A decentralized unmanned aerial vehicle task allocation method and system for low-altitude economy

Abstract

The application discloses a kind of decentralized unmanned aerial vehicle task allocation methods and systems for low-altitude economy, the method is first smart contract to UAV registration, and start anti-witch attack mechanism, publisher uploads task information and pays deposit, and smart contract verifies task, and carries out delay decision.Secondly, UAV confuses and then encrypts own geographic position coordinates, and the task publisher clusters encrypted coordinates to obtain encrypted clustering data.Finally, according to the encrypted clustering data, the probability of each UAV reaching the task location is calculated by the smart contract, the matching degree of the UAV is evaluated by the smart contract, and the candidate UAV set is screened out;According to the delay decision result, the smart contract allocates tasks to the candidate UAV set.The application considers the motion state of task and UAV, ensures the rationality and executability of task allocation, effectively enhances the accuracy and real-time performance of task matching.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of task allocation, and particularly relates to a decentralized unmanned aerial vehicle task allocation method and system for low-altitude economy. BACKGROUND

[0002] Task allocation and scheduling is a technology that allocates tasks to available execution units and completes specified operations by them through reasonable planning and resource allocation. Based on this, the task allocation and scheduling technology under space-time constraints is developed, which requires the execution unit to go to a specific location within a specific time to complete a specified task, such as typical application scenarios such as online car-hailing services and food delivery. With the continuous development and application of space-time crowdsourcing technology and the expansion of application scenarios, under the promotion of low-altitude economy (including unmanned aerial vehicle logistics) and UAV (unmanned aerial vehicle) technology, the task execution subject has expanded from traditional ground workers to UAV clusters, forming a heterogeneous task execution network of "air-ground" cooperation.

[0003] In the actual task execution process, there is often a demand for multiple UAVs to cooperate to complete complex tasks, or the task requires UAVs to have specific skills (carry different equipment or perform specific flight area tasks) to ensure smooth completion. This problem is called the multi-capability constrained UAV cooperative scheduling problem. Among them, the dynamic method takes into account the complexity of the real environment, and the generation of tasks is often dynamic and unpredictable, while the state of UAVs (including geographic location, availability and skills) also changes over time.

[0004] In the architecture of the unmanned aerial vehicle scheduling and allocation system, the real-time changes of tasks and UAVs in the dynamic environment put higher requirements on the response speed of the system, and the traditional centralized platform is difficult to effectively cope with the airspace scheduling of the dynamic UAV cluster due to the single-point coordination bottleneck, further exposing its reliability defects: the traditional centralized scheduling platform usually relies on the centralized platform as the coordination core, but once the central platform fails, the entire scheduling and allocation system may be paralyzed. In addition, the central platform lacks transparency and decentralized supervision mechanism in the evaluation of tasks and UAVs, and the evaluation results are therefore easy to be manipulated, making it difficult to guarantee its objectivity and accuracy. In terms of task allocation method, the current dynamic scheduling and allocation algorithm excessively focuses on the cost optimization index, and lacks comprehensive consideration of task completion quality in dynamic scenarios, lacking a multi-dimensional task completion quality evaluation system.

[0005] As a decentralized distributed ledger technology, blockchain is expected to effectively overcome the drawbacks of centralized systems by deploying the dispatching and allocation system based on smart contracts on the blockchain. Through distributed storage and decentralized consensus mechanism, blockchain avoids the single point failure problem that may occur on traditional platforms; at the same time, the smart contract can record information on the chain to ensure the authenticity and non-tamperability of data, thereby realizing the transparency and fairness of the evaluation process. Although blockchain technology provides a potential solution to the above problems, its application still needs to overcome the following obstacles, the first of which is cost. The gas cost of smart contract is positively correlated with the computational complexity, and the high-frequency task allocation request in dynamic dispatching and allocation may lead to exponential growth of gas cost. In addition, the high transparency of blockchain may also lead to the risk of sensitive information leakage, and in the low-altitude economic scenario, the security protection of the real-time location, task data and other privacy information of UAV is particularly important.

