A dynamic cooperative network communication method and system for low-altitude three-dimensional coverage

Abstract

The application relates to a dynamic cooperative network communication method and system for low-altitude stereoscopic coverage, wherein the method fundamentally breaks through the limitations of the prior art in low-altitude stereoscopic coverage by constructing an intelligent closed-loop process of "perception-prediction-decision-execution", and realizes a mode change from "passive response" to "active planning". Through three-dimensional trajectory prediction and dynamic clustering, the moving trend of a user group can be grasped in advance, and a basis for resource pre-allocation is provided; through stereoscopic interference graph construction and perception, traditional local and static interference coordination is improved to global and dynamic intelligent interference management; through cooperative scheduling processing, precise regulation and control of deep integration are realized in the spatial domain, the cooperative domain and the resource domain, and the system has self-evolution ability and can adapt to complex and changeable low-altitude environments; through execution mobility management, the problem of service interruption caused by high-speed movement is effectively solved, and the problem of how to improve communication quality in related technologies is solved.

Description

Technical Field

[0001] This application relates to the field of wireless communication technology, and in particular to a dynamic cooperative network communication method and system for low-altitude three-dimensional coverage. Background Technology

[0002] With the development of the low-altitude economy, low-altitude equipment such as drones and aerial logistics have placed high demands on communication networks for high mobility, high reliability, and low latency. Traditional ground base station antennas with down-tilted beams are unable to effectively cover the low-altitude airspace, resulting in large signal fluctuations, frequent handovers, and severe co-channel interference.

[0003] While existing technologies such as Cooperative Multipoint Transmission (CoMP), Carrier Aggregation (CA), and Dynamic Reference Signals (DRS) can partially alleviate the problem, they still have the following limitations: First, traditional CoMP relies on ideal backhaul conditions, making it difficult to achieve low-latency coordination under non-ideal transmission latency across base stations; second, DRS lacks the ability to predict and dynamically cluster low-altitude three-dimensional motion trajectories, making it difficult to meet low-latency requirements; third, it lacks interference modeling and multi-dimensional coordination mechanisms for low-altitude-ground three-dimensional space, making it difficult to cope with the complex interference topology brought about by the altitude dimension; fourth, existing resource scheduling and handover management do not fully consider the altitude dimension of low-altitude users, resulting in insufficient low-altitude airspace signal coverage and network performance that cannot meet the needs of low-altitude services.

[0004] Therefore, no effective solution has yet been proposed for improving communication quality in dynamic cooperative network communication methods for low-altitude three-dimensional coverage. Summary of the Invention

[0005] This application provides a dynamic collaborative network communication method and system for low-altitude three-dimensional coverage, which at least solves the problem of how to improve communication quality in related technologies.

[0006] In a first aspect, embodiments of this application provide a dynamic cooperative network communication method for low-altitude three-dimensional coverage, the method comprising: Multidimensional feature data of low-altitude users are obtained, and three-dimensional trajectory prediction of the low-altitude users is performed based on the multidimensional feature data to obtain predicted trajectory data. Clustering processing is performed based on the predicted trajectory data to obtain clustering results. A three-dimensional interference map including base station nodes and low-altitude user nodes is constructed, and the three-dimensional interference map is processed by a graph neural network to generate an interference coordination strategy. Based on the predicted trajectory data, the clustering results, and the interference coordination strategy, collaborative scheduling processing is performed to obtain network scheduling parameters. Mobility management is performed on the low-altitude users based on the predicted trajectory and the network scheduling parameters.

[0007] In one embodiment, the step of acquiring multidimensional feature data of low-altitude users and performing three-dimensional trajectory prediction on the low-altitude users based on the multidimensional feature data to obtain predicted trajectory data includes: Acquire multidimensional feature data for each low-altitude user. The multidimensional feature data includes real-time status data, historical trajectory data, and context data. The real-time status data includes three-dimensional coordinates, instantaneous velocity vector, and heading angle. The context data includes mission type, preset waypoint, and airspace restrictions. The multidimensional feature data is input into the prediction model, and the three-dimensional trajectory probability distribution of the low-altitude user in the future time period is output as the predicted trajectory data.

[0008] In one embodiment, the prediction model includes an encoder, a social pooling layer, and a decoder; wherein, The encoder extracts the temporal dependencies in the multidimensional feature data to generate hidden states. Through the social pooling layer, based on the hidden state and the graph structure relationship between the low-altitude user and its spatial neighboring users, the influence weight of each neighboring user on the future trajectory of the low-altitude user is calculated using an attention mechanism to obtain the social pooling feature. A graph attention network is introduced into the social pooling layer to construct a graph structure for each low-altitude user and spatially neighboring users. The influence weight of neighboring users on future trajectories is quantified through the attention mechanism to obtain social pooling features. Based on the hidden state and the social pooling feature, the decoder recursively outputs predicted trajectory data.

[0009] In one embodiment, the clustering process based on the predicted trajectory data to obtain clustering results includes: Based on the predicted trajectory data, clustering features are extracted for each low-altitude user. The clustering features include the centroid of the future location sequence, the average vector of the direction of movement, the trajectory shape features, and the relative geometric relationship with neighboring base stations. Based on the clustering features, the trajectory similarity between the two low-altitude users is calculated using a dynamic time warping algorithm to generate a trajectory similarity matrix. Based on the similarity matrix, a spectral clustering algorithm is used to cluster low-altitude users to obtain cluster division results; wherein, the spectral clustering algorithm maps the data to a low-dimensional space through Laplacian matrix eigenvalue decomposition, and then performs clustering through K-means; Based on the preset period, the confidence level of the predicted trajectory, and the location status of low-altitude users, the cluster division results are dynamically managed to obtain clustering results. The dynamic cluster management includes merging, splitting, and disbanding clusters.

[0010] In one embodiment, a three-dimensional interference map including base station nodes and low-altitude user nodes is constructed. The three-dimensional interference map is then processed using a graph neural network to generate an interference coordination strategy, including: Base station nodes and low-altitude user nodes are respectively used as graph nodes. Each graph node includes corresponding node features. Service edges and interference edges are constructed according to the service relationship and interference relationship between nodes. The interference edge weight is calculated according to the SINR measurement value. Based on the graph nodes, the service edges, the interference edges and the interference edge weight, a three-dimensional interference map is obtained. The stereo interference image is input into a graph neural network, and the features of each node in the stereo interference image are aggregated and updated through the message passing mechanism of the graph neural network to obtain the embedding vector of each graph node. Based on the embedded vector, link performance is predicted or interference hotspots are classified, and interference coordination strategies are generated based on the prediction or classification results.

[0011] In one embodiment, the step of performing cooperative scheduling processing based on the predicted trajectory data, the clustering results, and the interference coordination strategy to obtain network scheduling parameters includes: Based on the predicted trajectory data and the clustering results, the average altitude distribution and future spatial distribution trend of low-altitude users are determined. The comprehensive coverage rate of the low-altitude coverage area is used as the optimization target, and the beam tilt angle, azimuth angle and beamwidth are used as optimization variables. Combined with the beam interference avoidance requirements in the interference coordination strategy, the optimal beam configuration is solved by genetic algorithm or particle swarm algorithm, and the optimal beam configuration is used as the beam configuration parameter. Based on the predicted trajectory data, the clustering results, and the interference coordination strategy, cross-station asynchronous cooperative processing is performed to obtain the cross-station cooperative transmission strategy. Based on the beam configuration parameters, the cross-site cooperative transmission strategy, the clustering results, and the interference coordination strategy, reinforcement learning resource allocation processing is performed to obtain the network scheduling parameters.

