such as 64QAM and Turbo coding, 16QAM and convolution coding), retransmission mechanism parameter (retransmission mode, timeout time, maximum retransmission number) and flow control threshold value (uplink / downlink bandwidth threshold), in the extraction process, taking the equipment pair identifier as the index, and binding the parameters with the corresponding task roles (such as core data transmission, cooperative operation, auxiliary monitoring) and channel quality indicators (signal-to-noise ratio, bit error rate), so as to ensure the pertinence of parameter configuration.

[0215] The symbol mapping rule is configured according to the modulation and coding scheme. According to the signal constellation characteristics of different modulation types, different symbol mapping logics are formulated for various modulation and coding schemes. For high-density level data transmission equipment pairs (such as command relay machines and reconnaissance machines) using 64QAM, the coded binary data is mapped to the corresponding constellation point symbol in groups of 6 bits, taking into account the balance between transmission rate and anti-interference; for 16QAM modulated medium-density level data equipment pairs, mapping is performed in groups of 4 bits to optimize the transmission efficiency of medium-bandwidth data; for QPSK modulated low-density level data equipment pairs, mapping is performed in groups of 2 bits to prioritize correct demodulation of signals in weak channel environments. The mapping rule synchronously embeds symbol synchronization identifiers to ensure that the receiving end can quickly identify the modulation type and complete symbol demodulation, adapting to the time-varying characteristics of mountainous channels.

[0216] The automatic repeat request strategy is set according to the retransmission mechanism parameters. For real-time task equipment pairs (such as command relay machines and delivery machines) using the stop-and-wait retransmission mechanism, the retransmission timeout time is strictly matched with the time slot length (such as 50ms) of dynamic frame scheduling, and the maximum number of retransmissions is set to 3 to avoid the cumulative impact of retransmission delay on the timeliness of delivery instructions; for non-real-time data equipment pairs (such as reconnaissance machines and command relay machines) using the sliding window retransmission mechanism, the retransmission trigger condition is set according to the window size (such as 8 data blocks), and when the number of unconfirmed received data blocks in the window exceeds 3, batch retransmission is started, the timeout time is set to 100ms, and the maximum number of retransmissions is 5 to ensure the integrity of large-volume data such as high-resolution images; the retransmission priority is set synchronously in the retransmission strategy, and the retransmission request of high-density level data is prioritized over medium and low-density level data to avoid loss of core data due to retransmission queuing.

[0217] The data buffer size is determined according to the flow control threshold. Combined with the uplink / downlink flow threshold of each equipment pair (such as the uplink threshold of 100Mbps for command machines and reconnaissance machines), the ring buffer of the receiving end and the sending end is configured according to the buffer capacity = flow threshold x 1.2 ratio, and 20% of the redundant space is reserved to cope with the sudden flow fluctuations of mountainous channels; for high-density level data transmission equipment pairs, a double-buffering mechanism (separation of receiving buffer and processing buffer) is used to avoid conflicts between data reception and demodulation processing; for low-density level data equipment pairs, a single buffer is configured to save storage resources of unmanned aerial vehicles, and the upper limit of the buffer size is strictly controlled within 10% of the device storage capacity to balance data caching needs and energy consumption. After all configurations are completed, the symbol mapping rule, retransmission strategy parameters, and buffer configuration instructions are issued to the corresponding matching equipment pairs through the control channel to ensure that the devices on both ends are synchronized and take effect.

[0218] Step 6.2, according to the spatial position information of the matched device pair and the spatial dimension component in the channel state tensor, calculate the beamforming weight matrix; based on the beamforming weight matrix, adjust the radiation mode of the phased array antenna on board the unmanned aerial vehicle to form a directional beam, specifically including:

[0219] Extract the real-time spatial position information of the sending end (such as a reconnaissance machine) and the receiving end (such as a command relay machine) in the matched device pair, which is generated by the on-board GPS / IMU combined positioning system, contains three-dimensional coordinates (precision 0.1m), pitch / roll / heading attitude angle (precision 0.01°) and motion velocity vector, and the update frequency is 10Hz to adapt to the dynamic flight state of the unmanned aerial vehicle; From the spatial dimension component of the channel state tensor, filter the core parameters between the sending end and the receiving end, including the spatial attenuation coefficient (reflecting the signal loss degree of line-of-sight / non-line-of-sight link), multipath propagation path identifier (distinguishing direct wave and RIS reflected wave path) and spatial correlation coefficient of adjacent device pairs (judging beam interference overlap risk); Call the reflection state data of the RIS array in the corresponding coverage area, including the unit activation state of each RIS array, the current reflection phase configuration and the link availability label (such as unobstructed-usable, partially obstructed-reduced efficiency).