[0006] In summary, the current dynamic dispatching and allocation system is facing the problems of insufficient task completion quality, weak privacy protection and high blockchain cost, and it is urgent to explore an efficient and feasible solution. Especially in the background of the rapid development of low-altitude economy and UAV technology, the system needs to optimize the cooperative task allocation of UAVs, improve the task completion quality in dynamic environment, and balance between privacy protection and system performance, to ensure that the dispatching and allocation system has safety, reliability and good scalability. SUMMARY

[0007] The present application is directed to the deficiencies of the prior art, and proposes a decentralized UAV task allocation method and system for low-altitude economy. In the scenario where both UAVs and tasks are in dynamic change, this method not only ensures that each task meets the corresponding skill requirements, but also fully considers the dynamic movement characteristics of UAVs to achieve more accurate and efficient task allocation. At the same time, the present application introduces an innovative privacy protection scheme, effectively protecting the location privacy of UAVs, reducing the risk of privacy leakage while minimizing the task allocation deviation caused by the privacy protection mechanism, thereby improving the task quality while ensuring the fairness and reliability of the system.

[0008] In the technical solution of the application, when a task arrives, the task publisher first generates a pair of keys and sends the public key to all UAVs in the ready state. After receiving the public key, the UAV performs multiple obfuscation processing on its geographic location information, and then encrypts and returns it to the task publisher. Subsequently, the task publisher uses a homomorphic encryption algorithm to perform encrypted clustering calculation on the task location and UAV trajectory information to generate the clustering center coordinates and the trajectory distribution of each clustering area in the encrypted state. The task publisher then decrypts the clustering center coordinates to obtain the real clustering center location and the UAV weight of each clustering area, and sends this information to the smart contract for predicting the probability of each UAV reaching the task location. Then, the smart contract selects the optimal UAV set based on the allocation algorithm and feeds back the allocation result to the task publisher and the relevant UAVs, finally assigning the corresponding UAV to the task location to perform the task. After the UAV completes the task, the system will give the corresponding final reward according to its behavior performance.

[0009] The application adopts the following technical solution: a decentralized UAV task allocation method for low-altitude economy, comprising the following steps:

[0010] (1) Before the process starts, the smart contract registers the UAV. The UAV needs to upload its non-sensitive information to the smart contract. To reduce the occurrence of dishonest behavior and prevent some malicious UAVs from possibly forging multiple identities to cheat rewards (i.e., Sybil attack), in addition to requiring the UAV to transfer a certain amount of deposit to the smart contract, the smart contract should also start an anti-Sybil attack mechanism to ensure that each UAV identity is real and unique.

[0011] (2) When a task arrives, the publisher uploads the task information to the smart contract and pays the deposit. The smart contract will verify the task to ensure that it meets the relevant compliance requirements. Then, based on the candidate UAV set, the smart contract dynamically determines when to allocate UAVs for the task according to the actual start time of the task, i.e., makes a task delay decision. The application determines when each task is allocated to a UAV through the task delay decision mechanism, achieving dynamic optimization of task scheduling.

[0012] (3) The registered UAV first performs multiple obfuscation on its geographic location coordinates and returns the encrypted coordinates to the task publisher.

[0013] (4) The task publisher uses an encrypted clustering algorithm to perform clustering calculation on the encrypted coordinates in step (3) to obtain the clustering center coordinates in the encrypted state and the UAV weight of each clustering area as encrypted clustering data, and sends these data to the smart contract.

[0014] (5) The smart contract calculates the probability of each UAV reaching the task location based on the encrypted clustering data obtained in step (4).

[0015] (6) The smart contract combines the probability of the UAV reaching the task location, the skill coverage rate, and the probability of completing the task within the specified time calculated in step (5) to comprehensively evaluate the matching degree of each UAV and selects a candidate UAV set that best meets the task requirements.

[0016] (7) According to the delay decision result, at the task allocation decision time point, the task is allocated, and the result is notified to the UAV. After the UAV completes the task, the system will perform reward settlement and distribution.

[0017] Further, in step (1), the UAV needs to provide non-sensitive information, mainly including: its own skill set, wallet address, and the time required by the UAV to complete the task according to the skills it possesses. In addition, the historical moving track of the UAV is sensitive information and does not need to be uploaded at this time.