[0012] In one embodiment, the step of performing cross-site asynchronous cooperative processing based on the predicted trajectory data, the clustering results, and the interference coordination strategy to obtain a cross-site cooperative transmission strategy includes: Based on the clustering results, the set of cooperating base stations corresponding to low-altitude users in the same cluster is determined. Based on the predicted trajectory data, the cross-site movement trend of low-altitude users is predicted. Combined with the joint transmission interference optimization requirements in the interference coordination strategy, the primary serving base station and cooperating base stations are determined. The primary serving base station schedules low-altitude user data at the current moment and presets the transmission time to be the sum of the current moment and the maximum delay of the cooperating base stations, wherein the maximum delay of the cooperating base stations is the maximum backhaul delay from the primary serving base station to all the cooperating base stations; The primary serving base station sends the low-altitude user data and resource allocation information to all cooperating base stations through the base station interface to complete the data pre-distribution. The primary serving base station instructs the low-altitude user via radio resource control signaling to mix the inactive secondary carrier with the automatic repeat request process to receive data arriving at the transmission time on the primary carrier. At the transmission time, the primary serving base station and all cooperating base stations use the same time and frequency resources to jointly transmit the same user data to the low-altitude user, thus obtaining a cross-site cooperative transmission strategy.

[0013] In one embodiment, the step of performing reinforcement learning resource allocation processing based on the beam configuration parameters, the cross-site cooperative transmission strategy, the clustering results, and the interference coordination strategy to obtain the network scheduling parameters includes: Based on the beam configuration parameters, the cross-site cooperative transmission strategy, the clustering results, and the interference coordination strategy, a state space of the Markov decision process is constructed. The state space includes the channel quality, buffer state, cluster identifier, embedding vector in the stereo interference map, predicted trajectory timing features, and resource avoidance rules in the interference coordination strategy for each low-altitude user. The resource block allocation strategy and modulation and coding scheme are constructed as the action space of the Markov decision process. Based on the comprehensive reward function, the Markov decision process is solved using a deep reinforcement learning algorithm, and the resource block allocation strategy and the modulation and coding scheme are output as the network scheduling parameters. The comprehensive reward function is obtained by weighting the total throughput, fairness index and number of handover interruptions.

[0014] In one embodiment, performing mobility management on the low-altitude user based on the predicted trajectory and the network scheduling parameters includes: Based on the predicted trajectory data, the target cell edge area that the low-altitude user will enter in a future preset time period is identified, and the target cell corresponding to the target cell edge area is determined as a potential handover target cell. Based on the network scheduling parameters, obtain the resource availability status of the current serving cell and the potential handover target cell; Based on the resource availability status of the potential handover target cell, before the low-altitude user enters the edge area of ​​the target cell, a dual connection is established between the current serving cell and the potential handover target cell, and the data to be transmitted by the low-altitude user is forwarded to the potential handover target cell to complete the predictive seamless handover. or, Based on the real-time service load and user distribution information in the network scheduling parameters, a network controller is defined to aggregate multiple physically separated physical base stations into a single logical cell, wherein the single logical cell is a super-large coverage cell. Based on the predicted trajectory data, the coverage boundary of the individual logical cell is dynamically adjusted; When the low-altitude user moves between the multiple physical base stations, the low-altitude user is always within the coverage area of ​​the single logical cell, and the low-altitude user stops performing cell handover.

[0015] Secondly, embodiments of this application provide a dynamic cooperative network communication system for low-altitude three-dimensional coverage. The system is used to execute the above-described method and includes a prediction and clustering module, an interference coordination strategy module, a network scheduling parameter module, and a communication configuration module; wherein: The prediction and clustering module is used to acquire multi-dimensional feature data of low-altitude users, and based on the multi-dimensional feature data, to perform three-dimensional trajectory prediction of the low-altitude users, acquire predicted trajectory data, and perform clustering processing based on the predicted trajectory data to obtain clustering results. The interference coordination strategy module is used to construct a three-dimensional interference map including base station nodes and low-altitude user nodes, and to process the three-dimensional interference map through a graph neural network to generate an interference coordination strategy. The network scheduling parameter module is used to perform cooperative scheduling processing based on the predicted trajectory data, the clustering results, and the interference coordination strategy to obtain network scheduling parameters. The communication configuration module is used to perform mobility management on the low-altitude user based on the predicted trajectory and the network scheduling parameters.

[0016] Thirdly, embodiments of this application provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements a dynamic collaborative network communication method for low-altitude three-dimensional coverage as described in the first aspect above.

[0017] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a dynamic cooperative network communication method for low-altitude three-dimensional coverage as described in the first aspect above.

[0018] The dynamic collaborative network communication method and system for low-altitude three-dimensional coverage provided in this application embodiment have at least the following technical effects.

[0019] By constructing an intelligent closed-loop process of "perception-prediction-decision-execution," the limitations of existing technologies in low-altitude three-dimensional coverage are fundamentally overcome, realizing a shift from "passive response" to "proactive planning." Specifically, through three-dimensional trajectory prediction and dynamic clustering, the movement trends of user groups can be proactively grasped, providing a basis for resource pre-allocation; through the construction and perception of three-dimensional interference maps, traditional local and static interference coordination is upgraded to global and dynamic intelligent interference management; through collaborative scheduling processing, precise control with deep integration in the spatial domain, collaborative domain, and resource domain is achieved, and it has self-evolution capabilities, adapting to complex and ever-changing low-altitude environments; through the execution of mobility management, the problem of service interruption caused by high-speed movement is effectively solved, addressing the issue of how to improve communication quality in related technologies.

[0020] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. Attached Figure Description

[0021] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of a dynamic collaborative network communication method for low-altitude three-dimensional coverage; Figure 2 This is a schematic diagram illustrating three-dimensional trajectory prediction and clustering processing according to an exemplary embodiment; Figure 3 This is a schematic diagram illustrating the construction of a stereo interference map and graph neural network processing according to an exemplary embodiment; Figure 4 This is a schematic diagram illustrating cross-site asynchronous collaborative processing according to an exemplary embodiment; Figure 5 This is a schematic diagram illustrating predictive seamless switching according to an exemplary embodiment; Figure 6 This is a system structure block diagram of a dynamic collaborative network communication system for low-altitude three-dimensional coverage, according to an exemplary embodiment. Figure 7 This is a structural block diagram of an electronic device according to an exemplary embodiment. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this application clearer, the application is described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0023] Obviously, the accompanying drawings described below are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar scenarios based on these drawings without any inventive effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, any changes to design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as insufficient disclosure of the content of this application.

[0024] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0025] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms “a,” “an,” “an,” “the,” and similar words used in this application do not indicate quantity limitation and may indicate singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms “connected,” “linked,” “coupled,” and similar words used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following objects are in an "or" relationship. The terms "first," "second," and "third" used in this application are merely to distinguish similar objects and do not represent a specific ordering of the objects.

[0026] In this document, it should be understood that the terms used may be technical means used to implement part of the present invention or other summary technical terms. For example, the terms may include: Predicted trajectory data: Spatiotemporal sequence data obtained by estimating the future three-dimensional position, velocity, and heading of low-altitude users (such as drones) through a predictive model. It typically includes the three-dimensional coordinates and corresponding probability distributions for multiple future time steps.

[0027] Three-dimensional interference graph: a heterogeneous graph structure containing base station nodes and low-altitude user nodes, wherein the edges are divided into service edges (connecting users to their serving base stations) and interference edges (connecting users to base stations that cause co-channel interference to them), used to characterize the radio interference topology in the low-altitude-ground three-dimensional space.

[0028] Interference coordination strategies: control decisions used to reduce or avoid co-channel interference, such as suggesting that multiple base stations perform joint transmission (JT) on a designated resource block, or adjusting the beam downtilt angle of base stations to reduce interference to specific users.