[0220] All collected data need to be time-stamped to ensure that the sending end position, receiving end position, channel parameters and RIS state data have the same time and space reference at the same time; For the position data of the terrain obstruction area, combine the terrain three-dimensional model in step 1.1 to correct it, for example, when the unmanned aerial vehicle flies into the valley, correct the coordinate deviation caused by GPS signal drift through terrain matching algorithm to ensure that the spatial position parameters are completely matched with the actual communication scene.

[0221] Based on the preprocessed multi-source data, determine the communication link type of the matched device pair and filter the target path for weight matrix calculation. Combine the terrain three-dimensional model and the spatial coordinates of the sending end-receiving end to first judge the line-of-sight accessibility of the direct communication link: if there is no terrain obstruction (such as landslide, mountain) in the range of the line connecting the two, and the spatial attenuation coefficient is less than 80dB (adapted to the signal transmission characteristics of the emergency communication frequency band), then determine the direct link as the main transmission path; If there is line-of-sight obstruction or the direct link attenuation coefficient is greater than 80dB, start the RIS auxiliary link filtering process.

[0222] The RIS-assisted link screening takes the principle of minimum total path loss and highest interference orthogonality, and the specific operation includes: according to the spatial position of the sending end and the receiving end, locking 1-2 RIS arrays covering the area (such as RIS at the entrance of the canyon and RIS on the highland); calculating the total path attenuation of the sending end-RIS array-receiving end, that is, the sum of the attenuation value from the sending end to the RIS and the attenuation value from the RIS to the receiving end; in combination with the reflection efficiency parameter fed back by the RIS array, eliminating invalid paths with total attenuation greater than 100 dB; for the remaining valid paths, the interference risk is evaluated through the spatial correlation coefficient of adjacent device pairs, and the path with a correlation coefficient less than 0.3 with other device pair links is selected as the target reflection link. If there are multiple valid reflection paths, the path with fewer RIS arrays and fewer reflection links is preferred to reduce signal transmission delay.

[0223] With the characteristics of the target transmission path (direct link or RIS-assisted link) as the core, combined with the interference suppression demand, the beamforming weight matrix is calculated in layers. The dimension of the matrix is consistent with the number of elements of the airborne phased array antenna (such as 128 elements of the antenna corresponding to a 128x1 weight matrix), and each matrix element corresponds to the amplitude and phase control parameter of an antenna element.

[0224] The first layer is the basic gain weight calculation, which ensures that the beam main lobe accurately points to the target direction. For the direct link, the direction from the sending end to the receiving end is taken as the beam pointing reference, and the main lobe gain requirement is determined in combination with the spatial attenuation coefficient: for every 10dB increase in attenuation coefficient, the main lobe gain is increased by 3dB, but the maximum gain does not exceed the hardware upper limit of the antenna (such as 30dB); by adjusting the phase difference of each element, the radiation signals of the antenna elements form in-phase superposition in the direction of the receiving end, for example, the left element leads the right element in phase to ensure that the beam main lobe points to the right receiving end. For the RIS-assisted link, the beam pointing reference is divided into two stages: the first stage points to the incident region of the RIS array, and the second stage combines the reflection phase parameters of the RIS array to ensure that the signal reflected by the RIS accurately points to the receiving end through link phase compensation calculation; at the same time, according to the reflection efficiency of the RIS array, the main lobe gain is adjusted, for example, when the RIS reflection efficiency is 80%, the main lobe gain is increased by 5dB compared with the direct link to compensate for the reflection loss.

