Trusted resource allocation method and system in unmanned aerial vehicle assisted edge computing scene

Abstract

The invention discloses a trusted resource allocation method and system in an unmanned aerial vehicle auxiliary edge computing scene. The method comprises the following steps: step 1, constructing a multi-provider auxiliary edge computing model based on a user layer, a server layer and a block chain layer structure; step 2, a provider collects task requests of users and performs task release to form a to-be-processed task queue; the provider considers the income, the time delay and the energy consumption of each server and constructs an optimization target; 3, converting the optimization problem into a potential game problem, solving the potential game problem and giving out a distribution algorithm; and step 4, introducing a block chain technology, deploying an allocation algorithm on the smart contract, and automatically executing resource transaction and task allocation by the smart contract. According to the method, related algorithms are designed based on the potential game theory, so that the problems of trust and incentive compatibility among multiple providers are solved, the resource utilization rate is increased, and it is ensured that the providers can share resources fairly and effectively while meeting own interests as much as possible.

Description

Technical Field

[0001] The present invention belongs to the field of edge computing and blockchain technology, and specifically relates to a trusted resource allocation method and system in a drone-assisted edge computing scenario. Background Art

[0002] Edge computing frameworks are increasingly being used in emerging fields such as virtual reality, autonomous driving, and industrial IoT. They meet real-time requirements by offloading computing tasks from the cloud to the edge, reducing the need for many devices to access the cloud and large amounts of data to transmit. However, traditional edge computing frameworks usually rely on ground edge resources such as base stations and edge servers, which may not be applicable in remote areas or disaster areas. In addition, traditional ground edge servers often lack flexibility in dealing with network workloads and service coverage. To compensate for this shortcoming, drones are introduced as mobile edge servers or relays. Compared with ground servers, drones have greater mobility and can dynamically adjust their positions and assign computing tasks to cover a specified communication range. Current drones are equipped with powerful computing capabilities and can process and analyze data in flight. Therefore, air-assisted edge computing frameworks can provide lower latency services, more flexible deployment, and dynamic management.

[0003] In practical applications, many edge servers belong to different edge service providers. As a rational individual, each provider is self-driven and aims to maximize its revenue. Traditional centralized optimization methods are not applicable to edge distributed systems and fail to consider competition between providers. In a multi-provider environment, there is often a competitive relationship between providers. Each provider hopes to attract more users by providing high-quality services to increase its own market share and revenue. Under this competitive relationship, providers will believe that sharing resources will weaken their competitive advantage and benefit other providers, so they choose not to share resources. Considering the selfishness of providers, it is crucial to design an incentive mechanism to encourage resource sharing for task offloading. In addition, when providers interact, in order to obtain greater benefits, providers may deliberately provide incorrect calculation results or fail to complete calculations within the specified time in order to save computing resources, thereby undermining the fairness and efficiency of the system. Therefore, in a multi-provider scenario, a trusted interaction method is essential.

[0004] In order to solve the security and privacy issues in multi-provider scenarios, blockchain technology is considered. The decentralized nature of blockchain can ensure the transparency and immutability of all transaction and task offloading records, thereby enhancing the trust of the system. Providers can automatically execute resource sharing agreements through smart contracts to ensure that task offloading and fee settlement are carried out on the basis of consensus. This approach can not only reduce fraud, but also improve resource utilization efficiency.

[0005] However, the design of drone-assisted edge computing with multi-provider participation faces two challenges. First, due to the selfishness of providers, providers are reluctant to share task information with other providers, which poses a major challenge to the design of a resource sharing incentive mechanism. Second, some providers may prioritize maximizing their own interests, provide false information, or fail to fulfill their promises. Therefore, it is very important to establish trust among providers. Based on this, the present invention models the competition problem between providers as a non-cooperative game, aiming to establish trust among providers. In order to ensure trusted sharing among providers, the present invention designs an effective allocation method combining reputation mechanism and blockchain. Summary of the invention