[0029] Primary serving base station: In cross-base station collaborative transmission (such as CoMP) scenarios, the base station responsible for scheduling user data, initiating collaborative processes, and serving as the main data transmission anchor point.

[0030] Cooperative base stations: Other base stations that participate in joint transmission or joint reception for the same low-altitude user under the coordination of the primary serving base station. Cooperative base stations and the primary serving base station typically exchange data and scheduling information through non-ideal backhaul (such as the X2 interface).

[0031] Time-frequency resources: Two-dimensional physical resources in wireless communication, defined by time units (such as symbols and time slots) and frequency units (such as subcarriers and resource blocks), are used to distinguish the transmissions of different users or links.

[0032] Cross-site cooperative transmission strategy: When low-altitude users cross the coverage areas of multiple base stations, a multi-base station joint transmission / reception scheme is formulated to overcome non-ideal backhaul latency and improve edge performance. It includes specific actions such as latency compensation scheduling, data pre-distribution, HARQ process borrowing, and synchronous joint transmission.

[0033] Predictive seamless handover: Based on three-dimensional trajectory prediction, a handover method is used to establish dual connections with the target base station and complete data forwarding before low-altitude users enter the edge of the target cell, so that the user plane interruption latency during the handover process is close to zero (seamless).

[0034] A single logical cell is a logical entity with a unified physical cell identifier (PCI) formed by dynamically merging the coverage areas of multiple physical base stations through software-defined networking (SDN) technology. When a user moves within this logical cell, physical layer handover is not triggered, thereby reducing the number of handovers.

[0035] Mobility management: In this embodiment, it specifically refers to a proactive and intelligent mobility control process for low-altitude users (such as drones) based on three-dimensional predicted trajectories and network scheduling parameters, targeting low-altitude three-dimensional coverage scenarios. This includes, but is not limited to, the following two implementation methods: Predictive seamless handover: Based on the predicted trajectory, handover points are identified in advance, and dual connections are established and data is transmitted before the user enters the edge of the target cell, achieving handover with zero interruption.

[0036] Virtual cell merging: Multiple physical base stations are dynamically aggregated into a single logical cell through a software-defined network controller. When a user moves within this logical cell, physical layer handover is not triggered, thereby reducing the number of handovers and interruptions.

[0037] The mobility management in this patent differs from traditional handover mechanisms based on real-time measurement and event triggering. Its core feature is to use predicted trajectories to anticipate mobile behavior, combine network scheduling parameters to obtain resource status, and proactively execute handover preparation or logical cell reorganization, thereby ensuring communication continuity and service quality for low-altitude users in highly dynamic, three-dimensional mobile environments.

[0038] Firstly, embodiments of this application provide a dynamic collaborative network communication method for low-altitude three-dimensional coverage. Figure 1 This is a flowchart of a dynamic collaborative network communication method for low-altitude three-dimensional coverage, such as... Figure 1 As shown, the method includes: Step S101: Obtain multidimensional feature data of low-altitude users, and perform three-dimensional trajectory prediction on low-altitude users based on the multidimensional feature data to obtain predicted trajectory data. Perform clustering processing based on the predicted trajectory data to obtain clustering results.

[0039] Step S102: Construct a three-dimensional interference map including base station nodes and low-altitude user nodes, process the three-dimensional interference map using a graph neural network, and generate an interference coordination strategy.

[0040] Step S103: Based on the predicted trajectory data, clustering results, and interference coordination strategy, perform cooperative scheduling processing to obtain network scheduling parameters.

[0041] Step S104: Perform mobility management for low-altitude users based on the predicted trajectory and network scheduling parameters.

[0042] In summary, this application provides a dynamic collaborative network communication method for low-altitude three-dimensional coverage. By constructing an intelligent closed-loop process of "perception-prediction-decision-execution," it fundamentally overcomes the limitations of existing technologies in low-altitude three-dimensional coverage, realizing a shift from "passive response" to "proactive planning." Specifically, through three-dimensional trajectory prediction and dynamic clustering, it can proactively grasp the movement trends of user groups, providing a basis for resource pre-allocation; through three-dimensional interference map construction and perception, it elevates traditional local and static interference coordination to global and dynamic intelligent interference management; through collaborative scheduling processing, it achieves precise control with deep integration in the spatial domain, collaborative domain, and resource domain, and possesses self-evolution capabilities, adapting to complex and ever-changing low-altitude environments; and by executing mobility management, it effectively solves the problem of service interruption caused by high-speed movement, addressing the issue of how to improve communication quality in related technologies.

[0043] In one embodiment, Figure 2 This is a schematic diagram illustrating three-dimensional trajectory prediction and clustering processing according to an exemplary embodiment, such as... Figure 2 As shown, step S101 involves acquiring multidimensional feature data of low-altitude users, predicting their three-dimensional trajectories based on this data, obtaining predicted trajectory data, and performing clustering processing on the predicted trajectory data to obtain clustering results. Specifically, this includes the following steps: Step S1011: Obtain multidimensional feature data for each low-altitude user. The multidimensional feature data includes real-time status data, historical trajectory data, and context data. The real-time status data includes three-dimensional coordinates, instantaneous velocity vector, and heading angle. The context data includes mission type, preset waypoint, and airspace restrictions. Step S1012: Input the multidimensional feature data into the prediction model, and output the three-dimensional trajectory probability distribution of low-altitude users in the future time period as the predicted trajectory data. The prediction model includes an encoder, a social pooling layer, and a decoder; where... The encoder extracts temporal dependencies from multidimensional feature data to generate hidden states. By using the social pooling layer, based on the hidden state and the graph structure relationship between low-altitude users and spatially neighboring users, the influence weight of each neighboring user on the future trajectory of the low-altitude user is calculated using the attention mechanism to obtain the social pooling feature. A graph attention network is introduced into the social pooling layer to construct a graph structure for each low-altitude user and spatially neighboring users. The influence weight of neighboring users on future trajectories is quantified through the attention mechanism to obtain social pooling features. Based on the hidden state and social pooling features, the decoder recursively outputs the predicted trajectory data.

[0044] Optionally, when performing 3D trajectory prediction, multi-dimensional feature data is first collected for each low-altitude user, including real-time state data (3D coordinates, instantaneous velocity vector, heading angle), historical trajectory data (serialized position and velocity at a preset number of time steps), and contextual data (task type, preset waypoints, airspace restrictions). This data is then input into the prediction model. The prediction model employs an encoder-decoder architecture, where the encoder uses a gated recurrent unit or Transformer to extract the temporal dependencies of the user's own trajectory, generating hidden states. Simultaneously, a social pooling layer introduces a graph attention network, constructing a graph structure around each user and its spatial neighbors. The influence weight of each neighboring user on the user's future trajectory is calculated through an attention mechanism, yielding social pooling features. Based on the hidden states and social pooling features output by the encoder, the decoder recursively outputs the 3D trajectory probability distribution for a preset number of time steps in the future, serving as the predicted trajectory data for that low-altitude user.

[0045] Step S1013: Based on the predicted trajectory data, extract clustering features for each low-altitude user. The clustering features include the centroid of the future location sequence, the average vector of the direction of movement, the trajectory shape features, and the relative geometric relationship with neighboring base stations. Step S1014: Based on clustering features, calculate the trajectory similarity between two low-altitude users using the dynamic time warping algorithm, and generate a trajectory similarity matrix; Step S1015: Based on the similarity matrix, the spectral clustering algorithm is used to cluster low-altitude users to obtain the cluster division results; wherein, the spectral clustering algorithm maps the data to a low-dimensional space through Laplacian matrix eigenvalue decomposition, and then performs clustering through K-means; Step S1016: Based on the preset period, predicted trajectory confidence and low-altitude user location status, perform dynamic cluster management on the cluster division results to obtain clustering results. Dynamic cluster management includes merging, splitting and disbanding clusters.