[0225] The second layer is the optimization of interference suppression weight, which reduces the interference of beam sidelobes to adjacent devices. The spatial position and channel quality parameters of the adjacent device pair in step 4.5 are extracted to determine the interference target (such as another reconnaissance aircraft) that may be affected by the current beam; the azimuth and elevation angle range of the interference target is marked by the spatial correlation coefficient; based on the basic gain weight, the amplitude of the antenna elements pointing to the direction of the interference target is adjusted, for example, the amplitude of the corresponding elements is reduced by 50%, so that the gain of the beam sidelobes in the direction of the interference target is reduced by more than 15 dB, ensuring that the interference power of the sidelobes to the interference target does not exceed -90 dBm; at the same time, the element parameters in the main lobe direction are kept unchanged to avoid affecting the signal strength of the core communication link.

[0226] The third layer is the power constraint weight correction, which ensures that the weight matrix meets the device power control requirements. The transmit power level of the sending end in step 3.1 (such as 1 level 10 dBm, 3 level 20 dBm) is extracted, and the total power amplitude of the weight matrix is matched and calibrated with the transmit power level; if the calculated total power amplitude exceeds the threshold corresponding to the current level, the amplitude parameters of all elements are attenuated in proportion to ensure that the total power does not exceed the limit; for matching device pairs for high-density data transmission (such as reconnaissance aircraft-command relay aircraft), within the power allowed range, the main lobe gain is prioritized, and only the sidelobe element parameters are fine-tuned to balance communication quality and power consumption.

[0227] According to the dynamic changes of unmanned aerial vehicle flight and channel environment, the beamforming weight matrix is adjusted in real time. The attitude sensor of the on-board IMU monitors the attitude changes of the sending end unmanned aerial vehicle in real time, and if the pitch angle or heading angle changes more than 0.5°, the phase parameters of the weight matrix are corrected according to the attitude change, for example, when the heading angle of the unmanned aerial vehicle deflects 3° to the right, the phases of all elements are adjusted synchronously to make the beam main lobe deflect 3° to the right, offsetting the beam shift caused by attitude changes; combined with the real-time update data of the channel state tensor (updated every 10 ms), if the spatial attenuation coefficient of the target link suddenly increases, the amplitude of the element corresponding to the main lobe gain is immediately increased to ensure the stability of the signal strength at the receiving end.

[0228] The beamforming weight matrix is analyzed, and each element in the matrix corresponds to an element of the phased array antenna, and each element contains two groups of core parameters: amplitude coefficient and phase coefficient. During the analysis process, the amplitude coefficient is standardized as a power control ratio (range 0-1); the phase coefficient is standardized as an identifiable angle value (range 0-360°). After the analysis is completed, the structured parameter table of element number-amplitude ratio-phase angle is generated according to the physical arrangement order of the antenna elements (such as linear arrangement or surface array arrangement), avoiding the confusion of element parameter mapping.

[0229] Based on the structured parameter table, the digital control signals of each array element are generated. Among them, the phase control signal adopts 16-bit binary coding, directly mapping the phase angle of 0-360° (each 1 bit corresponds to 0.014°, meeting the 0.1° accuracy requirement); the amplitude control signal adopts 8-bit binary coding, mapping the power ratio of 0-1 (each 1 bit corresponds to a proportional accuracy of 0.0039). The control signal is transmitted through a high-speed serial peripheral interface (SPI), and the transmission rate is set to 10 Mbps, ensuring that the control signals of all array elements are transmitted within 1 ms, avoiding the deviation of array elements due to transmission delay. During transmission, the timestamp of the Beidou timing system is used as the synchronization reference, and a synchronization identification bit is added to the control signal frame header to ensure that all array element control signals are received and executed synchronously, with a deviation of not more than 10 μs.

[0230] After receiving the digital control signal, the phase difference of the array element output signal is adjusted according to the coding value of the phase control signal; the amplification factor is adjusted according to the coding value of the amplitude control signal, so that the array element output power meets the standardized power ratio, for example, an amplitude instruction of 0.6 corresponds to an amplification factor of 60% of the reference value, ensuring that the beam main lobe gain meets the design requirements (such as 25 dB). During hardware driving, the parameter adjustment response time of each array element is not more than 50 μs, ensuring the rapid formation of the beam and adapting to the time slot requirements of dynamic frame scheduling.

