An unmanned aerial vehicle trajectory planning and task offloading method for uneven terrain

Abstract

The application discloses a kind of unmanned plane trajectory planning and task unloading method for uneven terrain, it is related to unmanned plane communication and mobile edge computing technical field, fusion topography elevation, space geometry and line-of-sight / Non-line-of-sight propagation characteristics, the communication quality model of access and backhaul link is constructed, and the communication interruption caused by topographic obstruction is converted into spatial constraint;A graph space-time communication state model is constructed, which integrates the interaction of unmanned planes, terminals and base stations, and the topology structure and evolution law are coded using graph neural networks and time convolution networks to predict future communication states;A two-time scale collaborative optimization framework is established: the slow time scale uses model predictive control to roll optimization unmanned plane three-dimensional trajectory;The fast time scale uses Lyapunov optimization to schedule task unloading proportion, bandwidth and computing resources online.The collaborative optimization of trajectory control, resource allocation and task unloading under complex uneven terrain is realized, and the system communication reliability, task processing efficiency and operation stability are improved.

Description

Technical Field

[0001] This invention belongs to the field of UAV communication and mobile edge computing technology, specifically relating to a collaborative optimization method for UAV trajectory planning and task unloading driven by communication continuity perception in uneven terrain environments. Background Technology

[0002] With the development of wireless communication technology and mobile edge computing, unmanned aerial vehicles (UAVs) have been widely used in environmental monitoring, emergency communication, and intelligent services due to their high mobility and flexible deployment capabilities. In UAV-assisted edge computing systems, optimizing UAV flight trajectories and making task offloading decisions can effectively improve system service capabilities and resource utilization efficiency. However, most existing UAV trajectory planning and task offloading methods are based on idealized or simplified communication models, typically assuming that the communication coverage area is regular and stable, which has significant shortcomings in practical applications. Since the UAV communication environment is often affected by factors such as terrain undulations, signal obstruction, and dynamic changes in links, especially under uneven terrain conditions, communication quality exhibits complex spatial non-uniformity. Traditional methods cannot accurately reflect actual communication capabilities, thus affecting system performance.

[0003] In contrast, introducing communicable domain modeling in complex terrain environments helps to more accurately characterize the effective communication area between the UAV and ground users, thus providing a more reasonable basis for trajectory planning and task unloading. However, existing technologies have not fully considered the impact of uneven terrain on the communicable domain, and lack methods for co-optimizing communicable domain constraints with UAV trajectory planning and task unloading.

[0004] Therefore, there is an urgent need for a new technical solution that can construct a communicable domain model in uneven terrain environments and, on this basis, achieve joint optimization of UAV trajectory planning and task unloading to improve the system's adaptability and overall performance. Summary of the Invention