[0006] Aiming at the problems existing in resource allocation and incentive mechanism between multiple edge service providers in the scenario of drone-assisted mobile edge computing in the prior art, the present invention provides a reliable resource allocation method and system. The present invention designs relevant algorithms based on potential game theory to solve the trust and incentive compatibility problems between multiple providers, so as to improve resource utilization and ensure that each provider can share resources fairly and effectively while satisfying its own interests as much as possible. At the same time, combined with the reputation mechanism and blockchain technology, the trust and interaction problems between multiple providers are solved to ensure the security and transparency of resource transactions. The present invention refers to this framework in which multiple providers provide services and resources through competition or collaboration to meet user needs as a multi-provider framework, which includes four entities: providers, drones, ground servers and users, among which drones are used as air servers. Each user signs a cooperation agreement with a specific provider, and the user sends the task to be executed to the corresponding provider, and the provider completes the execution of the task through drones or ground servers.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] A trusted resource allocation method in a drone-assisted edge computing scenario includes the following steps:

[0009] Step 1: Initialize the system and determine the system structure, which is divided into the user layer, server layer, and blockchain layer. A multi-provider assisted edge computing model is established based on this structure. Among them, users in the user layer will generate computing tasks according to their own needs and send task information to the contracted providers. The server layer contains drones and ground servers to process the sent tasks. Each provider connects several drones and ground servers. The blockchain layer coordinates the resource allocation process between the user layer and the server layer and records relevant information.

[0010] Step 2: The provider collects task requests from its users and forms a queue of pending tasks. The provider publishes the received task requests to all other providers. The provider considers the revenue, latency, and energy consumption of each server and builds optimization goals to ensure that each provider achieves efficient use of overall resources while pursuing its own interests.

[0011] Step 3: Convert the optimization problem into a potential game problem. Solve the potential game problem and give an allocation algorithm. According to the allocation algorithm, if a provider completes the task of a user belonging to another provider, it needs to pay an agency fee to the other service provider. When allocating tasks between providers, the agency fee should be set reasonably to encourage resource sharing.

[0012] Step 4: Introduce blockchain technology and deploy the allocation algorithm on smart contracts. Smart contracts automatically execute resource transactions and task allocations, and record all transaction and allocation information. Tasks are scheduled according to the allocation algorithm, and drones or ground servers receive and execute tasks. Task progress is monitored in real time to ensure that it is completed on time, and relevant data is recorded.

[0013] Preferably, step 5 is performed after step 4, introducing a reputation scoring mechanism for each provider. The reputation score comprehensively considers task completion, user feedback and historical performance. The reputation score will affect the task allocation process, and providers with higher reputation scores will be given priority. The reputation score is saved on the blockchain. Through the reputation smart contract, the service quality of the provider is evaluated in real time, and the evaluation results are fed back to the task allocation module to optimize resource allocation decisions.

[0014] Preferably, in step 1, the system framework is divided into three layers: user layer, server layer and blockchain layer. It represents the set of mobile users that provider e regularly sends task requests to edge servers (edge ​​servers include drones and ground servers. Drones are aerial servers that can be flexibly moved. Ground servers refer to servers that are fixed in a certain place and do not move. They can all be used to process tasks. In the present invention, if there is no specific reference, the server refers to the edge server. Drones and ground servers constitute the edge server). At the server level, there are E service providers in the entire system area, and the set The ground server owned by any service provider e is represented by The owned drones are represented as Any server The computing power possessed is expressed as in Indicates the number of CPU cores, and assumes that each core provides the same computing resources. jAt the blockchain layer, the relevant task allocation algorithm is deployed on the smart contract to coordinate the resource allocation process between users and servers. At the server layer, there is a dedicated control node responsible for managing the communication between the server layer and the blockchain layer.

[0015] Time is evenly divided into T time slots, expressed as Since the length of each time slot is very short, the position of the drone can be approximately considered unchanged in each time slot. Any user periodically sends computing tasks to the server. Definition is a binary variable, where It means that user i of provider e successfully sends a task to server j of provider e' (which can be other providers or provider e) at time slot t. Indicates that the user has not sent a task to the corresponding server. Any user can only choose one server of a provider to offload at time slot t. At time slot t, the task posted by user i of provider e Information is expressed as They represent the amount of data, the amount of computation, and the computation time limit. Each task will have a bid It is used to pay the provider who provides the computing service for this task. The agency fee for each task is expressed as