[0018] In step (1), the smart contract effectively resists Sybil attacks by combining Proof of Location (PoL) and Zero-Knowledge Proof (ZKP). Specifically, the UAV must submit encrypted location snapshots of consecutive timestamps when registering to ensure the continuity and authenticity of its physical location. At the same time, the smart contract randomly selects three registered UAVs as verifiers to conduct distributed verification of the submitted location data, thereby enhancing the decentralized trust of the system. In terms of privacy protection, the UAV uses ZKP algorithm to generate location proof, proving only that it is within the specified geographic fence range to the smart contract without revealing sensitive information such as actual coordinates.

[0019] Further, in step (2), the task information includes: task location information, required skills, task start and end time, task generation public key, this task reward, and UAV basic salary.

[0020] Further, in step (2), the smart contract fully considers the impact of the task allocation decision time point when making task delay decisions. Based on the task start and end time, combined with the minimum remaining time for the UAV to move before the task starts, the longest time that can be effectively predicted, and the average time for historical completion of such tasks, the smart contract dynamically sets different allocation decision time points for each task.

[0021] Further, in step (3), the UAV needs to perform two times of location coordinate confusion. The first confusion uses polar coordinate differential privacy confusion, and the second confusion maps the differential privacy processed coordinates to the center coordinates of the corresponding grid according to the predefined grid range and grid width, thereby achieving further confusion of the location.

[0022] Further, in the step (3), the UAV needs to use the public key provided by the task publisher in step (2) to perform Paillier homomorphic encryption on the obfuscated location information to further enhance the strength of privacy protection and increase the difficulty of attack.

[0023] Further, in the step (4), the task publisher uses an encrypted k-means clustering algorithm to divide the trajectory data of the UAV into several clustering regions and calculate the weight of the UAV in each region. In order to reduce the consumption of smart contract Gas, the clustering operation is performed off-chain. The task publisher decrypts the result and transmits it to the smart contract.

[0024] Further, in the step (5), the smart contract predicts the probability of the UAV going to different regions in the future by analyzing the historical moving trajectory of the single UAV. Specifically, the region to which the task belongs is determined according to the weight data of the UAV in each region, the meta-path correlation and the clustering result; then all feasible paths of the UAV from the current location to the target region are found; the correlation of each path is calculated, and finally the probability of the UAV reaching the task location is calculated based on these results.

[0025] Further, in the step (6), the smart contract assigns the task, and before that, the UAV is preliminarily screened: first, it is judged whether the time required for the UAV to complete the task meets the following conditions:

[0026] The estimated completion time is less than (the specified completion time of the task - the current time - the shortest time required for the UAV to move). If the expected duration is greater than the above time, the UAV is excluded. Next, further screening is performed: the skill coverage of the UAV is calculated by the number of skills possessed by the UAV and the number of skills required by the task; the probability of completing the task within a specified time is calculated based on the estimated completion time uploaded by the UAV. Finally, the smart contract comprehensively evaluates the probability of the UAV completing the task within a specified time, and selects the set of UAVs with the highest utility based on this.

[0027] Further, in the step (7), the smart contract assigns the task according to the decision time point of the assignment, and after the task assignment is completed, the system notifies the UAV of the result of the task assignment. After the UAV completes the task, it uploads the current time, and the system compares the current time with the end time required by the task, and determines the final remuneration of each UAV according to the comparison result and the number of UAVs.

[0028] The application also provides a decentralized UAV task assignment system for low-altitude economy, which comprises the following modules:

[0029] The registration in the task publishing module is used for the UAV to be registered by the smart contract, and an anti-witch attack mechanism is started, the task information is uploaded to the smart contract by the publisher and a deposit is paid, the smart contract verifies the task, and a delay decision is made.

[0030] The coordinate encryption confusion module is used for the UAV after registration to confuse and encrypt the geographic position coordinates multiple times, and the task publisher uses an encryption clustering algorithm to cluster the encrypted coordinates, obtains the clustering center coordinates in the encrypted state and the UAV weight of each clustering area as encrypted clustering data.

[0031] The task allocation module is used for the smart contract to calculate the probability of each UAV reaching the task site according to the encrypted clustering data, the smart contract to evaluate the matching degree of the UAV based on the probability of the UAV reaching the task site, and the candidate UAV set is screened out; and the smart contract to allocate tasks to the candidate UAV set according to the delay decision result.