[0231] After beam generation, the physical characteristic parameters of the beam are collected in real time, including the main lobe pointing angle, beam width, main lobe gain and sidelobe suppression ratio. By receiving the preset pilot signal (Zadoff-Chu sequence consistent with step 4.1), the actual pointing of the beam main lobe is calculated and compared with the designed pointing (such as the receiving end position or the incident direction of the RIS array). If the deviation exceeds 0.5°, the phase parameters of the corresponding array elements are fine-tuned. This conversion process does not rely on external control nodes or public network support, and the parameter adjustment accuracy and hardware response speed adapt to the dynamic communication requirements in the public network environment, ensuring that the directional beam can quickly adapt to link changes and maintain a stable signal transmission path.

[0232] Step 6.3, the coded symbol stream is divided according to the matching device pair set to generate a corresponding sub-symbol stream for each matching device pair, and each sub-symbol stream is mapped to a physical resource block according to the symbol mapping rule to form a scheduling transmission unit, specifically including: based on the communication parameters (bandwidth requirement, flow control threshold, modulation and coding scheme) of the matching device pair set, the coded symbol stream generated in step 5.4 is processed by segmentation. The segmentation logic takes the exclusive needs of the device pair as the core: the symbol stream length is allocated according to the bandwidth threshold of each device pair, for example, the device pair bandwidth of the command relay machine and the reconnaissance machine is 20MHz, and the allocated sub-symbol stream length accounts for 40% of the total symbol stream (adapted to the high-resolution image transmission requirement); the device pair bandwidth of the delivery machine and the command machine is 5MHz, and the allocated sub-symbol stream length accounts for 10% of the total symbol stream (adapted to the control instruction transmission). Unique identification information is added to each sub-symbol stream during the segmentation process, including the sending end device ID, the receiving end device ID, the data security level, the symbol stream length and the check code, so as to ensure that the receiving end can accurately identify and receive the target data and avoid data confusion between multiple device pairs.

[0233] The segmented sub-symbol stream needs to be preprocessed and optimized: for high-density data sub-symbol streams, an additional frame synchronization identifier (16-bit fixed sequence) is added to improve the synchronization speed and anti-interference ability of the receiving end; for large-volume medium-density data sub-symbol streams, block processing is performed according to the size of the physical resource block (such as containing 1024 symbols per block), which facilitates subsequent mapping and retransmission management; for low-density data sub-symbol streams, the preprocessing process is simplified, only the core identification information is retained, and the processing efficiency is improved. The preprocessed sub-symbol stream is stored in the exclusive cache area of the corresponding device pair and waits for resource mapping.

[0234] Then, according to the symbol mapping rule and the time-frequency resource planning of the dynamic frame scheduling scheme, each sub-symbol stream is mapped to a physical resource block (PRB). The division of the physical resource block is based on the carrier frequency allocation table of step 2.4 and the time-frequency two-dimensional overlapping frame structure of step 3.2, each resource block corresponds to a fixed time-frequency resource unit (such as 1msx1MHz), and the channel quality level (good / normal / poor) of the resource block is marked. The mapping logic follows the principle of high-quality resource adapting to high-priority data: the sub-symbol stream of high-density data is preferentially mapped to the resource block with high channel quality (SNR≥20dB), and the resource block is allocated continuously to reduce switching loss; the sub-symbol stream of medium-density data is mapped to the resource block with good channel quality (10dB≤SNR<20dB), and non-continuous allocation can be used to improve resource utilization; the sub-symbol stream of low-density data is mapped to the remaining resource block to adapt to its low transmission requirement.

[0235] The time-frequency resource occupation table of the dynamic frame scheduling scheme is used to monitor the allocation state of each resource block in real time, to ensure that the same resource block is only mapped to the sub-symbol stream of one device pair in the same time slot; if the sub-symbol stream of a certain device pair encounters resource conflict during mapping, the resource block allocation of the low-priority device pair is adjusted first to ensure the transmission demand of the high-priority device pair; after the mapping is completed, a mapping relationship table of sub-symbol stream identifier-physical resource block number-time-frequency coordinate-channel quality level is generated and sent to the sending end device and the command relay machine, which is used for subsequent transmission scheduling and link monitoring. Each mapped resource block combination forms an independent scheduling transmission unit, which contains complete sub-symbol stream blocks, identification information and check codes, and has independent transmission and demodulation capabilities.