[0046] Optionally, during dynamic clustering, firstly, based on the predicted trajectory data of each low-altitude user, clustering features are extracted for that user, including the centroid of the future location sequence, the average vector of the direction of movement, the shape features of the trajectory (such as curvature), and the relative geometric relationship with neighboring base stations. Then, based on the clustering features of all users, the Dynamic Time Warping (DTW) algorithm is used to calculate the trajectory similarity between every two users, constructing a trajectory similarity matrix. Next, using the similarity matrix as input, Spectral Clustering is used for clustering: the original feature data is mapped to a low-dimensional space through Laplacian matrix eigenvalue decomposition, and then the K-means algorithm is applied in this low-dimensional space to perform cluster partitioning, obtaining the initial cluster partitioning results. Finally, the above clustering process is re-executed according to a preset period (such as every 10 seconds), and existing clusters are dynamically merged, split, or disbanded based on the confidence level of the predicted trajectory and whether the low-altitude user deviates from the boundary of the current cluster, generating the final clustering results.

[0047] Step S101 employs a GRU / Transformer neural network for multi-step trajectory prediction, and performs dynamic spectral clustering based on the similarity of the predicted trajectories, transforming spatiotemporal sequence data into group movement pattern recognition. This enables proactive understanding of user group movement trends, providing a basis for resource pre-allocation.

[0048] In one embodiment, Figure 3 This is a schematic diagram illustrating the construction of a stereo interference map and graph neural network processing according to an exemplary embodiment, such as... Figure 3 As shown, step S102 involves constructing a three-dimensional interference map including base station nodes and low-altitude user nodes, and processing the three-dimensional interference map using a graph neural network to generate an interference coordination strategy. Specifically, this includes the following steps: Step S1021: Base station nodes and low-altitude user nodes are respectively used as graph nodes. Each graph node includes corresponding node features. Service edges and interference edges are constructed according to the service relationship and interference relationship between nodes. The interference edge weight is calculated according to the SINR measurement value. Based on the graph nodes, service edges, interference edges and interference edge weights, a three-dimensional interference map is obtained.

[0049] Optionally, each base station and each low-altitude user are treated as graph nodes. Base station nodes are assigned features such as location, transmit power, and antenna configuration, while user nodes are assigned features such as location, altitude, serving cell, RSRP (received power) / RSRQ (received quality) / SINR. Then, service edges are constructed based on service relationships (between low-altitude users and their serving base stations), and interference edges are constructed based on interference relationships (between low-altitude users and base stations that cause co-channel interference). The weight of each interference edge is calculated by back-calculating the measured SINR (Signal to Interference plus Noise Ratio), thus obtaining a three-dimensional interference graph including two types of nodes, two types of edges, and edge weights.

[0050] Step S1022: Input the stereo interference map into the graph neural network. Through the message passing mechanism of the graph neural network, aggregate and update the features of each node in the stereo interference map to obtain the embedding vector of each graph node. The graph neural network is a heterogeneous graph neural network, simultaneously processing base station nodes and low-altitude user nodes. Specifically, it includes: Through a message passing mechanism, each graph node aggregates information from its neighboring graph nodes; among them, low-altitude user nodes aggregate signal information from serving base station nodes and interference signal information from interfering base station nodes. Through iterative updates of a multi-layer graph neural network, each graph node obtains an embedding vector that includes global topological relationships and interference relationships.

[0051] Optionally, the three-dimensional interference map is input into a heterogeneous graph neural network, and each node aggregates information from its neighboring nodes through a message-passing mechanism: for example, a low-altitude user node will aggregate useful signal features from the serving base station node connected to it, and at the same time aggregate interference signal features from multiple interfering base station nodes; after iterative updates of the multi-layer graph neural network, the feature vector of each node gradually merges the global topological relationship and the interference relationship, and finally obtains the embedding vector.

[0052] Step S1023: Perform link performance prediction or interference hotspot classification based on the embedded vector, and generate interference coordination strategy based on the prediction or classification results.

[0053] Optionally, based on the embedding vector, link performance prediction (such as predicting the throughput or SINR of a link) or interference hotspot classification (such as identifying areas where the SINR is consistently below a threshold) can be performed, and interference coordination strategies can be generated accordingly, such as "suggesting that base station A and base station B start joint transmission on resource block X" or "suggesting that the beam downtilt angle of base station C be adjusted to avoid interference to low-altitude user D".

[0054] Step S102 explicitly models the radio interference relationship in the low-altitude-ground three-dimensional space as a heterogeneous graph and utilizes the message passing mechanism of graph neural networks for global interference perception, thus overcoming the limitations of traditional interference management that relies solely on local measurements or static planning. Link performance prediction and interference hotspot classification based on embedded vectors can proactively identify potential interference areas and generate targeted collaborative strategies (such as joint transmission or beam adjustment), thereby reducing co-channel interference intensity, improving signal quality and spectral efficiency for low-altitude users, and ultimately enhancing network reliability and coverage in complex low-altitude environments.

[0055] In one embodiment, step S103 involves performing cooperative scheduling processing based on predicted trajectory data, clustering results, and interference coordination strategies to obtain network scheduling parameters. Specifically, this includes the following steps: Step S1031: Based on the predicted trajectory data and clustering results, determine the average altitude distribution and future spatial distribution trend of low-altitude users. With the comprehensive coverage rate of the low-altitude coverage area as the optimization objective, and the beam tilt angle, azimuth angle and beamwidth as optimization variables, combined with the beam interference avoidance requirements in the interference coordination strategy, solve the optimal beam configuration through a genetic algorithm or particle swarm algorithm, and use the optimal beam configuration as the beam configuration parameter.

[0056] Optionally, the conventional beam downtilt angle is fixed, but the embodiments of this application make it dynamic. Specifically, when performing altitude-adaptive beamforming, the average altitude distribution of all low-altitude users within the service area is first calculated based on predicted trajectory data and clustering results, along with future spatial distribution trends (e.g., predicting that the user group will move northeast and slightly ascend within the next 5 seconds). The signal quality coverage rate of the low-altitude coverage area is used as the objective function (e.g., maximizing the proportion of locations with signal received power greater than -95dBm within the area). The downtilt angle, azimuth angle, and beamwidth of the base station beam are used as optimization variables. Simultaneously, beam interference avoidance requirements from the interference coordination strategy are introduced (e.g., avoiding beam main lobe pointing to known strong interference sources or overlapping coverage with beams from neighboring base stations), constructing a constrained optimization problem. A genetic algorithm or particle swarm optimization algorithm is used to solve this problem, iteratively searching for the optimal beam configuration parameters that maximize low-altitude coverage while satisfying interference avoidance constraints. Finally, the optimal beam parameters (beam downtilt angle, azimuth angle, and beamwidth) obtained from the solution are output as beam configuration parameters for real-time adjustment of the base station's beam direction and shape.

[0057] Beneficial effects: It can dynamically adjust beam parameters according to user distribution, align the main lobe of the beam with the altitude layer where low-altitude users are most concentrated, and at the same time use the beam width to cover the user's altitude fluctuation range, avoiding signal attenuation caused by beam pointing deviation or insufficient width.