[0005] The technical problem this invention aims to solve is to provide a UAV trajectory planning and task unloading method for uneven terrain, addressing the shortcomings of previous technologies. This method constructs a communication model that integrates terrain elevation information, spatial geometric relationships, and LosS / NLoS propagation characteristics. It explicitly models the impact of terrain occlusion on the communication link and further transforms it into dynamic communicable domain constraints. This enables collaborative optimization between UAV trajectory planning and task unloading processes, thereby improving the communication reliability and computing service performance of the system in complex environments.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: A method for UAV trajectory planning and task offloading for uneven terrain includes a UAV-assisted mobile edge computing system, wherein the UAV-assisted mobile edge computing system includes a UAV, a mobile terminal, a base station, and an uneven terrain environment; Among them, UAV is equipped with an edge computing server and a wireless communication module to provide communication and computing services to end users; mobile terminals include smartphones and in-vehicle terminals, which are user devices with task processing needs; The base station is used to provide backhaul link support for UAVs; Uneven terrain, composed of surfaces at varying heights, can obstruct and cause path loss in communication links between UAVs and end users, as well as between UAVs and base stations, resulting in dynamic discontinuous accessibility of these links in space. Within the same time slot, each end user can only establish a communication connection with one UAV, while each UAV can serve multiple end users simultaneously. Specifically, the following steps are included: Step 1: The UAV flies within the target area and provides communication and computing services to the end user, while collecting terminal location, terrain elevation, channel status and mission information. Step 2: Based on the spatial geometric relationship between UAV and end user and between UAV and base station, determine the LoS / NLoS status of communication link, and construct a communication quality model that incorporates terrain occlusion effect by combining path loss model; jointly construct dynamic communicable domain by combining the communication quality constraints of access link and backhaul link, so as to transform the communication blind spot and link failure problem caused by occlusion in complex and uneven terrain environment into spatial communication constraints, thereby realizing the dynamic description and constraint expression of the communication reachable area in complex environment; Step 3: Construct a graph-based spatiotemporal communication topology model that integrates the interaction relationships between UAVs, end users, and base stations; Specifically, a graph convolutional network (GCN) is used to spatially model the communication topology, link states, and communicable domain characteristics to extract the dynamic communication relationships between different network entities. Furthermore, a temporal convolutional network (TCN) is combined to model the evolution of communication states over time, enabling the prediction of future communication states and communication interruption risks. Step 4: Within each decision time slot, the UAV constructs a two-time-scale collaborative optimization mechanism based on the current system state, communication topology state, and communicable domain information. On the slow time scale, based on the prediction results of future communication states and the evolution trend of communicable domains, the MPC method is used to perform rolling optimization of the UAV's three-dimensional flight trajectory, enabling the UAV to avoid potential communication interruption areas in advance and maintain long-term communication continuity. On the fast time scale, combined with the real-time channel state, task queue state, and computing resource state, an online resource scheduling method based on Lyapunov optimization is used to dynamically optimize the task offloading ratio, bandwidth resources, and computing resources, thereby achieving collaborative optimization between trajectory planning and task offloading processes under communication continuity constraints.

[0007] As a further preferred embodiment of the UAV trajectory planning and task unloading method for uneven terrain according to the present invention, in step 1, the environment is modeled using a three-dimensional Cartesian coordinate system, the target area is discretized into a two-dimensional grid map, and corresponding terrain elevation values ​​are assigned; the UAV performs trajectory planning in three-dimensional space, and the mobile terminals are distributed in the ground area; considering the short time slot interval, it is assumed that the position state of each network entity remains unchanged within a single time slot.

[0008] As a further preferred embodiment of the UAV trajectory planning and task unloading method for uneven terrain according to the present invention, in step 2, the communication model comprehensively considers the terrain occlusion effect, LoS / NLoS propagation state and path loss characteristics, and describes the link communication state through communication quality indicators. As a further preferred embodiment of the UAV trajectory planning and task unloading method for uneven terrain according to the present invention, in step 2, the communicable domain is obtained by jointly determining the communication quality constraints of the access link and the backhaul link, which is used to constrain the communication relationship between the UAV and the terminal user and the base station, and to characterize the spatial continuous reachability of the communication link in complex terrain environment.

[0009] As a further preferred embodiment of the UAV trajectory planning and task unloading method for uneven terrain according to the present invention, in step 3, the collaborative optimization strategy includes: under the premise of satisfying the communication domain constraints, jointly adjusting the UAV flight trajectory, task unloading ratio and communication resource allocation, so that the UAV maintains a stable communication connection and reduces the probability of communication interruption during flight, while improving task processing efficiency, resource utilization efficiency and system operation stability.