[0016] Preferably, step 2 is as follows: Due to limited computing resources of the device, it is assumed that all tasks need to be offloaded to the edge server for computing. It is assumed that the positions of the user and the ground server are stationary. Vector Indicates that user i is at a stationary horizontal coordinate position, vector represents the location of the ground server g, and the distance from user i to the ground server g in time slot t can be obtained as Consider the channel gain from user i to ground server g, expressed as where β 0 Indicates the received power. represents the horizontal position of the drone belonging to provider e at time slot t. Assuming that the drone is always flying at a fixed height H, the distance from user i to the drone can be calculated as In drone networks, line-of-sight links are critical to the quality and stability of signal transmission. Since drones usually fly in the air, line-of-sight links can ensure that signals are not blocked by ground obstacles such as buildings, trees, and terrain changes during transmission, thereby reducing signal attenuation and multipath effects and improving communication performance. Compared with non-line-of-sight links, line-of-sight links have more direct and clear transmission paths, providing higher transmission rates and lower latency. Therefore, the channel gain from user i to drone f can be expressed as By using OFDMA, we can get user i to the server The transmission rate is expressed as:

[0017]

[0018] in, It is represented as the channel bandwidth from user i to server j, represents the transmission power of user i, is the channel gain, σ 2 Potential noise power.

[0019] When the task When it is transmitted to a server for calculation, data transmission delay and task calculation delay will occur. The transmission delay is expressed as The calculation delay is expressed as in Represents the set of drones and ground servers. Therefore, the delay incurred by the execution of the entire task is as follows:

[0020]

[0021] The energy consumption of the server to perform this task is:

[0022]

[0023] where κ j is the chip correlation coefficient for server j.

[0024] Preferably, in step 3, the optimization objective is transformed into a potential game problem. First, the three basic elements of the game are clarified: participants, method space, and utility function. It can be expressed as in, represents the set of participants, which is the set of providers in this invention. Π represents the set of method spaces for each provider. U represents the set of utilities for each provider. Method space Π e It is a collection of tasks The total method space Π is the combination of all providers’ methods, that is, Π = Π 1 ×…×Π E ,in, The utility of each player is u e,t (π e ), so U={u 1,t ,u 2,t ,…,u E,t}.

[0025] If we want to prove that a function is a potential function, we need to satisfy

[0026] The present invention proposes a potential function

[0027]

[0028] It is proved that this function satisfies the definition of potential function, so the game has a Nash equilibrium solution. After solving it, the specific allocation algorithm is given:

[0029] 1) Initialize time slice t, user i sends a task to the corresponding provider e, all competing providers publish the task information sent by the user to the blockchain, and calculate the intermediary fee for each task. The task quotation sent by user i of provider e is Calculate the agency fee for publishing tasks as

[0030] 2) All providers request to be allocated, and the winner will be given priority to be allocated. The blockchain uses a reputation selection algorithm. For each provider, its reputation score is first updated based on user factors, and then the total reputation score of all providers is calculated. Then a random value l is generated, ranging from 0 to the total reputation score. All providers are traversed and their reputation scores are accumulated. If the accumulated value reaches or exceeds l, the current provider is selected as the winner. The winner information is recorded on the blockchain and the winner is returned. The higher the reputation value, the greater the probability of being called the winner.

[0031] 3) The winner then calculates his optimal allocation method π e , solved by mixed integer linear programming method, and the optimal solution is obtained using the branch and bound algorithm with the help of standard solvers such as CVX or MOSEK.

[0032] 4) The blockchain assigns the task to the winner and records the assignment result and task information. At the same time, the winner is removed from the competitor set and the assigned task is removed from the task set. The above process is repeated until all competitors or tasks are assigned.

[0033] Preferably, in step 5, the reputation scoring mechanism is as follows: first, consider the execution of the current task, and for each completed task, set a basic reputation score s base , and additional credit points are awarded based on task performance bonus , task performance refers to the time it takes to complete a task. Therefore, when a server successfully completes a task, the reputation score it obtains can be expressed as s = s base +s bonus Considering the long-term stability of the server, a factor m is introduced to represent the number of interactions between the server and the user. The long-term stability factor is defined as The more tasks the server completes for users, the more rewards it will receive. bonus The specific calculation is as follows:

[0034]

[0035] Among them, θ is the reward parameter, t act is the actual execution time of the task. Within the time limit, the shorter the time it takes to complete the task, the greater the extra reward.