[0058] Figure 4This is a schematic diagram illustrating cross-site asynchronous collaborative processing according to an exemplary embodiment, such as... Figure 4 As shown, in step S1032, based on the predicted trajectory data, clustering results, and interference coordination strategy, cross-site asynchronous cooperative processing is performed to obtain the cross-site cooperative transmission strategy. Specifically, this includes: Based on the clustering results, the set of cooperating base stations corresponding to low-altitude users in the same cluster is determined. Based on the predicted trajectory data, the cross-site movement trend of low-altitude users is predicted. Combined with the joint transmission interference optimization requirements in the interference coordination strategy, the main serving base station and cooperating base stations are determined. The primary serving base station schedules low-altitude user data at the current moment and presets the transmission time to the sum of the current moment and the maximum latency of the cooperating base stations. The maximum latency of the cooperating base stations is the maximum backhaul latency from the primary serving base station to all cooperating base stations. The primary serving base station sends low-altitude user data and resource allocation information to all cooperating base stations through the inter-base station interface to complete data pre-distribution; The primary serving base station instructs low-altitude users via radio resource control signaling to mix inactive secondary carriers with automatic repeat request processes to receive data arriving at the transmission time on the primary carrier. At the transmission time, the primary serving base station and all cooperating base stations use the same time and frequency resources to jointly transmit the same user data to low-altitude users, thus obtaining a cross-site cooperative transmission strategy.

[0059] Optionally, when performing cross-site asynchronous cooperative transmission, the set of cooperative base stations corresponding to low-altitude users within the same cluster is first determined based on the clustering results. The cross-site movement trend of low-altitude users is predicted based on predicted trajectory data (e.g., predicting that a drone will enter the coverage edge of base station B in 300ms). This is combined with the joint transmission interference optimization requirements in the interference coordination strategy (e.g., the interference coordination strategy indicates that joint transmission between base station A and base station B on a resource block can avoid interference to user D), thereby locking in the primary serving base station and cooperative base stations. Subsequently, the primary serving base station (SeNB) schedules the data of a low-altitude user at the current time (e.g., t0), and simultaneously calculates the return latency from itself to all cooperative base stations (Co-NB), taking the maximum value as the maximum return latency Δt_max. The actual data transmission time is preset to the current time plus this maximum return latency (i.e., t0). +Δt_max); then, the primary serving base station immediately sends the user data packet and resource allocation information to all cooperating base stations in advance through the inter-base station interface (such as the X2 interface) to complete the data pre-distribution; at the same time, the primary serving base station notifies the low-altitude user through Radio Resource Control (RRC) signaling to borrow the Hybrid Automatic Repeat Request (HARQ) process on the secondary carrier (SCC) that has not yet been activated, in order to receive the data that will arrive on the primary carrier at the preset transmission time; finally, at the preset transmission time t0 + Δt_max, the primary serving base station and all cooperating base stations simultaneously transmit the exact same data to the low-altitude user on the same time and frequency resources, thereby achieving joint transmission and obtaining the cross-site cooperative transmission strategy.

[0060] It should be noted that the resource allocation information includes the time and frequency resource location parameters, transmission format parameters, synchronization control parameters, user identification parameters, beam direction constraint parameters, and interference avoidance rules related to this joint transmission.

[0061] Beneficial effects: By pre-setting the transmission time to compensate for the non-ideal backhaul delay between base stations, and by using the hybrid automatic repeat request process of the auxiliary carrier to open a dedicated receiving channel for this transmission, multiple base stations can achieve synchronous joint transmission under non-ideal backhaul conditions. This avoids the problem of other services being blocked due to waiting for data from the farthest base station in traditional cooperative multi-point systems, and effectively reduces the delay and interruption probability when moving across base stations.

[0062] Step S1033: Based on beam configuration parameters, cross-site cooperative transmission strategy, clustering results, and interference coordination strategy, perform reinforcement learning resource allocation processing to obtain network scheduling parameters. Specifically, this includes: Based on beam configuration parameters, cross-site cooperative transmission strategy, clustering results, and interference coordination strategy, a state space for a Markov decision process is constructed. The state space includes the channel quality, buffer state, cluster identifier, embedding vector in the stereo interference map, predicted trajectory timing characteristics, and resource avoidance rules in the interference coordination strategy for each low-altitude user. The resource block allocation strategy and modulation and coding strategy are constructed as the action space of a Markov decision process; Based on the comprehensive reward function, a deep reinforcement learning algorithm is used to solve the Markov decision process, and the output resource block allocation strategy and modulation coding strategy are used as network scheduling parameters. The comprehensive reward function is obtained by weighting the total throughput, fairness index and number of handover interruptions.

[0063] Optionally, when performing reinforcement learning resource allocation processing, the state space of the Markov decision process is first constructed based on the previously obtained beam configuration parameters, cross-site cooperative transmission strategy, clustering results, and interference coordination strategy. This state space includes the channel quality, buffer state, cluster identifier, embedding vector obtained from the stereo interference map, predicted trajectory temporal characteristics, and resource avoidance rules in the interference coordination strategy for each low-altitude user, to comprehensively characterize the current network resource status and interference environment. Then, the resource block allocation strategy and modulation and coding scheme strategy are defined as the action space, that is, the scheduler can select specific resource blocks (RBs) and modulation and coding schemes (MCS) for each user. Next, a comprehensive reward function is designed, which obtains the immediate reward value by weighted summing of the total system throughput, the fairness index among low-altitude users (such as Jain's Fairness Index), and the number of handover interruptions (as a penalty term), which can be expressed by the following formula: R = w1 * total system throughput + w2 * fairness index (such as Jain's Fairness Index) - w3 * number of handover interruptions. Where w1, w2, and w3 are weight coefficients. Finally, a deep reinforcement learning algorithm (such as Deep Q-Network (DQN) or Proximal Policy Optimization (PPO)) is used to solve the Markov decision process. The agent learns the optimal scheduling strategy through continuous interaction with the network environment and outputs the resource block allocation and modulation coding scheme for each low-altitude user as network scheduling parameters.

[0064] It is important to note that a resource block (RB) is a two-dimensional resource unit consisting of 12 consecutive subcarriers in the frequency domain and one time slot (usually containing 14 OFDM symbols) in the time domain. It is the smallest physical granularity for resource allocation by the base station scheduler. A modulation and coding scheme (MCS) is an index value (e.g., 0-31), with each index corresponding to a specific set of modulation schemes (e.g., QPSK, 16QAM, 64QAM), target code rate, and spectral efficiency. The deep neural network of the deep reinforcement learning agent (scheduler) takes the current state space (including the channel quality CQ1 of each user, the buffer state BSR: cluster identifier, the embedding vector output by the GNN, etc.) as input. Through forward propagation calculation, it directly selects the specific resource block allocation (e.g., allocating RB 10-20 for user A) and modulation and coding scheme (e.g., selecting MCS 15) that maximizes long-term rewards for each user as the action to be executed at the current scheduling moment.

[0065] Beneficial effects: By incorporating beam configuration and cross-site cooperation strategies into the state space, the reinforcement learning scheduler can perceive the impact of the current beam pointing and cooperative transmission mode on user channel quality and interference, thereby making more coordinated resource allocation decisions.

[0066] In one embodiment, step S104 involves performing mobility management for low-altitude users based on the predicted trajectory and network scheduling parameters. Specifically, this includes the following steps: Step S104a1: Based on the predicted trajectory data, identify the target cell edge area that the low-altitude user will enter in the future preset time period, and determine the target cell corresponding to the target cell edge area as the potential handover target cell; Step S104a2: Obtain the resource availability status of the current serving cell and the potential handover target cell according to the network scheduling parameters; Step S104a3, based on the resource availability status of the potential handover target cell, establishes a dual connection between the current serving cell and the potential handover target cell before the low-altitude user enters the edge area of ​​the target cell, and forwards the data to be transmitted by the low-altitude user to the potential handover target cell to complete the predictive seamless handover. or, Step S104b1 defines a network controller based on the real-time service load and user distribution information in the network scheduling parameters, and aggregates multiple physically separated physical base stations into a single logical cell, which is a super-large coverage cell. Step S104b2 dynamically adjusts the coverage boundary of a single logical cell based on the predicted trajectory data; In step S104b3, when a low-altitude user moves between multiple physical base stations, the low-altitude user is always within the coverage area of ​​a single logical cell, and the low-altitude user stops performing cell handover.