[0010] Compared with the prior art, the present invention, employing the above technical solution, has the following technical effects: 1. This invention proposes a UAV trajectory planning and task unloading method for uneven terrain. This method constructs a communication model that integrates terrain elevation information, spatial geometric relationships and LoS / NLoS propagation characteristics, explicitly models the impact of terrain occlusion on the communication link, and further transforms it into dynamic communicable domain constraints, thereby achieving collaborative optimization between UAV trajectory planning and task unloading processes, thus improving the communication reliability and computing service performance of the system in complex environments. 2. Compared with existing technologies, this invention fully considers the impact of uneven terrain on the obstruction and attenuation of communication links between UAVs and end users, as well as between UAVs and base stations. By constructing a dynamic communicable domain, it transforms the traditional communication reachability determination based on empirical channel models into a spatial communication constraint expression, improving the accuracy and constraint description capability of communication modeling in complex terrain environments. At the same time, this invention constructs a graph spatiotemporal communication topology model that integrates the dynamic interaction relationships between UAVs, end users, and base stations. It uses GCN to perform spatial correlation modeling of communication topology, link status, and communicable domain characteristics, and TCN to model the evolution of communication status over time. This enables the system to effectively characterize the dynamic connectivity relationships and communication status evolution process in complex environments, thereby improving the system's state perception capability and communication state prediction capability. 3. This invention constructs a two-timescale collaborative optimization mechanism constrained by the communicable domain. On the slow timescale, the MPC method is used to perform rolling optimization of the UAV's three-dimensional trajectory. On the fast timescale, an online resource scheduling method based on Lyapunov optimization is used to dynamically optimize the task offloading ratio, bandwidth resources, and computing resources, thereby achieving collaborative optimization between trajectory planning, communication resource allocation, and task offloading processes. Compared with existing methods, this invention can effectively improve the communication performance, task processing efficiency, and operational stability of the system in complex and uneven terrain environments while ensuring the feasibility of communication constraints. Attached Figure Description

[0011] Figure 1 This is a schematic diagram illustrating an application scenario of the UAV trajectory planning and task unloading method for uneven terrain according to the present invention. Figure 2 This is a flowchart of dual-timescale collaborative optimization based on graph spatiotemporal prediction. Figure 3 This is a flowchart of a method for UAV trajectory planning and task unloading for uneven terrain according to the present invention; Figure 4This paper compares the communication interruption probability of the CC-MPC-Lyapunov algorithm proposed in this invention with other baseline methods (Greedy Offloading + Shortest Path, Connectivity-unaware Planning, Traditional MPC, and Lyapunov-only Optimization) in complex and uneven terrain environments. Detailed Implementation

[0012] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings: The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention. The present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments. The purpose and effects of the present invention will become clearer. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0013] like Figures 1 to 4 As shown, this invention provides a method for UAV trajectory planning and task unloading in uneven terrain. This method constructs a communication model that integrates terrain elevation information, spatial geometric relationships, and line-of-sight (LoS) / non-line-of-sight (NLoS) propagation characteristics. It jointly models the communication link status and channel quality between the UAV and the end user, and between the UAV and the base station. Based on the communication quality constraints of the access link and backhaul link, it defines a dynamic communicable domain. This transforms the link interruption and communication blind spot problems caused by terrain occlusion in uneven terrain environments into spatial communication constraints, enabling a dynamic description of communication reachability in complex environments.

[0014] Building upon this foundation, this invention further constructs a graph-based spatiotemporal communication state model that integrates the dynamic interactions between UAVs, end users, and base stations. By jointly modeling the communication topology, link states, and the evolution of communicable domains, it achieves spatiotemporal characteristic representation of communication states and prediction of future communication states in complex dynamic network environments. Specifically, a graph neural network is used to encode the network topology and communicable domain features to extract dynamic relationships and spatial connectivity features between different network entities, thereby enhancing the system's perception and state representation capabilities of dynamic communication topology changes. Furthermore, this invention constructs a two-timescale collaborative optimization framework constrained by the communicable domain. At the slow timescale, Model Predictive Control (MPC) is used to perform rolling optimization of the UAV's three-dimensional flight trajectory, enabling the UAV to avoid potential future communication interruption areas in advance, thus maintaining long-term communication continuity. At the fast timescale, combining real-time channel states, task queue states, and computing resource states, an online resource scheduling method based on Lyapunov optimization is used to dynamically optimize the task offloading ratio, bandwidth resources, and computing resources to reduce system task latency and improve resource utilization efficiency. Through the above method, the present invention achieves coordinated optimization between UAV trajectory control, communication resource allocation and task unloading process in complex and uneven terrain environments. While ensuring the feasibility of communication constraints, it effectively improves the system's communication reliability, task processing efficiency and operational stability.