[0236] Step 6.4, based on the frame scheduling time sequence table in the dynamic frame scheduling scheme, allocate transmission time slots for each scheduling transmission unit; in combination with the directional beam, send the scheduling transmission unit in the allocated transmission time slot to realize directional transmission scheduling, which specifically includes: extracting the frame scheduling time sequence table in the dynamic frame scheduling scheme, which clearly shows the time slot allocation priority, available time slot segment and time slot length of each device pair (based on the decision tree optimization result of step 3.4). The time slot allocation takes "priority classification and time sequence staggered" as the core principle: the first priority device pair (high-density data transmission of the command machine and the reconnaissance machine) is allocated to the core time slot segment (such as the first 30% time slots in the frame period) with the best channel quality and the least interference, and the time slot length is adapted to the size of the sub-symbol stream block (such as 500ms time slot for large volume image data); the second priority device pair (cooperative data transmission between reconnaissance machines) is allocated to the secondary core time slot segment (the middle 40% time slots in the frame period), and the time slot length is adjusted as needed; the third priority device pair (low-density data transmission of the delivery machine and the command machine) is allocated to the edge time slot segment (the last 30% time slots in the frame period), to ensure that the core task occupies high-quality time sequence resources first.

[0237] When allocating specific transmission time slots for each scheduling transmission unit, the time-frequency coordinates of the physical resource blocks need to be combined for synchronous planning: to ensure that the time slot allocation of the scheduling transmission unit is consistent with the time dimension of the corresponding physical resource block, to avoid time-frequency resource mismatch; at the same time, based on the spatial position and beam pointing direction of the matching device pair, to avoid the beam coverage time slots of other device pairs, to reduce the cross interference between directional beams. For example, the cooperative transmission time slots of two adjacent reconnaissance machines need to be staggered by more than 50ms, and the beam pointing angle is greater than 15°, to ensure that they do not interfere with each other; after the time slot allocation is completed, a transmission scheduling table of "scheduling transmission unit-transmission time slot-physical resource block-beam parameter" is generated and sent to each sending end device, to update the global scheduling state of the command relay machine synchronously.

[0238] The transmission execution stage is centered on a directional beam, and data transmission is started in combination with a transmission schedule table: the sending end device activates the directional beam in a preset time slot according to the starting time of the transmission time slot (based on Beidou timing, with an accuracy of 10 mu s), and loads the symbol stream of the scheduled transmission unit to the corresponding time-frequency resource according to the physical resource block mapping relationship; during the transmission process, the updated data of the channel state tensor is monitored in real time, and if the channel quality suddenly deteriorates (such as a sharp drop in SNR), the fast retransmission strategy in the retransmission mechanism parameters is immediately started to complete retransmission in the current time slot or the next standby time slot, avoiding data loss; for a link involving an RIS array, the reflection state of the RIS array is monitored synchronously during the transmission process to ensure the stability of the beam reflection path.

[0239] The receiving end device adjusts the beam pointing direction (in cooperation with the sending end beam) based on the preset information of the transmission schedule table, receives the scheduled transmission unit in the corresponding time slot, filters the target data through the sub-symbol stream identifier, demodulates and decodes in combination with the symbol mapping rule, and finally restores the original data; after receiving, a receiving confirmation signal (including the receiving state and the bit error rate) is fed back to the sending end, and the sending end updates the buffer area state according to the confirmation signal, and the relay machine synchronously updates the global transmission state to form a closed-loop scheduling mechanism of “allocation-transmission-feedback-adjustment”. The whole process does not need to rely on public network infrastructure, and through self-organized time slot allocation and directional beam cooperation, efficient, low-interference parallel transmission of multiple devices in complex mountainous environments is realized, and the continuous development of emergency rescue tasks is ensured.