[0067] Optionally, either predictive seamless handover or virtual cell merging can be used. Figure 5 This is a schematic diagram illustrating predictive seamless switching according to an exemplary embodiment, such as... Figure 5 As shown, for predictive seamless handover, firstly, based on the three-dimensional predicted trajectory of each low-altitude user, the system identifies the target cell edge area that the user will enter within a preset time period (e.g., 500 milliseconds) in the future, and determines the target base station corresponding to this area as the potential handover target cell. Then, based on network scheduling parameters, the system obtains the resource availability status (such as the number of remaining resource blocks and load status) of the current serving cell and the potential handover target cell. If resources are sufficient, a dual connection is established between the current serving cell and the potential handover target cell before the low-altitude user actually enters the edge area, and the user's data to be transmitted is pre-transmitted to the target cell, thus achieving predictive seamless handover when the user crosses the cell boundary. For example, for a logistics drone flying along a fixed route, the system predicts that it will enter the coverage edge of base station B in 300 milliseconds, so it establishes a dual connection with base station B in advance and caches the data, making the handover seamless.

[0068] For virtual cell merging, based on real-time service load and user distribution information in network scheduling parameters, a network controller (such as a software-defined network controller) dynamically aggregates multiple physically separate physical base stations into a single logical cell (i.e., a super-large coverage cell). Then, based on predicted trajectory data, the coverage boundary of this logical cell is dynamically adjusted to ensure it always surrounds the currently active low-altitude user group. When low-altitude users move between these multiple physical base stations, they do not need to perform physical layer cell handover because they are always within the coverage area of ​​the same logical cell. For example, in areas with dense drone delivery, the controller merges three adjacent base stations into one logical cell, and drones flying within it do not require any handover operations.

[0069] Predictive seamless handover prepares for proactive handover based on trajectory prediction by pre-emptively triggering handover based on real-time measurement, eliminating data interruptions and signaling delays caused by handover decision lags. This enables low-altitude users to achieve zero-interruption service continuity during high-speed cross-cell movement. Virtual cell merging abstracts multiple physical base stations into a single logical cell, fundamentally avoiding frequent handovers between base stations and reducing handover signaling overhead and the probability of interruption. This is particularly suitable for dense flight scenarios involving drone swarms within fixed areas. The two mechanisms complement each other: predictive handover handles movement across logical cells, while virtual cell merging reduces unnecessary handovers at the source, jointly ensuring communication quality and user experience for low-altitude users in highly dynamic, three-dimensional mobile environments.

[0070] The following are specific embodiments of this application in practice: 1. Overall Architecture A complete low-altitude intelligent communication system was constructed and deployed in a smart city drone logistics network. It consists of a closed-loop control system comprised of four core modules, enabling complete intelligent management from perception to execution.

[0071] 1) System configuration parameters: Coverage area: 50 square kilometers of the city center; Number of base stations: 64 5G-A base stations (64TR AAU); Service targets: 30 logistics drones; Network architecture: dual carrier (3.4GHz terrestrial coverage + 3.5GHz low-altitude coverage).

[0072] 2) 3D trajectory prediction and dynamic intelligent clustering module ① Trajectory prediction algorithm logic: A GRU neural network employing an encoder-decoder architecture.

[0073] Input features: 9-dimensional vector (3D coordinates, 3D velocity, heading angle, acceleration, mission type).

[0074] Processing flow: 50 historical time steps → encoder → context vector → decoder → prediction of the next 30 time steps.

[0075] Key parameters: Hidden layer dimension 128, number of network layers 3, learning rate 0.001.

[0076] ② Dynamic clustering algorithm logic A spectral clustering algorithm based on predicted trajectories is used for similarity calculation: dynamic time-warped distance is used to measure trajectory similarity. The clustering process involves: constructing a similarity matrix → Laplacian matrix eigenvalue decomposition → K-means clustering. The update mechanism involves re-clustering every 10 seconds, adaptively determining the number of clusters (2-8). 3) 3D Interference Map Construction and Sensing Module ① Graph construction logic Node types: base station nodes (64), user nodes (300).

[0077] Edge relationships: service edge (base station - user), interference edge (interference relationship).

[0078] Feature dimensions: 8 dimensions for base station nodes and 6 dimensions for user nodes.

[0079] ② Graph Neural Network Processing Logic Architecture: 3-layer graph convolutional network.

[0080] Message passing: Neighbor node information aggregation → Feature update → Global relationship awareness.

[0081] Output: Interference hotspot identification, collaborative strategy generation.

[0082] Processing cycle: Model updates every 60 seconds, real-time inference response time <80ms 4) Multi-layered collaborative scheduling engine ① Highly adaptive beamforming Optimization objective: Maximize the proportion of low-altitude coverage areas with RSRP > -95dBm.

[0083] Adjust parameters: downtilt angle (-5° to +15°), beamwidth (15° to 65°).

[0084] Decision frequency: Re-optimize every 1 second based on user height distribution ② Cross-site asynchronous CoMP mechanism Delay compensation: Preset transmission time = scheduling time + maximum return transmission delay.

[0085] Process reuse: The HARQ process on the CA secondary carrier is used to serve the data transmission of the primary carrier.

[0086] Collaborative process: Data pre-distribution → Delay alignment → Multi-station synchronous transmission.

[0087] ③ Strengthen the allocation of learning resources State space: 300 users × 15-dimensional features (CQI, buffer, cluster affiliation, etc.).

[0088] Action space: resource block allocation × modulation and coding scheme × power control.

[0089] Reward function: Throughput weight 0.6 + Fairness weight 0.3 - Switching interruption weight 0.1.

[0090] Learning algorithm: Q-learning, learning rate 0.1, discount factor 0.9.

[0091] 5) Intelligent handover and mobility management ① Predictive switching logic Prediction window: Predicts the movement path for the next 10 seconds based on the trajectory.

[0092] Switching trigger: Identify potential switching points 3 seconds in advance.

[0093] Execution mechanism: Target base station pre-connection + data forwarding ② Virtual Community Management Merging strategy: Dynamically combine 3-8 physical base stations based on service density.

[0094] PCI Configuration: Uniform PCI within the virtual cell eliminates boundaries.

[0095] Reorganization assessment: Re-optimize based on user distribution every 30 seconds.

[0096] In summary, this application provides a dynamic collaborative network communication method for low-altitude three-dimensional coverage. By constructing an intelligent closed-loop process of "perception-prediction-decision-execution," it fundamentally overcomes the limitations of existing technologies in low-altitude three-dimensional coverage, realizing a shift from "passive response" to "proactive planning." Specifically, through three-dimensional trajectory prediction and dynamic clustering, it can proactively grasp the movement trends of user groups, providing a basis for resource pre-allocation; through three-dimensional interference map construction and perception, it elevates traditional local and static interference coordination to global and dynamic intelligent interference management; through collaborative scheduling processing, it achieves precise control with deep integration in the spatial domain, collaborative domain, and resource domain, and possesses self-evolution capabilities, adapting to complex and ever-changing low-altitude environments; and by executing mobility management, it effectively solves the problem of service interruption caused by high-speed movement, addressing the issue of how to improve communication quality in related technologies.