[0015] A method for UAV trajectory planning and task offloading based on communicable domain modeling includes a UAV-assisted mobile edge computing system, which includes a UAV, a mobile terminal, a base station, and an uneven terrain environment.

[0016] In this environment, UAVs are equipped with edge computing servers and wireless communication modules to provide communication and computing services to end users. Mobile terminals include smartphones, vehicle-mounted terminals, and other user devices with task processing needs. Base stations provide backhaul link support for UAVs. Uneven terrain, consisting of surfaces at varying heights, can cause obstruction and path loss in communication links between UAVs and end users, as well as between UAVs and base stations, resulting in dynamically discontinuous reachability of these communication links. Within a single time slot, each end user establishes a communication connection with only one UAV, while each UAV can serve multiple end users simultaneously.

[0017] like Figure 3 As shown, the specific steps include: (1) The UAV flies within the target area and provides communication and computing services to the end user, while collecting terminal location, terrain elevation, channel status and mission information. Based on the spatial geometric relationship between the UAV and the end user and between the UAV and the base station, the LoS / NLoS status of the communication link is determined, and a communication quality model that incorporates terrain occlusion effect is constructed in combination with the path loss model. On this basis, the communication quality constraints of the access link and the backhaul link are combined to construct a dynamic communicable domain, which is used to transform the communication blind spots and link failures caused by occlusion in complex and uneven terrain environments into spatial communication constraints, thereby realizing the dynamic description and constraint expression of the communicable reachable area in complex environments.

[0018] (2) To address the issues of significant dynamic changes in communication links and strong time-varying network topology in complex terrain environments, a graph spatiotemporal communication topology model integrating the interaction relationships between UAVs, end users, and base stations is constructed. Specifically, a graph convolutional network (GCN) is used to spatially correlate the communication topology, link states, and communicable domain characteristics to extract dynamic communication relationships between different network entities. Furthermore, a temporal convolutional network (TCN) is combined to model the evolution of communication states over time, enabling the prediction of future communication states and communication interruption risks, thereby enhancing the system's ability to perceive the continuity of dynamic communication in complex environments and its spatiotemporal state representation capabilities.

[0019] (3) Within each decision time slot, the UAV constructs a two-time-scale collaborative optimization mechanism based on the current system state, communication topology state, and available communication domain information: On the slow time scale, based on the prediction results of future communication state and the evolution trend of available communication domain, the MPC method is used to perform rolling optimization of the UAV's three-dimensional flight trajectory, enabling the UAV to avoid potential communication interruption areas in advance and maintain long-term communication continuity; On the fast time scale, combined with the real-time channel state, task queue state, and computing resource state, the online resource scheduling method based on Lyapunov optimization is used to dynamically optimize the task offloading ratio, bandwidth resources, and computing resources, thereby realizing collaborative optimization between trajectory planning and task offloading process under communication continuity constraints.

[0020] Furthermore, in step (1), the environment is modeled using a three-dimensional Cartesian coordinate system, the target area is discretized into a two-dimensional grid map, and corresponding terrain elevation values ​​are assigned; the UAV performs trajectory planning in three-dimensional space, and the mobile terminals are distributed in the ground area. Considering the short time slot interval, it is assumed that the position state of each network entity remains unchanged within a single time slot.

[0021] Furthermore, the communication model in step (2) comprehensively considers the terrain shading effect, LoS / NLoS propagation state and path loss characteristics, and describes the link communication state through communication quality indicators; the communicable domain is obtained by jointly judging the communication quality constraints of the access link and the backhaul link, which is used to constrain the communication relationship between UAV and terminal users and base stations, and to characterize the spatial continuous reachability of communication links in complex terrain environments.