[0032] Compared with the prior art, the beneficial technical effects of the present application are that:

[0033] 1. Different from the technical route of traditional static scene modeling and single-dimensional dynamic analysis, the present application innovatively constructs a space-time coupling model of double dynamic elements of tasks and workers, fully considers the motion state of tasks and UAVs, introduces a task delay decision strategy based on time constraints, accurately judges the legitimacy of dynamic arrival tasks, and ensures the rationality and executability of task allocation. At the same time, the present application aims at the dynamic moving characteristics of workers, constructs a meta-path analysis model based on space-time association rules by mining the moving mode features in historical task data, predicts the probability of UAVs reaching the task site, and thus selects the most suitable UAVs to execute the task. This method effectively enhances the accuracy and real-time performance of task matching, and provides a more optimal scheduling scheme for the scheduling and allocation system.

[0034] 2. The technical solution is aimed at the privacy protection requirements of the decentralized scheduling and allocation system, and a hierarchical dynamic privacy protection system is constructed. The present application introduces differential privacy technology, dynamically injects noise into the worker position within the required accuracy range of task matching, generates perturbed coordinates that meet the differential privacy requirements, and realizes the absolute position confusion under the premise of geographical correlation. Further, the dynamic grid division technology is used to map the continuous geographic space into discrete grid units with adaptive service radius, and the UAV position is mapped to the appropriate grid center for secondary confusion. The actual position of the UAV is effectively hidden by confusion and encryption clustering, and the data exposure risk is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0036] Figure 1 This is a schematic diagram of the process of the present invention;

[0037] Figure 2 This is a timing diagram of the model of the present invention;

[0038] Figure 3 This is a schematic diagram of the grid mapping in the position confusion process of the present invention. The present invention adopts a gridded position confusion mechanism to divide the target area into several uniform grids. Detailed Implementation

[0039] The specific embodiments of the present invention will be described below with reference to the accompanying drawings and examples. However, the following examples are only used to illustrate the present invention in detail and do not limit the scope of the present invention in any way.

[0040] This invention discloses a decentralized unmanned aerial vehicle (UAV) task allocation method and system for the low-altitude economy. Figure 1 , Figure 2 The system architecture and method flow of the present invention are given respectively. The present invention will now be described in detail with reference to Embodiment 1, specifically including the following steps:

[0041] Step 1: The system registers the UAV. Internally, the UAV is represented as follows:

[0042]

[0043] in It is the unique identifier of UAV. This is the UAV's wallet address. The UAV can use this wallet address to transfer funds to the smart contract or receive rewards. It is a skill set, representing the skill points mastered by the UAV, used to match tasks that can be completed. It is the estimated completion time for a UAV to fulfill its own skill requirements.

[0044] Proof-of-location is used to combat Sybil attacks. When a new UAV registers, a set of encrypted location snapshots with consecutive timestamps is submitted to the smart contract, generated as follows: First, the UAV uses a mobile terminal device to collect GPS coordinates at fixed time intervals of 10 seconds, and then... (The sentence is incomplete and requires further context to translate accurately.) Applying salted hashing results in ,in For a random number, The timestamp of the upload time. Given the GPS horizontal accuracy value, an irreversible position proof sequence is finally generated.

[0045]

[0046] in This means at least 4 minutes, increasing the cost of counterfeiting. Uploaded along with non-sensitive data from UAV. Then, the smart contract randomly selects at least three existing UAV nodes as validators. The verification logic includes verification... The time interval between adjacent hashes is kept constant, and forged tracks are excluded by a preset movement speed threshold. After successful verification, the verifier node signs the result, and the smart contract confirms the identity by collecting ≥2 / 3 of the signatures.

[0047] Step 2: The task publisher waits for the task to be delivered, uploads the task to the smart contract, and makes a delay decision on the task.

[0048] For task set any element in It is expressed as follows:

[0049]

[0050]

[0051] Where 𝑎dd𝑟𝑒𝑠𝑠𝑖 represents the wallet address for the task. and The latitude and longitude of the mission location. These are the skills required for the task. and These are the start and end times of the task. This represents the rewards that can be distributed for this task. Represents base salary, This is the public key generated for the task. Since each UAV should receive at least a base salary, and each task can be completed by at most a number of UAVs with skills related to the task, this invention requires that the reward amount for each task must be greater than or equal to the product of the number of skills and the base salary.