[0240] As shown in Figure 2 , the embodiment of the present application also proposes a public network-free communication area unmanned aerial vehicle near field communication matching processing system, comprising:

[0241] The control module is used for adjusting and controlling the near field electromagnetic wave propagation environment to obtain three-dimensional channel characteristic data; the three-dimensional channel characteristic data is collected to generate a channel state tensor;

[0242] The optimization module is used for constructing a quantum annealing optimization model based on the channel state tensor; and solving a frequency spectrum resource allocation problem by simulating a quantum tunneling effect according to the quantum annealing optimization model to obtain a carrier frequency, a time slot offset and a power control parameter set of each unmanned aerial vehicle;

[0243] The construction module is used for constructing an overlapping communication frame structure according to the carrier frequency, the time slot offset and the power control parameter set; and generating a dynamic frame scheduling scheme according to the overlapping communication frame structure;

[0244] The matching module is used for extracting a channel reciprocity feature sequence based on the dynamic frame scheduling scheme; generating quantum random numbers according to the channel reciprocity feature sequence; generating a session key and a device permission topology graph according to the quantum random numbers; and performing communication capability matching on the devices in the unmanned aerial vehicle cluster based on the device permission topology graph and the channel state tensor to generate a matched device pair set and a communication parameter mapping relationship thereof.

[0245] The processing module is used for jointly encoding the reconnaissance data, the control instruction and the state information according to the session key, the device permission topology graph and the matched device pair set to generate an encoded symbol stream; and performing directional transmission scheduling on the encoded symbol stream according to the communication parameter mapping relationship of the matched device pair set to complete the near-field communication matching processing between the unmanned aerial vehicles.

[0246] The above is the preferred embodiment of the present application. It should be noted that, for those skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, which should also be considered within the scope of protection of the present application.