[0097] Secondly, embodiments of this application provide a dynamic collaborative network communication system for low-altitude three-dimensional coverage. Figure 6 This is a system architecture block diagram illustrating a dynamic collaborative network communication system for low-altitude three-dimensional coverage, according to an exemplary embodiment. For example... Figure 6 As shown, the system includes a prediction and clustering module 610, an interference coordination strategy module 620, a network scheduling parameter module 630, and a communication configuration module 640; wherein: The prediction and clustering module 610 is used to acquire multi-dimensional feature data of low-altitude users, and to perform three-dimensional trajectory prediction of low-altitude users based on the multi-dimensional feature data, to acquire predicted trajectory data, and to perform clustering processing based on the predicted trajectory data to obtain clustering results. The interference coordination strategy module 620 is used to construct a three-dimensional interference map including base station nodes and low-altitude user nodes, and to process the three-dimensional interference map through a graph neural network to generate an interference coordination strategy. The network scheduling parameter module 630 is used to perform cooperative scheduling processing based on predicted trajectory data, clustering results and interference coordination strategies to obtain network scheduling parameters. The communication configuration module 640 is used to perform mobility management for low-altitude users based on predicted trajectories and network scheduling parameters.

[0098] In summary, this application provides a dynamic collaborative network communication system for low-altitude three-dimensional coverage. By constructing an intelligent closed-loop system of "perception-prediction-decision-execution," it fundamentally overcomes the limitations of existing technologies in low-altitude three-dimensional coverage, realizing a shift from "passive response" to "proactive planning." Specifically, through three-dimensional trajectory prediction and dynamic clustering, it can proactively grasp the movement trends of user groups, providing a basis for resource pre-allocation; through three-dimensional interference map construction and perception, it elevates traditional local and static interference coordination to global and dynamic intelligent interference management; through collaborative scheduling processing, it achieves precise control with deep integration in the spatial domain, collaborative domain, and resource domain, and possesses self-evolution capabilities, adapting to complex and ever-changing low-altitude environments; and through mobility management, it effectively solves the problem of service interruption caused by high-speed movement, addressing the issue of how to improve communication quality in related technologies.

[0099] It should be noted that the dynamic collaborative network communication system for low-altitude three-dimensional coverage provided in this embodiment is used to implement the above-described embodiments, and details already described will not be repeated. As used above, terms such as "module," "unit," and "subunit" can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the above embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0100] Thirdly, embodiments of this application provide an electronic device, Figure 7 This is a block diagram illustrating an electronic device according to an exemplary embodiment. (e.g.) Figure 7 As shown, the electronic device may include a processor 71 and a memory 72 storing computer program instructions.

[0101] Specifically, the processor 71 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0102] The memory 72 may include a mass storage device for data or instructions. For example, and not limitingly, the memory 72 may include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 72 may include removable or non-removable (or fixed) media. Where appropriate, the memory 72 may be internal or external to a data processing device. In a particular embodiment, the memory 72 is non-volatile memory. In a particular embodiment, the memory 72 includes read-only memory (ROM) and random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), an electrically alterable read-only memory (EAROM), or flash memory, or a combination of two or more of these. Where appropriate, the RAM can be Static Random-Access Memory (SRAM) or Dynamic Random-Access Memory (DRAM). DRAM can be Fast Page Mode Dynamic Random-Access Memory (FPMDRAM), Extended Data Out Dynamic Random-Access Memory (EDODRAM), Synchronous Dynamic Random-Access Memory (SDRAM), etc.

[0103] The memory 72 can be used to store or cache various data files that need to be processed and / or communicated, as well as possible computer program instructions executed by the processor 71.

[0104] The processor 71 reads and executes computer program instructions stored in the memory 72 to implement any of the dynamic cooperative network communication methods for low-altitude three-dimensional coverage in the above embodiments.

[0105] In one embodiment, a dynamic collaborative network communication device for low-altitude three-dimensional coverage may further include a communication interface 73 and a bus 70. Wherein, as Figure 7 As shown, the processor 71, memory 72, and communication interface 73 are connected through bus 70 and complete communication with each other.

[0106] The communication interface 73 is used to enable communication between the various modules, devices, units, and / or equipment in the embodiments of this application. The communication interface 73 can also enable data communication with other components such as external devices, image / data acquisition devices, databases, external storage, and image / data processing workstations.

[0107] Bus 70 includes hardware, software, or both, that couples together components of a dynamic cooperative network communication device for low-altitude three-dimensional coverage. Bus 70 includes, but is not limited to, at least one of the following: data bus, address bus, control bus, expansion bus, and local bus. For example, and not as a limitation, bus 70 may include Accelerated GraphicsPort (AGP) or other graphics buses, Extended Industry Standard Architecture (EISA) buses, Front Side Bus (FSB), HyperTransport (HT) interconnects, Industry Standard Architecture (ISA) buses, InfiniBand interconnects, Low Pin Count (LPC) buses, memory buses, Micro Channel Architecture (MCA) buses, Peripheral Component Interconnect (PCI) buses, PCI-Express (PCI-X) buses, Serial Advanced Technology Attachment (SATA) buses, Video Electronics Standards Association Local Bus (VLB) buses, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 70 may include one or more buses. Although specific buses are described and illustrated in the embodiments of this application, this application considers any suitable bus or interconnection.

[0108] Fourthly, embodiments of this application provide a computer-readable storage medium storing a program thereon, which, when executed by a processor, implements a dynamic collaborative network communication method for low-altitude three-dimensional coverage provided in the first aspect.

[0109] The readable storage medium may be more specifically adopted, including but not limited to: portable disk, hard disk, random access memory, read-only memory, erasable programmable read-only memory, optical storage device, magnetic storage device, or any suitable combination thereof.

[0110] In a possible implementation, the present invention can also be implemented as a program product comprising program code, which, when the program product is run on a terminal device, causes the terminal device to perform steps implementing the dynamic cooperative network communication method for low-altitude three-dimensional coverage provided in the first aspect.

[0111] The program code for executing the present invention can be written in any combination of one or more programming languages. The program code can be executed entirely on the user device, partially on the user device, as a standalone software package, partially on the user device and partially on a remote device, or entirely on a remote device.

[0112] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0113] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A dynamic collaborative network communication method for low-altitude three-dimensional coverage, characterized in that, The method includes: Multidimensional feature data of low-altitude users are obtained, and three-dimensional trajectory prediction of the low-altitude users is performed based on the multidimensional feature data to obtain predicted trajectory data. Clustering processing is performed based on the predicted trajectory data to obtain clustering results. A three-dimensional interference map including base station nodes and low-altitude user nodes is constructed, and the three-dimensional interference map is processed by a graph neural network to generate an interference coordination strategy. Based on the predicted trajectory data, the clustering results, and the interference coordination strategy, collaborative scheduling processing is performed to obtain network scheduling parameters. Mobility management is performed on the low-altitude users based on the predicted trajectory and the network scheduling parameters.

2. The dynamic collaborative network communication method for low-altitude three-dimensional coverage according to claim 1, characterized in that, The process of acquiring multidimensional feature data of low-altitude users and predicting their three-dimensional trajectories based on this data to obtain predicted trajectory data includes: Acquire multidimensional feature data for each low-altitude user. The multidimensional feature data includes real-time status data, historical trajectory data, and context data. The real-time status data includes three-dimensional coordinates, instantaneous velocity vector, and heading angle. The context data includes mission type, preset waypoint, and airspace restrictions. The multidimensional feature data is input into the prediction model, and the three-dimensional trajectory probability distribution of the low-altitude user in the future time period is output as the predicted trajectory data.

3. The dynamic collaborative network communication method for low-altitude three-dimensional coverage according to claim 2, characterized in that, The prediction model includes an encoder, a social pooling layer, and a decoder; wherein... The encoder extracts the temporal dependencies in the multidimensional feature data to generate hidden states. Through the social pooling layer, based on the hidden state and the graph structure relationship between the low-altitude user and its spatial neighboring users, the influence weight of each neighboring user on the future trajectory of the low-altitude user is calculated using an attention mechanism to obtain the social pooling feature. A graph attention network is introduced into the social pooling layer to construct a graph structure for each low-altitude user and spatially neighboring users. The influence weight of neighboring users on future trajectories is quantified through the attention mechanism to obtain social pooling features. Based on the hidden state and the social pooling feature, the decoder recursively outputs predicted trajectory data.