[0052] The smart contract will then verify the task, ensuring it complies with relevant compliance requirements and determining when to allocate a UAV (User Available) to it—a process known as task delay decision-making. Before this, the task decision-making process requires calculating the following parameters: This is the minimum remaining time for the UAV to move before the mission begins; The longest duration for which effective prediction can be made. The average time taken to complete such tasks in history. The calculation formula is as follows:

[0053]

[0054]

[0055]

[0056] where is a set of pairs of consecutively visited regions in the historical trajectory of the UAV, is the arithmetic mean of the multiple historical spatio-temporal movement time periods, and represents the time of consecutively visiting two regions and the time of consecutively visiting three regions, is the average time of completing such tasks in history.

[0057] The specific process of the task delay decision is as follows: receiving the minimum deposit required for the transfer , and the set of arrived tasks as input. For each element , first check whether it is assigned, if not assigned, check whether the deposit amount provided by the publisher is greater than , and after confirmation, proceed with the transfer. Then judge whether the task meets the standard of delay decision: for the current time , if it meets (the time of starting the task the time of arriving at the task location predicted by the UAV current time), and meets (the time of ending the task the minimum remaining time of UAV movement the average time of completing such tasks in history), it is determined that the task is allowed to enter, and the task is added to the result task set , and finally

[0058] .

[0059] Step 3, coordinate confusion

[0060]

[0061]

[0062] resulting in:

[0063]

[0064]

[0065] wherein , is the coordinate that the UAV needs to perform differential privacy, is the base point for polar coordinate conversion. and is the Laplace noise for polar radius and polar angle. The formula first calculates the polar radius and the polar angle , and adds and to obtain , is the coordinate after differential privacy. The present application adds a large amount of noise to the polar angle and ensures that the noise of the relevant position polar angle is consistent, adds a small amount of noise to the polar radius, and the noise is added independently and randomly to ensure that the impact on subsequent clustering is reduced as much as possible.

[0066] Step 3.2: Perform secondary confusion on the coordinates after differential privacy , . The UAV maps the coordinates after differential privacy to the center coordinates of the corresponding grid according to the predefined grid range and grid width. Specifically, given the geographical range and the grid width , for any point , the following conversion is performed:

[0067]

[0068]

[0069] to obtain:

[0070]

[0071]

[0072] where and are the grid indexes in the longitude and latitude directions respectively, is the corresponding grid center coordinate, which is taken as the final confused coordinate.

[0073] As shown in Figure 3 , an example area is divided into four grids, labeled , , and respectively. In the grid mapping process, any position point falling within a certain grid is mapped to the center point of the grid. For example, the original position point A (coordinates ) is located in , and after mapping, the confused position of A is the center point of Similarly, position points B, C, and the like also follow the same mapping rules, and finally correspond to the center positions of their respective grids.

[0074] Finally, the obfuscated position information is Paillier homomorphically encrypted using the public key provided by the task publisher in step (2), and the encrypted position information is used as the final transmission data.

[0075] Step 4: The task publisher uses the encrypted clustering to calculate the encrypted position information of the UAV obtained in step 3 to obtain the final clustering center and the weight of the UAV in each region. The specific algorithm is: taking the UAV set and the obfuscated trajectory of the UAV as input, where is the set of obfuscated coordinates of the UAV in step (3), first add all the obfuscated positions of the UAV to the encrypted coordinate set , then randomly select encrypted coordinates as initial clustering centers , and enter the loop, resetting the weight of each UAV in each clustering region and the total number of points in each clustering region each time the loop is executed. For each candidate , calculate the square of the encrypted distance of point to each clustering center, and assign point to the nearest cluster, and update and if the cluster assignment of point changes, and exit the loop; otherwise, for all , calculate the encrypted sum of all points in the cluster, calculate the average of the encrypted sum and update the clustering center, and continue to execute the loop. After the loop exits, decrypt each clustering center, output its plaintext coordinates, and store them in , and finally output the weight of each group in each cluster and .

[0076] Step 5: The smart contract calculates the probability of each UAV reaching the task location using the meta-path prediction method based on the encrypted clustering data obtained. Define is a set of nodes, is a set of edges, and the meta-path is the sequence of paths of the UAV between nodes, represented as:

[0077]

[0078] where, For To a node type path pattern. According to the three-degree influence theory, the invention infers that when the meta-path length exceeds 3, the correlation between the beginning and the end is significantly weakened. Therefore, the invention mainly focuses on two types of key meta-paths: and . Among them, represents the meta-path of the user set , represents the meta-path of the user set .