[0022] Furthermore, the collaborative optimization strategy in step (3) includes: under the premise of satisfying the constraints of the communicable domain, jointly adjusting the UAV flight trajectory, task offloading ratio and communication resource allocation, so that the UAV can maintain a stable communication connection and reduce the probability of communication interruption during flight, while improving task processing efficiency, resource utilization efficiency and system operation stability.

[0023] The flowchart of dual-timescale collaborative optimization based on graph spatiotemporal prediction is as follows: Figure 2 As shown, the CC-MPC-Lyapunov algorithm proposed in this invention compares the communication interruption probability with other baseline methods (Greedy Offloading + Shortest Path, Connectivity-unaware Planning, Traditional MPC, and Lyapunov-only Optimization) in complex and uneven terrain environments. Figure 4 As shown.

[0024] 1. System Model: like Figure 1 As shown, the UAV-assisted mobile edge computing system consists of a UAV equipped with an edge computing server, The system consists of one mobile terminal, one base station, and an uneven terrain environment. The total mission duration is assumed to be... Divide it into equal parts There are 1 equal-length time slot, each time slot having a length of 1. The location and state of each network entity in the system are modeled using a time-discrete approach. (In time slots...) The spatial location of a UAV is represented as , No. The spatial location of a mobile terminal is represented as follows: The fixed location of the base station is represented as Among them, uneven terrain is represented by a two-dimensional elevation field function. Description: The current horizontal projection position of the UAV is... Its corresponding terrain elevation is To ensure flight safety and trajectory feasibility, the UAV's flight altitude relative to the ground must meet the following requirements. ,in, and These are the minimum safe flight altitude and the maximum permissible flight altitude of the UAV relative to the ground, respectively.

[0025] The motion state of the UAV between adjacent time slots satisfies the following discrete kinematic model:

[0026] in, and These represent the UAV in the time slots. The velocity vector and acceleration vector. To ensure stability during flight, its motion must satisfy the following physical constraints:

[0027]

[0028]

[0029] in, and These represent the maximum permissible flight speed and maximum flight acceleration of the UAV, respectively. Regarding communication correlation, a binary variable is used. , indicating mobile terminal In the time slot Whether to establish a communication connection with the current UAV, when When, it indicates the end user The UAV provides communication and computing services; otherwise, no association is established. The system satisfies the single association constraint, meaning that each mobile terminal can only establish a connection with one UAV in the same time slot.

[0030]

[0031] Furthermore, considering that the UAV needs to interact with the base station via the backhaul link during the task unloading process, a binary variable is defined. Indicates UAV in time slot Whether to maintain a valid communication connection with the base station, when If the condition is met, it indicates that the current UAV meets the backhaul communication constraints and can perform task backhaul and edge collaborative computing; otherwise, the backhaul link is considered interrupted. Therefore, system communication reachability must simultaneously satisfy the constraints of the terminal access link and the base station backhaul link, i.e. At that time, the system considers the current task offloading link to be available. Considering the short duration of a single time slot, the positional state of the UAV and mobile terminal can be assumed to remain unchanged within each time slot, thus providing a discrete state basis for subsequent communication modeling and trajectory optimization.

[0032] The above kinematic model and physical constraints can accurately describe the three-dimensional spatial position change process of UAV in uneven terrain environments, providing basic support for subsequent communication domain modeling, trajectory optimization and task unloading decisions.

[0033] 2. Modeling of the communicable domain: In uneven terrain environments, communication links between UAVs and end users, as well as between UAVs and base stations, are affected by terrain undulations and obstructions, causing the links to dynamically switch between line-of-sight (LoS) and non-line-of-sight (NLoS) propagation states. Traditional communication models based solely on Euclidean distance attenuation are insufficient to accurately describe the spatial discontinuities of link reachability in complex terrain. Therefore, this invention constructs a dynamic communicable domain model that integrates terrain obstruction determination and probabilistic propagation characteristics to describe end-to-end communication constraints in complex environments.