[0036] Further preferably, user evaluation is considered, as follows: In order to prevent malicious users from reducing the reputation of servers with good performance, the user's credibility is included in the evaluation. Assume that after ξ-1 rounds of evaluation, user i's ratings of the interactive servers form a set M i , whose average score is calculated as make represents the score of each round, and the standard deviation is defined as:

[0037]

[0038] Then the accuracy of user ratings is expressed as The credit score obtained by the user in the first round of evaluation can be obtained as Considering the performance and user evaluation of server j, the total reputation evaluation of server j in round m is:

[0039]

[0040] Refer to the exponential moving average technique to construct the reputation score:

[0041]

[0042] Among them, α+β=1, α∈[0,1] represents the weight of past reputation scores. Increasing the weight of α means that recent transactions have a greater impact on the reputation score, ensuring that only continued good performance can maintain a high reputation score. By considering both task performance and user evaluation, the overall reputation score of provider e can be calculated as

[0043]

[0044] The present invention focuses on maximizing the utility of each provider while meeting the user's communication and computing time requirements. This includes the sum of the utilities of each provider's internal servers and the intermediary fees obtained by providing tasks to other providers. The utility of server j of provider e is defined as:

[0045]

[0046] where γ is the cost per unit of energy.

[0047] The total utility obtained by each provider can be expressed as:

[0048]

[0049] The second term on the right side of the equation represents the intermediary fee charged for providing tasks to other providers, and the third term represents the intermediary fee paid for selecting tasks from other providers.

[0050] The present invention uses π e Denote the decision of provider e to allocate tasks to its own servers, denoted by π -e represents the task allocation decision of all other providers except provider e. The multi-provider task offloading optimization problem is formulated as follows:

[0051]

[0052] Constraint (12) states that each task can only be sent to one server.

[0053] Constraint (13) indicates whether to offload task i of provider e to server j.

[0054] Constraint (14) ensures that the delay must be less than the deadline of each task.

[0055] Constraint (15) ensures that the number of tasks executed by each server does not exceed the upper limit of the number of cores of the server.

[0056] Use potential game theory to solve it, specifically in the problem of multi-edge computing task offloading, by building a potential game model and solving the potential game problem, to achieve efficient task allocation and optimal resource utilization. The specific method is:

[0057] First, a provider game model is established. Each provider Method π e Defined as a collection of tasks The subset of tasks selected in That is, select the task to be uninstalled. Method space Π e For task set All possible combinations of , the overall method space Π is the Cartesian product of all provider methods, that is, Π = Π 1 ×Π 2 ×…×Π E . Let the utility function of each provider be u e,t (π e ), the overall utility set is U = {u 1,t ,u 2,t ,…,u E,t}.

[0058] Nash equilibrium refers to a stable state in which no single provider can improve its utility by changing its own method. Next, we have:

[0059]

[0060] Then construct the potential function and define the potential function φ(π e ,π -e ) is the sum of all provider utilities minus the sum of all transaction intermediary fees, that is:

[0061]

[0062] Finally, the correctness of the potential function is verified. By proving that when any provider changes its method π e to π e′ When , the change of its utility is equivalent to the change of the potential function, that is:

[0063] u e,t (π e′ ,π -e )-u e,t (π e ,π -e )=φ(π e′ ,π -e )-φ(π e ,π -e )

[0064] Thus, it is verified that this game model is a potential game. According to the properties of potential games, it is ensured that there are one or more Nash equilibria. It is verified that the potential function holds and the game is a potential game with a Nash equilibrium solution.

[0065] Next, we build the allocation algorithm and embed it into the blockchain smart contract. The specific steps of the algorithm are as follows:

[0066] 1) At the beginning of each time slice, all competing providers first publish their task requests and calculate the agency fee for each task. The agency fee is the bid for each task multiplied by a certain ratio.

[0067] 2) Providers request to update their allocation method on the blockchain. The blockchain selects the winner based on the reputation selection contract.

[0068] 3) The winner then calculates his optimal allocation method π e , and is solved by a mixed integer linear programming approach, using a branch and bound algorithm with the help of standard solvers such as CVX or MOSEK to obtain the optimal solution.

[0069] 4) The blockchain records the relevant information and assigns the task to the winner. At the same time, the winner is removed from the competitor set and the assigned task is removed from the task set.