[0079] First, calculate the probability of transition between nodes , the formula is as follows:

[0080]

[0081] Where is a set containing nodes connected to node and of the same type as node , is the weight based on frequency calculation from to . The invention defines as the correlation between the first node and the tail node in a path based on transition probability :

[0082]

[0083] Where is the shortest time from the first node to the tail node in the path . is a custom time threshold for limiting the time of user movement between two areas. is a specific instance of the meta-path in , containing all possible paths in a given network environment. The eigenvalue of the meta-path can be represented as the sum of the correlation of all path instances in the set :

[0084]

[0085] According to the meta-path set , the probability of the UAV to complete the task can be regarded as the correlation between them, as follows:

[0086]

[0087] wherein is a hyperparameter for adjusting the impact factor.

[0088] Step 6: The smart contract combines the obtained probability of the UAV reaching the task location to perform task allocation. The task allocation algorithm is as follows: obtain the task , the UAV , the probability of the UAV reaching the task location as input. For each , there is a number of skills required to complete the task and a skill set , when is not empty, enter the loop, for any UAV , calculate the following three parameters: the skill coverage rate of

[0089] ,

[0090] wherein is the skill set held by the UAV; the probability of completing the task , obtained by normalizing the estimated completion time of the UAV, wherein N is the number of UAVs; and the final comprehensive selection UAV probability formula is:

[0091]

[0092] wherein , , are the weights of different constituent elements, , , are the exponents of the elements, is the normalization exponent. The overall formula adopts a weight tuning and exponent plasticity strategy, dynamically adjusts the priority of different objectives by adjusting the weights, and changes the data distribution by adjusting the exponents to make the excellent ones more prominent.

[0093] Select the highest UAV in the traversal round as the candidate object, and loop until the skills of the UAVs in the candidate set completely cover .

[0094] ​Step 7: After the task is completed, the system will carry out reward settlement and distribution. The UAV will upload the current time, if the current time exceeds the end time required by the task, it means that the task cannot be completed on time, and the UAV will only get the basic base salary; if the current time is before the end time required by the task, it means that the UAV has successfully completed the task, in addition to the basic salary, the remaining part of the total reward declared by the task publisher after deducting the base salary of all UAVs will be distributed according to the utility ratio of the UAV. The calculation formula of the reward is:

[0095]

[0096]

[0097] Wherein is the basic base salary, is the actual end time of the task, is the scheduled end time of the task, is the remaining part of the total reward after deducting the base salary of all UAVs, is the number of UAVs participating in the current task, is the total utility of completing the task at the current moment, is the ratio of utility.

[0098] The application also provides a decentralized UAV task allocation system for low-altitude economy, comprising the following modules:

[0099] The registration in the task publishing module is used for the registration of UAVs by the smart contract, and the anti-malicious attack mechanism is started, the task information is uploaded to the smart contract by the publisher, and the deposit is paid, the task is verified by the smart contract, and the delay decision is made.

[0100] The coordinate encryption confusion module is used for the UAV to confuse and encrypt the geographical position coordinates of itself multiple times after registration, and the task publisher uses an encryption clustering algorithm to cluster the encrypted coordinates, obtains the clustering center coordinates in the encrypted state and the UAV weight of each clustering area, and uses the encrypted clustering data.

[0101] The task allocation module is used for the smart contract to calculate the probability of each UAV reaching the task site according to the encrypted clustering data, the smart contract to evaluate the matching degree of the UAV based on the probability of the UAV reaching the task site, and the candidate UAV set is screened out; according to the delay decision result, the smart contract carries out task allocation to the candidate UAV set.

[0102] Next, the implementation results are shown. The simulation data set of the present technology is implemented, and the data distribution is shown in Table 1. The implementation process selects all UAVs and tasks that are active within a radius of 10 kilometers from a reference point at a certain time. The specific settings of the required skills of the UAV and the task are that the number of skills is 3, the skill range is 1 to 5, and the skill distribution is randomly generated within this range. The estimated completion time of the UAV is generated with a wave up and down based on the time required by the task.