Claims

1. A method for matching a UAV with a near field communication in a public network communication area, characterized in that, The method comprises: The method comprises: The method comprises: Based on the channel state tensor, a quantum annealing optimization model is constructed; according to the quantum annealing optimization model, the frequency spectrum resource allocation problem is solved by simulating the quantum tunneling effect, so as to obtain the carrier frequency, time slot offset and power control parameter set of each unmanned aerial vehicle, comprising: based on the spatial correlation and time-frequency characteristics in the channel state tensor, a target function of frequency spectrum resource allocation is constructed; according to the communication demand and topology structure of the unmanned aerial vehicle cluster, a constraint condition set of frequency spectrum resource allocation is determined; the target function and constraint condition set are converted into the expression form of Ising model to construct the Hamiltonian of the quantum annealing optimization model; according to the ground state energy distribution characteristics of the Hamiltonian, the spin state and coupling strength parameters of the quantum bit are initialized; by simulating the quantum tunneling effect, the transverse magnetic field strength is dynamically adjusted in the quantum annealing process, and when the energy converges to a stable threshold, the final spin state configuration of the quantum bit is recorded; according to the final spin state configuration of the quantum bit, the corresponding frequency spectrum resource allocation scheme of each unmanned aerial vehicle is decoded; the frequency spectrum resource allocation scheme is mapped into a specific carrier frequency allocation table, a time slot offset configuration table and a power control parameter set to form a complete frequency spectrum resource scheduling instruction set; According to the carrier frequency, time slot offset and power control parameter set, an overlapping communication frame structure is constructed; according to the overlapping communication frame structure, a dynamic frame scheduling scheme is generated; Based on the dynamic frame scheduling scheme, a channel reciprocity feature sequence is extracted; according to the channel reciprocity feature sequence, a quantum random number is generated; according to the quantum random number, a session key and a device authority topology graph are generated; based on the device authority topology graph and the channel state tensor, the devices in the unmanned aerial vehicle cluster are matched in communication capability to generate a matched device pair set and a communication parameter mapping relationship thereof; According to the session key, the device permission topology graph and the matched device pair set, reconnaissance data, control instructions and state information are jointly encoded to generate an encoded symbol stream; According to the communication parameter mapping relationship of the matched device pair set, the encoded symbol stream is directionally transmitted and scheduled to complete the near-field communication matching processing between unmanned aerial vehicles. 2.The method of claim 1, wherein, Based on the spatial correlation and time-frequency characteristics in the channel state tensor, a target function for spectrum resource allocation is constructed, including: Separate the spatial dimension component, the time dimension component and the frequency dimension component from the channel state tensor; calculate the spatial isolation matrix between unmanned aerial vehicle nodes based on the spatial dimension component, extract the channel coherence time sequence based on the time dimension component, and obtain the frequency domain flatness index based on the frequency dimension component; According to the spatial isolation matrix, the normalized spatial orthogonality coefficient between each pair of unmanned aerial vehicles is calculated; according to the channel coherence time sequence, the normalized time window length is determined; according to the frequency domain flatness index, the normalized subcarrier aggregation area is divided; The normalized spatial orthogonality coefficient is mapped to the spatial multiplexing gain index, the normalized time window length is mapped to the time domain scheduling efficiency index, and the normalized subcarrier aggregation area is mapped to the spectrum utilization index; The spatial multiplexing gain index, the time domain scheduling efficiency index and the spectrum utilization index are multiplied by the corresponding preset weight coefficients and summed to construct a basic target function; interference suppression constraint terms and power consumption constraint terms are set in the basic target function to form the target function for spectrum resource allocation. 3.The method of claim 2, wherein, According to the carrier frequency, the time slot offset and the power control parameter set, an overlapping communication frame structure is constructed; According to the overlapping communication frame structure, a dynamic frame scheduling scheme is generated, including: Based on the carrier frequency allocation table, an independent subcarrier group is allocated to each unmanned aerial vehicle to obtain a subcarrier group allocation result; based on the time slot offset configuration table, a frame start time offset is set for each unmanned aerial vehicle; based on the power control parameter set, a corresponding transmission power level is configured for each unmanned aerial vehicle; According to the subcarrier group allocation result and the frame start time offset, a time-frequency two-dimensional overlapping frame structure is constructed; in the time-frequency two-dimensional overlapping frame structure, a control channel area, a data channel area and a pilot channel area are set; according to the transmission power level, a power allocation ratio is configured for different areas; Based on the time-frequency two-dimensional overlapping frame structure, the inter-frame interference mode and the channel occupation situation are analyzed to obtain the channel occupation situation and the interference mode; according to the channel occupation situation and the interference mode, frame scheduling conflict detection rules and priority determination criteria are determined; According to the frame scheduling conflict detection rules and the priority determination criteria, a frame scheduling decision tree is constructed; based on the frame scheduling decision tree, combined with real-time channel state feedback information, an adaptive dynamic frame scheduling scheme is generated, including a frame scheduling time sequence table, a power adjustment instruction and a conflict avoidance strategy. 4.The method of claim 3, wherein, Based on the dynamic frame scheduling scheme, a channel reciprocity feature sequence is extracted; According to the channel reciprocity feature sequence, a quantum random number is generated; According to the quantum random number, a session key and a device permission topology graph are generated, including: Based on the frame scheduling timing table in the dynamic frame scheduling scheme, a bidirectional channel sounding time slot is configured; in the bidirectional channel sounding time slot, reciprocal channel measurement is performed between the UAV pairs to obtain uplink channel response data and downlink channel response data; The uplink channel response data and the downlink channel response data are subjected to reciprocity checking processing, and a channel reciprocity error index is calculated; when the channel reciprocity error index is less than a preset threshold, a channel amplitude fluctuation sequence and a phase jitter sequence are extracted, and a channel reciprocity feature sequence is generated by merging; According to the channel reciprocity feature sequence, an original random bit stream is generated; the original random bit stream is post-processed to generate a quantum random number meeting the quantum randomness standard; Based on the quantum random number, a session key is generated; according to the topology structure and task role of the UAV cluster, a device permission topology graph is generated in combination with the session key, which defines the communication permission level and data access range between devices. 5.The method of claim 4, wherein, Based on the device permission topology graph and the channel state tensor, the communication capability of the devices in the UAV cluster is matched to generate a matched device pair set and a communication parameter mapping relationship, including: The communication permission level, data access range and task role identifier of each UAV device are extracted from the device permission topology graph; the channel quality index, spatial correlation coefficient and delay spread parameter between each UAV pair are extracted from the channel state tensor; According to the communication permission level and the task role identifier, a device communication demand matrix is constructed; according to the channel quality index, the spatial correlation coefficient and the delay spread parameter, a channel matching degree evaluation matrix is constructed; the device communication demand matrix and the channel matching degree evaluation matrix are weighted and fused to generate a comprehensive matching score matrix; Based on the comprehensive matching score matrix, the UAV devices are paired two by two to obtain a matched device pair set; The matched device pair set is subjected to communication parameter mapping, and each matched device pair is assigned a corresponding modulation and coding scheme, retransmission mechanism parameter and flow control threshold; the modulation and coding scheme, retransmission mechanism parameter and flow control threshold are bound with the matched device pair to generate a communication parameter mapping relationship table containing device identifier, matching weight and communication parameter. 6.The method of claim 5, wherein, According to the session key, the device permission topology graph and the matched device pair set, the reconnaissance data, control instructions and state information are jointly encoded to generate an encoded symbol stream, including: According to the data access range in the device permission topology graph, the reconnaissance data, control instructions and state information are classified by level to obtain classified data; the classified data is divided into high, medium and low security level data according to the security level; Based on the session key, the high, medium and low security level data are respectively encrypted to obtain first, second and third encrypted data; According to the communication capability and task role of the devices in the matched device pair set, forward error correction codes are respectively assigned to the high, medium and low security level data to obtain first, second and third error correction codes; According to the communication capability and task role of the devices in the matched device pair set, forward error correction codes are respectively assigned to the high, medium and low security level data to obtain first, second and third error correction codes; The first, second and third encrypted data are interleaved and fused with the corresponding first, second and third error correction codes to form a layered encoding data block; the layered encoding data block is sequentially weighted and sorted, and is converted into an encoding symbol stream, which contains data symbols, control symbols and synchronization symbols. 7.The method of claim 6, wherein, According to the communication parameter mapping relationship of the matched device pair set, the encoding symbol stream is subjected to directional transmission scheduling, which includes: Based on the communication parameter mapping relationship table, the modulation and coding scheme, the retransmission mechanism parameters and the flow control threshold of each matched device pair are extracted; the symbol mapping rule is configured according to the modulation and coding scheme, the automatic repeat request strategy is set according to the retransmission mechanism parameters, and the data buffer size is determined according to the flow control threshold; According to the spatial position information of the matched device pair and the spatial dimension component in the channel state tensor, a beamforming weight matrix is calculated; based on the beamforming weight matrix, the radiation mode of the unmanned aerial vehicle onboard phased array antenna is adjusted to form a directional beam; The encoding symbol stream is segmented according to the matched device pair set to generate a corresponding sub-symbol stream for each matched device pair; according to the symbol mapping rule, each sub-symbol stream is mapped to a physical resource block to form a scheduling transmission unit; Based on the frame scheduling time sequence table in the dynamic frame scheduling scheme, each scheduling transmission unit is allocated a transmission time slot; in combination with the directional beam, the scheduling transmission unit is sent in the allocated transmission time slot to realize directional transmission scheduling.