4. The dynamic collaborative network communication method for low-altitude three-dimensional coverage according to claim 3, characterized in that, The clustering process based on the predicted trajectory data to obtain the clustering results includes: Based on the predicted trajectory data, clustering features are extracted for each low-altitude user. The clustering features include the centroid of the future location sequence, the average vector of the direction of movement, the trajectory shape features, and the relative geometric relationship with neighboring base stations. Based on the clustering features, the trajectory similarity between the two low-altitude users is calculated using a dynamic time warping algorithm to generate a trajectory similarity matrix. Based on the similarity matrix, a spectral clustering algorithm is used to cluster low-altitude users to obtain cluster division results; wherein, the spectral clustering algorithm maps the data to a low-dimensional space through Laplacian matrix eigenvalue decomposition, and then performs clustering through K-means; Based on the preset period, the confidence level of the predicted trajectory, and the location status of low-altitude users, the cluster division results are dynamically managed to obtain clustering results. The dynamic cluster management includes merging, splitting, and disbanding clusters.

5. A dynamic collaborative network communication method for low-altitude three-dimensional coverage according to claim 1, characterized in that, A three-dimensional interference map including base station nodes and low-altitude user nodes is constructed. The three-dimensional interference map is then processed using a graph neural network to generate an interference coordination strategy, including: Base station nodes and low-altitude user nodes are respectively used as graph nodes. Each graph node includes corresponding node features. Service edges and interference edges are constructed according to the service relationship and interference relationship between nodes. The interference edge weight is calculated according to the SINR measurement value. Based on the graph nodes, the service edges, the interference edges and the interference edge weight, a three-dimensional interference map is obtained. The stereo interference image is input into a graph neural network, and the features of each node in the stereo interference image are aggregated and updated through the message passing mechanism of the graph neural network to obtain the embedding vector of each graph node. Based on the embedded vector, link performance is predicted or interference hotspots are classified, and interference coordination strategies are generated based on the prediction or classification results.

6. The dynamic collaborative network communication method for low-altitude three-dimensional coverage according to claim 1, characterized in that, Based on the predicted trajectory data, the clustering results, and the interference coordination strategy, the cooperative scheduling process is performed to obtain network scheduling parameters, including: Based on the predicted trajectory data and the clustering results, the average altitude distribution and future spatial distribution trend of low-altitude users are determined. The comprehensive coverage rate of the low-altitude coverage area is used as the optimization target, and the beam tilt angle, azimuth angle and beamwidth are used as optimization variables. Combined with the beam interference avoidance requirements in the interference coordination strategy, the optimal beam configuration is solved by genetic algorithm or particle swarm algorithm, and the optimal beam configuration is used as the beam configuration parameter. Based on the predicted trajectory data, the clustering results, and the interference coordination strategy, cross-station asynchronous cooperative processing is performed to obtain the cross-station cooperative transmission strategy. Based on the beam configuration parameters, the cross-site cooperative transmission strategy, the clustering results, and the interference coordination strategy, reinforcement learning resource allocation processing is performed to obtain the network scheduling parameters.

7. A dynamic collaborative network communication method for low-altitude three-dimensional coverage according to claim 6, characterized in that, Based on the predicted trajectory data, the clustering results, and the interference coordination strategy, cross-site asynchronous cooperative processing is performed to obtain a cross-site cooperative transmission strategy, including: Based on the clustering results, the set of cooperating base stations corresponding to low-altitude users in the same cluster is determined. Based on the predicted trajectory data, the cross-site movement trend of low-altitude users is predicted. Combined with the joint transmission interference optimization requirements in the interference coordination strategy, the primary serving base station and cooperating base stations are determined. The primary serving base station schedules low-altitude user data at the current moment and presets the transmission time to be the sum of the current moment and the maximum delay of the cooperating base stations, wherein the maximum delay of the cooperating base stations is the maximum backhaul delay from the primary serving base station to all the cooperating base stations; The primary serving base station sends the low-altitude user data and resource allocation information to all cooperating base stations through the base station interface to complete the data pre-distribution. The primary serving base station instructs the low-altitude user via radio resource control signaling to mix the inactive secondary carrier with the automatic repeat request process to receive data arriving at the transmission time on the primary carrier. At the transmission time, the primary serving base station and all cooperating base stations use the same time and frequency resources to jointly transmit the same user data to the low-altitude user, thus obtaining a cross-site cooperative transmission strategy.

8. A dynamic collaborative network communication method for low-altitude three-dimensional coverage according to claim 6, characterized in that, Based on the beam configuration parameters, the cross-site cooperative transmission strategy, the clustering results, and the interference coordination strategy, reinforcement learning resource allocation processing is performed to obtain the network scheduling parameters, including: Based on the beam configuration parameters, the cross-site cooperative transmission strategy, the clustering results, and the interference coordination strategy, a state space of the Markov decision process is constructed. The state space includes the channel quality, buffer state, cluster identifier, embedding vector in the stereo interference map, predicted trajectory timing features, and resource avoidance rules in the interference coordination strategy for each low-altitude user. The resource block allocation strategy and modulation and coding scheme are constructed as the action space of the Markov decision process. Based on the comprehensive reward function, the Markov decision process is solved using a deep reinforcement learning algorithm, and the resource block allocation strategy and the modulation and coding scheme are output as the network scheduling parameters. The comprehensive reward function is obtained by weighting the total throughput, fairness index and number of handover interruptions.

9. A dynamic collaborative network communication method for low-altitude three-dimensional coverage according to claim 1, characterized in that, The step of performing mobility management on the low-altitude user based on the predicted trajectory and the network scheduling parameters includes: Based on the predicted trajectory data, the target cell edge area that the low-altitude user will enter in a future preset time period is identified, and the target cell corresponding to the target cell edge area is determined as a potential handover target cell. Based on the network scheduling parameters, obtain the resource availability status of the current serving cell and the potential handover target cell; Based on the resource availability status of the potential handover target cell, before the low-altitude user enters the edge area of ​​the target cell, a dual connection is established between the current serving cell and the potential handover target cell, and the data to be transmitted by the low-altitude user is forwarded to the potential handover target cell to complete the predictive seamless handover. or, Based on the real-time service load and user distribution information in the network scheduling parameters, a network controller is defined to aggregate multiple physically separated physical base stations into a single logical cell, wherein the single logical cell is a super-large coverage cell. Based on the predicted trajectory data, the coverage boundary of the individual logical cell is dynamically adjusted; When the low-altitude user moves between the multiple physical base stations, the low-altitude user is always within the coverage area of ​​the single logical cell, and the low-altitude user stops performing cell handover.

10. A dynamic collaborative network communication system for low-altitude three-dimensional coverage, characterized in that, The system is used to execute the method according to any one of claims 1 to 9, and the system includes a prediction and clustering module, an interference coordination strategy module, a network scheduling parameter module, and a communication configuration module; wherein: The prediction and clustering module is used to acquire multi-dimensional feature data of low-altitude users, and based on the multi-dimensional feature data, to perform three-dimensional trajectory prediction of the low-altitude users, acquire predicted trajectory data, and perform clustering processing based on the predicted trajectory data to obtain clustering results. The interference coordination strategy module is used to construct a three-dimensional interference map including base station nodes and low-altitude user nodes, and to process the three-dimensional interference map through a graph neural network to generate an interference coordination strategy. The network scheduling parameter module is used to perform cooperative scheduling processing based on the predicted trajectory data, the clustering results, and the interference coordination strategy to obtain network scheduling parameters. The communication configuration module is used to perform mobility management on the low-altitude user based on the predicted trajectory and the network scheduling parameters.

Images