[0034] Specifically, let UAV and end user be set up The spatial connection between them is The terrain elevation at any point along the propagation path is determined by a two-dimensional elevation field function. Given that the link's elevation along the propagation path is consistently higher than the corresponding terrain elevation, the current link is determined to be in a Loss of Space (LoS) state; otherwise, it is determined to be in an Non-LoS (NLOS) state. This is based on the link elevation angle. The probability of Loss is established as follows: The expected channel power gain between the UAV and the end user is expressed as:

[0035]

[0036] in, This indicates the additional attenuation factor for NLoS. This represents the channel gain at the reference distance. This represents the path loss index. This indicates the communication distance between the UAV and the end user.

[0037] Similarly, considering the impact of terrain on the backhaul link between the UAV and the base station, the channel gain between the UAV and the base station can be established:

[0038] in, This indicates the communication distance between the UAV and the base station.

[0039] During task unloading, the communication bandwidth of the access link and backhaul link is dynamically allocated proportionally based on the task load of each terminal. Then, the terminal user... The access link transmission rate between the UAV and the UAV, and the backhaul link transmission rate between the UAV and the base station are expressed as follows:

[0040]

[0041]

[0042] in, and These represent the dynamic bandwidth allocation results for the access link and the backhaul link, respectively. and These represent the transmit power of the end user and the UAV, respectively. This represents noise power. Considering that task offloading needs to simultaneously meet the communication requirements of the access link and the backhaul link, the end-to-end effective transmission rate is defined as:

[0043]

[0044] To ensure the reliability of task unloading, a minimum communication rate threshold is set. When satisfied At this point, the current UAV location is considered to satisfy the end-to-end communication constraints. Furthermore, the set of spatial locations satisfying the above communication constraints is defined as a dynamically communicable domain:

[0045]

[0046] The modeling method described above directly maps the impact of uneven terrain on communication links into end-to-end communication rate constraints, and further transforms this into spatial communication domain constraints. This allows the UAV to dynamically identify spatial regions that meet communication conditions during trajectory planning. When communication is interrupted due to terrain obstruction, the UAV can adjust its flight altitude or horizontal position to enter a new communication domain, thus restoring the communication link spatially and ensuring communication continuity and system service stability during mission unloading.

[0047] 3. Graph-based spatiotemporal state modeling: Due to the dynamic changes in communication links caused by UAV movement, end-user movement, and terrain occlusion in uneven terrain environments, the system communication topology exhibits significant spatiotemporal coupling characteristics. Traditional static graph models are insufficient to characterize the dynamic evolution of communication relationships over time in complex environments. Therefore, this invention constructs a graph spatiotemporal state model oriented towards the evolution of the communicable domain to describe the dynamic changes in network topology and communication states.

[0048] Specifically, in time slots Construct the system communication topology diagram:

[0049] Among them, the node set Includes UAV nodes, mobile terminal nodes, and base station nodes; edge set This represents the communication link relationship between nodes. When the communication domain constraint is satisfied between nodes, i.e., the link transmission rate satisfies... At that time, it is considered that the node With nodes There must be a communication edge between nodes; otherwise, the corresponding communication edge is disconnected. Therefore, the communication topology dynamically evolves with changes in UAV location, user movement, and terrain occlusion. Furthermore, to characterize the communication state and spatial attributes of each node, the node feature vector is defined as:

[0050]

[0051] in, Indicates the spatial location of the node. Indicates the motion state of the node. Indicates the status of the task queue. Indicates the current communication rate. This indicates the communicable domain status corresponding to the node.

[0052] In the spatial dimension, GCN is introduced to extract features from the communication topology graph, learning spatial relationships within the network by aggregating neighbor node information. The graph convolution process is represented as:

[0053]

[0054] in, This represents the dynamic adjacency matrix generated by the constraints of the communicable domain. Represents the degree of correspondence matrix. Indicates the first Layer node feature matrix Represents the network weight parameters. This represents a non-linear activation function.