[0103] Table 1 Data set data parameter distribution

[0104]

[0105] The implementation results are shown in Tables 2, 3, and 4. Table 2 shows the influence of different extreme radius noise parameters on task completion rate. According to the data in Table 2, the influence of noise parameters on system performance shows a clear gradient effect. In the low noise interval (0.001-0.01), the system maintains a high success rate of 75%-81.82%, and the stability is good, which confirms that this interval can effectively balance privacy protection and task allocation efficiency. Moderate noise (0.1) causes a 5%-10% decrease in success rate, but the system exhibits certain adaptive ability. When the noise increases to 1, the success rate drops to about 60%, and the fluctuation is significant, indicating that excessive noise interference has seriously affected the path prediction accuracy. Table 3 shows the influence of different grid side lengths on task completion rate. Experimental data shows that the grid side length significantly affects the task success rate. A 100-meter side length has the highest success rate (85%) in 10 tasks, but the privacy protection is weak; a 500-meter side length maintains an 80% success rate, achieving a balance between privacy and efficiency; a 1000-meter side length reduces the success rate to 70%, especially in large task quantities. The results show that a 500-meter side length achieves the best trade-off between privacy protection and task success rate. Table 4 shows the comparison with other task allocation schemes, including DMCQG, a static decentralized multi-capability constrained collaborative scheduling scheme that guarantees task quality and Gas optimization, BMMC, a static scene method proposed by Xu W, and the present technology-V1, which does not use any privacy protection mechanism, and the present technology-V2, which only uses grid coordinate conversion for privacy protection. The experiment compares the performance of the five task allocation schemes under different task quantities. The static methods DMCQG and BMMC have the lowest success rate (25%-66.67%) and significant fluctuations, reflecting their limitations in dynamic environments. The present technology series performs best, with a stable success rate of 75%-80% for the basic version and 87.5% for V1 in 10 tasks. The data show that dynamic allocation methods are significantly better than static schemes, and even after adding privacy protection mechanisms, their success rates remain at a high level (20%-30% higher than static schemes), achieving a good balance between privacy protection and allocation efficiency.

[0106] Table 2 Influence of different polar radius noise parameters on task completion rate

[0107]

[0108] Table 3 Influence of different grid side lengths on task completion rate

[0109]

[0110] Table 4 Comparison with other task allocation schemes

[0111]

[0112] The above describes the embodiments and implementation process of the present application in detail in combination with the drawings and tables, but the present application is not limited to the described embodiments. For those skilled in the art, various changes, modifications, replacements and variations of the embodiments including components can be made without departing from the principles and spirits of the present application, and still fall within the protection scope of the present application.

Claims

1. A decentralized unmanned aerial vehicle (UAV) task allocation method for the low-altitude economy, characterized in that, Includes the following steps: Step 1: The smart contract registers the UAV and activates the anti-Syllabus attack mechanism; Step 2: The publisher uploads the task information to the smart contract and pays a deposit. The smart contract verifies the task and makes a delay decision. Step 3: After registration, the UAV obfuscates and encrypts its own geographic coordinates multiple times. The specific implementation is as follows: The UAV performs two obfuscations on its own geographic location coordinates. The first obfuscation uses polar coordinate differential privacy obfuscation, while the second obfuscation maps the differentially privacy-processed coordinates to the center coordinates of the corresponding grid according to the predefined grid range and grid width. The obfuscated location coordinates are homomorphically encrypted using the public key provided by the task publisher. Step 4: The task publisher uses an encrypted clustering algorithm to cluster the encrypted coordinates, and obtains the cluster center coordinates and UAV weights of each cluster region in the encrypted state, which are used as encrypted clustering data. Step 5: The smart contract calculates the probability of each UAV reaching the mission location based on the encrypted clustering data. The specific implementation is as follows: The smart contract analyzes the historical movement trajectory of a single UAV to predict the probability of a worker going to different areas in the future. Specifically, it first determines the region to which the task belongs based on the weight data of the UAV in each region, the meta-path correlation, and the clustering results; then it finds all feasible paths for the UAV to travel from its current location to the target region. Then, the correlation is calculated for each path to finally obtain the probability that the UAV will reach the mission location; Step 6: The smart contract evaluates the matching degree of the UAVs based on the probability of the UAVs arriving at the task location, and filters out a set of candidate UAVs; according to the delay decision result, the smart contract allocates tasks to the set of candidate UAVs, as specifically implemented as follows: The smart contract evaluates the suitability of each UAV by combining the probability of the UAV reaching the task location, its skill coverage, and the probability of completing the task within the specified time. If the expected completion time of a UAV is greater than the specified completion time of the task minus the current time and the shortest time required for the UAV to move, then the UAV is eliminated. Next, the skill coverage of the UAV is calculated by comparing the number of skills the UAV possesses with the number of skills required by the task. The probability of completing the task within the specified time is calculated based on the estimated completion time uploaded by the UAV. Finally, the smart contract comprehensively evaluates the probability of the UAV completing the task within the specified time and selects the set of UAVs with the highest utility based on this evaluation.