[0070] Repeat the above process until all competitors or tasks have been assigned.

[0071] Next, we build a reputation update algorithm, which is embedded in the blockchain and used to dynamically update the reputation score of each provider. For each provider, we first update its reputation change based on user factors, and then calculate the total reputation score of all providers. Then we generate a random value l, ranging from 0 to the total reputation score. We traverse all providers and accumulate their reputation scores. If the accumulated value reaches or exceeds l, we select the current provider as the winner. We record the winner information on the blockchain and return the winner.

[0072] Through the above-mentioned trusted task offloading scheme based on potential game, it is possible to effectively achieve fair distribution of tasks and optimal utilization of resources in a multi-provider competition environment. The potential game model ensures the stability and balance of the system, and the reputation update mechanism ensures the integrity and service quality of the participating providers, thereby improving the credibility and efficiency of the overall system.

[0073] The present invention also discloses a trusted resource allocation system in a drone-assisted edge computing scenario, which is used to execute the above method and includes the following modules:

[0074] Model building module: Construct the user layer, server layer, and blockchain layer structure, and establish a multi-provider assisted edge computing model based on this structure; users in the user layer generate computing tasks and send task information to the contracted providers; the server layer contains drones and ground servers to process the sent tasks, and each provider connects drones and ground servers; the blockchain layer coordinates the resource allocation process between the user layer and the server layer, and records information;

[0075] Optimization target building module: Providers collect task requests from their users and publish tasks to form a queue of pending tasks; Providers publish the received task requests to all other providers; Providers consider the revenue, latency, and energy consumption of each server to build optimization targets;

[0076] Problem conversion module: converts the optimization problem into a potential game problem, solves the potential game problem and gives an allocation algorithm;

[0077] Task scheduling and execution module: Introduce blockchain technology and deploy the allocation algorithm on smart contracts. Smart contracts automatically execute resource transactions and task allocation, and record all transaction and allocation information.

[0078] Preferably, a reputation evaluation module is also included: a reputation score is assigned to each provider, the reputation score takes into account task completion, user feedback and historical performance, and the reputation score is stored in the blockchain.

[0079] Compared with the prior art, the beneficial effects of the present invention are:

[0080] 1) Improve service performance: By adopting a task offloading scheme based on potential game, the task allocation among multiple edge service providers is effectively managed, significantly improving the overall service quality of the system.

[0081] 2) Enhance the flexibility of network load and service coverage: Use drones as mobile edge servers and dynamically adjust their locations to adapt to changes in network load and expand service coverage, thereby improving the adaptability and flexibility of the system.

[0082] 3) Ensure the security and credibility of resource transactions: Through the sharing mechanism designed by blockchain technology, the transparency and immutability of resource transactions between different providers are ensured.

[0083] 4) Achieve stability of task offloading strategy: Model the multi-provider task offloading problem as a potential game and prove the existence of its Nash equilibrium, ensuring that the task offloading strategy reaches a stable and optimized state with the participation of multiple parties, thereby improving the overall efficiency of the system.

[0084] 5) Provide a reputation-based service quality feedback mechanism: A reputation smart contract is designed to record and feedback the service quality provided by each provider, dynamically adjust task allocation based on historical performance, incentivize providers to improve service quality, and maintain the fairness and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 This is a flow chart of a trusted resource allocation method in a drone-assisted edge computing scenario according to a preferred embodiment of the present invention;

[0086] Figure 2 A system structure diagram of a preferred embodiment of the present invention;

[0087] Figure 3 This is a block diagram of a trusted resource allocation system in a drone-assisted edge computing scenario according to a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0088] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0089] This embodiment is aimed at the multi-provider drone-assisted edge computing scenario, involving a potential game method, taking into account competitive environment factors, maximizing the utility of each provider. At the same time, combined with blockchain technology, the credibility of the interaction between providers is ensured.