8. A system for matching unmanned aerial vehicles with near field communication in a public network communication zone, the system implementing the method of any one of claims 1 to 7, characterized in that, It includes: The regulation and control module is used for regulating and controlling the near-field electromagnetic wave propagation environment to obtain three-dimensional channel characteristic data; Collecting three-dimensional channel characteristic data to generate a channel state tensor; The optimization module is used for constructing a quantum annealing optimization model based on the channel state tensor; according to the quantum annealing optimization model, the frequency spectrum resource allocation problem is solved by simulating the quantum tunneling effect to obtain the carrier frequency, time slot offset and power control parameter set of each unmanned aerial vehicle; The construction module is used for constructing an overlapping communication frame structure according to the carrier frequency, time slot offset and power control parameter set; and generating a dynamic frame scheduling scheme according to the overlapping communication frame structure; The matching module is used for extracting a channel reciprocity characteristic sequence based on the dynamic frame scheduling scheme; According to the channel reciprocity characteristic sequence, a quantum random number is generated; According to the quantum random number, a session key and a device authority topology graph are generated; based on the device authority topology graph and the channel state tensor, the devices in the unmanned aerial vehicle cluster are subjected to communication capability matching to generate a matched device pair set and its communication parameter mapping relationship; The processing module is used for jointly encoding the reconnaissance data, control instructions and state information according to the session key, device authority topology graph and matched device pair set to generate an encoding symbol stream; and performing directional transmission scheduling on the encoding symbol stream according to the communication parameter mapping relationship of the matched device pair set to complete the near-field communication matching processing between unmanned aerial vehicles.

Images