2. The decentralized UAV task allocation method for the low-altitude economy according to claim 1, characterized in that, The process of registering a UAV includes: the UAV uploading its non-sensitive information to a smart contract and transferring a deposit to the smart contract; The non-sensitive information includes: one's own skill set, wallet address, and estimated task completion time for the UAV based on its skills; When making task delay decisions, smart contracts dynamically set different allocation decision points for each task based on the task's start and end times, combined with three parameters: the minimum remaining time for UAV movement before the task starts, the longest time that can be effectively predicted, and the average time to complete such tasks in the past.

3. The decentralized UAV task allocation method for the low-altitude economy according to claim 1, characterized in that, The aforementioned anti-Syllabus attack mechanism includes smart contracts resisting Syllabus attacks by combining proof-of-location and zero-knowledge proofs (ZKP). The specific implementation process is as follows: When registering, UAVs submit encrypted location snapshots with continuous timestamps. At the same time, the smart contract randomly selects three registered UAVs as validators to perform distributed verification of the submitted location data, thereby enhancing the decentralized trust of the system. In terms of privacy protection, UAV uses the ZKP algorithm to generate location proofs, which only prove to the smart contract that it is within the specified geofence, without revealing the actual coordinates.

4. The decentralized UAV task allocation method for the low-altitude economy according to claim 1, characterized in that, The task information includes: task location information, required skills, start and end times of the task, public key generated for the task, reward for this task, and base salary for the UAV.

5. The decentralized UAV task allocation method for the low-altitude economy according to claim 1, characterized in that, In step 4, the task publisher uses an encrypted k-means clustering algorithm to divide the UAV trajectory data into several clustering regions, calculates the weight of the UAV in each region, and performs the clustering operation off-chain. The task publisher then decrypts the results and transmits them to the smart contract.

6. The decentralized UAV task allocation method for the low-altitude economy according to claim 5, characterized in that, The specific implementation of the smart contract's task allocation based on the candidate UAV set according to the delay decision result is as follows: based on the delay decision result, when the task allocation decision time point is reached, the smart contract allocates tasks based on the candidate UAV set according to the allocation decision time point. After the task allocation is completed, the task allocation result is notified to the UAV. After the UAV completes the task, it uploads the current time, compares the current time with the task's required end time, and determines the final reward for each UAV based on the comparison result and the number of UAVs.

7. A decentralized drone task allocation system for the low-altitude economy, used to implement the decentralized drone allocation method according to any one of claims 1 to 6, characterized in that, Includes the following modules: Registered in the task publishing module, it is used by the smart contract to register the UAV and start the anti-Syllabus attack mechanism. The publisher uploads the task information to the smart contract and pays the deposit. The smart contract verifies the task and makes a delay decision. The coordinate encryption and obfuscation module is used to obfuscate and encrypt the geographical coordinates of the registered UAV multiple times. The task publisher uses an encrypted clustering algorithm to cluster the encrypted coordinates to obtain the cluster center coordinates and the UAV weights of each cluster region in the encrypted state, which are used as encrypted clustering data. The task allocation module is used by the smart contract to calculate the probability of each UAV arriving at the task location based on encrypted clustering data. The smart contract evaluates the matching degree of the UAVs based on the probability of the UAVs arriving at the task location and filters out a set of candidate UAVs. Based on the delay decision result, the smart contract allocates tasks to the set of candidate UAVs.

Images