[0055] Considering that communication topology evolves continuously over time, it is difficult to accurately predict future communication reachability based solely on the current state. Therefore, this invention further introduces TCN to analyze the historical topology state sequence. Temporal correlation modeling is performed to learn the changing trends of communication links and the evolution of the communicable domain, thereby predicting the system state in the future time domain. ,in, This indicates the length of the prediction time domain.

[0056] By using the above-mentioned spatiotemporal state modeling method, the system can jointly characterize the spatial correlation and temporal evolution characteristics of communication topology in complex terrain environments, and provide future communication state prediction information for subsequent UAV trajectory planning based on model predictive control, thereby improving the foresight of trajectory decision-making and communication stability.

[0057] 4. Two-timescale optimization mechanism: Because UAV trajectory changes and task unloading decisions differ significantly in time scale, unified joint optimization of the two would result in excessively high dimensionality of optimization variables and computational complexity, making it difficult to meet the real-time decision-making requirements of complex dynamic environments. Therefore, this invention constructs a two-time-scale optimization mechanism to collaboratively optimize UAV trajectory planning and task unloading decisions at different time scales.

[0058] UAV trajectory planning has a continuous impact on the long-term communication topology and the distribution of the communicable domain of the system. Its state changes are relatively slow, so it is optimized at a slow time scale. On the other hand, task offloading decisions are affected by instantaneous channel state, task arrivals and computational load, and their dynamic changes are more frequent. Therefore, they are adjusted in real time at a fast time scale.

[0059] At a slow time scale, this invention employs an MPC-based trajectory planning mechanism. Specifically, within each trajectory optimization cycle, the system predicts future trajectories based on a graph spatiotemporal state model. The evolution of the communication topology and the changing trend of the communicable domain within each time slot are analyzed, and the future trajectory of the UAV is optimized by rolling optimization within a finite prediction time domain. The optimization objective is expressed as:

[0060]

[0061] in, This represents the effective end-to-end transmission rate of the system within the future time slot. The optimization process must satisfy UAV kinematic constraints, flight altitude constraints, and dynamic communicable domain constraints. After each optimization, only the trajectory control action corresponding to the current time slot is executed, and prediction and optimization are re-performed in the next time slot based on the latest environmental conditions, thereby achieving rolling trajectory planning in complex dynamic environments. Through this mechanism, UAVs can anticipate future communication link changes and actively avoid communication interruption areas caused by terrain obstruction by adjusting flight altitude or horizontal position, improving the foresight of trajectory planning and communication stability.

[0062] On a fast timescale, this invention employs a task offloading mechanism optimized by Lyapunov to dynamically adjust the terminal task offloading ratio. Let... Indicates end user In the time slot If the task queue length is given, then the system queue state satisfies:

[0063]

[0064] in, Indicates the amount of newly arrived tasks. This indicates the amount of data that has been unloaded in the current time slot.

[0065] To simultaneously ensure system queue stability and task processing performance, Lyapunov functions are introduced. And construct a drift-penalty optimization objective. ,in, This represents the Lyapunov drift term. Represents the system utility function. This represents the trade-off parameter between system performance and queue stability. Based on the current channel state, task queue state, and communication resource allocation, the system calculates the optimal task offloading ratio in real time, thereby achieving online dynamic adjustment of task offloading decisions.

[0066] Through the above two-timescale optimization mechanism, the system can achieve coordinated optimization of UAV trajectory planning and task unloading strategy while ensuring that communication constraints are met in complex terrain environments, thereby improving system communication stability, task processing efficiency and overall service performance.

[0067] It will be understood by those skilled in the art that the above descriptions are merely preferred examples of the invention and are not intended to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, those skilled in the art can still modify the technical solutions described in the foregoing examples or make equivalent substitutions for some of the technical features. All modifications and equivalent substitutions made within the spirit and principles of the invention should be included within the scope of protection of the invention. All technical features in this embodiment can be freely combined according to actual needs.

[0068] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for UAV trajectory planning and task unloading for uneven terrain, characterized in that: The system includes a UAV-assisted mobile edge computing system, which comprises a UAV, a mobile terminal, a base station, and an uneven terrain environment. Among them, UAV is equipped with an edge computing server and a wireless communication module to provide communication and computing services to end users; mobile terminals include smartphones and in-vehicle terminals, which are user devices with task processing needs; The base station is used to provide backhaul link support for UAVs; Uneven terrain, composed of surfaces at varying heights, can obstruct and cause path loss in communication links between UAVs and end users, as well as between UAVs and base stations, resulting in dynamic discontinuous accessibility of these links in space. Within the same time slot, each end user can only establish a communication connection with one UAV, while each UAV can serve multiple end users simultaneously. Specifically, the following steps are included: Step 1: The UAV flies within the target area and provides communication and computing services to the end user, while collecting terminal location, terrain elevation, channel status and mission information. Step 2: Based on the spatial geometric relationship between UAV and end user and between UAV and base station, determine the LoS / NLoS status of communication link, and construct a communication quality model that incorporates terrain occlusion effect by combining path loss model; jointly construct dynamic communicable domain by combining the communication quality constraints of access link and backhaul link, so as to transform the communication blind spot and link failure problem caused by occlusion in complex and uneven terrain environment into spatial communication constraints, thereby realizing the dynamic description and constraint expression of the communication reachable area in complex environment; Step 3: Construct a graph-based spatiotemporal communication topology model that integrates the interaction relationships between UAVs, end users, and base stations; Specifically, a graph convolutional network (GCN) is used to spatially model the communication topology, link states, and communicable domain characteristics to extract the dynamic communication relationships between different network entities. Furthermore, a temporal convolutional network (TCN) is combined to model the evolution of communication states over time, enabling the prediction of future communication states and communication interruption risks. Step 4: Within each decision time slot, the UAV constructs a two-time-scale collaborative optimization mechanism based on the current system state, communication topology state, and communicable domain information. On the slow time scale, based on the prediction results of future communication states and the evolution trend of communicable domains, the MPC method is used to perform rolling optimization of the UAV's three-dimensional flight trajectory, enabling the UAV to avoid potential communication interruption areas in advance and maintain long-term communication continuity. On the fast time scale, combined with the real-time channel state, task queue state, and computing resource state, an online resource scheduling method based on Lyapunov optimization is used to dynamically optimize the task offloading ratio, bandwidth resources, and computing resources, thereby achieving collaborative optimization between trajectory planning and task offloading processes under communication continuity constraints.

2. The method for UAV trajectory planning and task unloading for uneven terrain according to claim 1, characterized in that: In step 1, the environment is modeled using a three-dimensional Cartesian coordinate system, the target area is discretized into a two-dimensional grid map, and corresponding terrain elevation values ​​are assigned; the UAV performs trajectory planning in three-dimensional space, and the mobile terminals are distributed in the ground area; considering the short time slot interval, it is assumed that the position state of each network entity remains unchanged within a single time slot.

3. The method for UAV trajectory planning and task unloading for uneven terrain according to claim 1, characterized in that: In step 2, the communication model comprehensively considers terrain occlusion effects, LosS / NLoS propagation status, and path loss characteristics, and describes the link communication status through communication quality indicators.

4. The method for UAV trajectory planning and task unloading for uneven terrain according to claim 1, characterized in that: In step 2, the communicable domain is obtained by jointly determining the communication quality constraints of the access link and the backhaul link. It is used to constrain the communication relationship between the UAV, the end user, and the base station, and to characterize the spatial continuity and reachability of the communication link in complex terrain environments.

5. The method for UAV trajectory planning and task unloading for uneven terrain according to claim 1, characterized in that: In step 3, the collaborative optimization strategy includes: under the premise of satisfying the constraints of the communicable domain, jointly adjusting the UAV flight trajectory, task offloading ratio and communication resource allocation, so that the UAV can maintain a stable communication connection and reduce the probability of communication interruption during flight, while improving task processing efficiency, resource utilization efficiency and system operation stability.

Images