[0090] Figure 1 The overall process of a trusted resource allocation method in a multi-provider drone-assisted edge computing scenario is presented, which mainly includes the following steps:

[0091] Step 1: Initialize the system and determine the system structure. Figure 2The system structure involved in this embodiment is shown, which is divided into a user layer, a server layer, and a blockchain layer. A multi-provider assisted edge computing model is established based on this structure. Among them, users in the user layer will generate computing tasks according to their own needs and send task information to the contracted provider. The server layer includes drones and ground servers for processing sent tasks. Each provider has several drones and ground servers. The blockchain coordinates the resource allocation process between users and servers and records relevant information. The details are as follows:

[0092] At the user level, use represents the set of mobile users that regularly send task requests to the edge server. At the server level, there are E service providers in the entire system area, and the set The ground server owned by any service provider e is represented by The owned drones are represented as Any server The computing power possessed is expressed as in, Indicates the number of CPU cores, set to 4 or 8, and assumes that each core provides the same computing resources. j For example, the total computing resources that a ground server can provide are set to 40GHz, and the total computing resources that a drone can provide are set to 20GHz. At the blockchain layer, the relevant allocation algorithm is deployed on the smart contract to coordinate the resource allocation process between users and servers. At the server layer, there is a dedicated control node responsible for managing the communication between the server layer and the blockchain layer.

[0093] Divide the time evenly into T time slots, expressed as Any user periodically sends computing tasks to the server. is a binary variable, where It is represented as user i of provider e successfully sending to server j of provider e' (which can be equal to e) in time slot t. Conversely, Indicates that the user has not sent a task to the corresponding server. Any user can only choose one server of a provider to offload at time slot t. At time slot t, the task posted by user i of provider e Information is expressed as They represent the amount of data, the amount of computation, and the computation time limit. Each task will have a bid Used to pay the corresponding providers. The agency fee for each task is expressed as This embodiment is specifically configured as One percent of.

[0094] Step 2: The provider collects task requests from its users and publishes them to form a queue of pending tasks. The provider publishes the received task requests to all other providers. The provider comprehensively considers server revenue, latency, and energy consumption, and builds optimization goals to ensure that each provider achieves efficient use of overall resources while pursuing its own interests. The details are as follows:

[0095] Using Vectors Indicates that user i is at a stationary horizontal coordinate position. Similarly, use the vector represents the location of the ground server g, and the distance from user i to the ground server g in time slot t is Calculate the channel gain from user i to ground server g as Among them, β 0 Indicates the received power at a reference distance of 1m and a transmit power of 1W, set to -50db. represents the horizontal position of the drone of provider e at time slot t. Assuming that the drone is always navigating at a fixed height H, set to 100m, the distance from user i to the drone is The channel gain from user i to drone f can be expressed as By using OFDMA, we can get user i to the server The transmission rate in It is represented as the channel bandwidth from user i to server j, represents the transmission power of user i, is the channel gain, σ 2 Potential noise power.

[0096] Computational tasks The transmission delay is The calculated delay is in Find the delay incurred by the execution of the entire task The energy consumption of the server to perform this task is Among them, κ j is the chip correlation coefficient.

[0097] The utility of server j in each provider is calculated as:

[0098]

[0099] where γ is the cost per unit of energy.

[0100] The total utility obtained by each provider can be expressed as:

[0101]

[0102] The second term on the right side of the equation represents the intermediary fee charged for providing tasks to other providers. The third term on the right side of the equation represents the intermediary fee paid for selecting tasks from other providers.

[0103] Use π e Denote the decision of provider e to allocate tasks to its own servers, denoted by π -e represents the task allocation decision of all other providers except provider e. represents the utility function of provider e at time slot t. The multi-provider task offloading optimization problem is formulated as follows:

[0104]

[0105] Step 3: Convert the optimization goal into a potential game problem. Solve the potential game problem and give an allocation algorithm. According to the allocation algorithm, if a provider completes the task of a user belonging to another provider, it needs to pay an agency fee to the other service provider. When allocating tasks between providers, the agency fee should be set reasonably to encourage resource sharing.

[0106] First, we need to clarify the three basic elements of the game: participants, method space, and utility function. We can express it as in represents the set of participants, which is the set of providers in this invention. Π represents the set of method spaces for each provider. U represents the set of utilities for each provider. Method space Π e It is a collection of tasks The total method space Π is the combination of all providers’ methods, that is, Π = Π 1 ×…×Π E ,in The utility of each player is u e,t (π e ), so U={u 1,t ,u 2,t ,…,u E,t}.

[0107] If we want to prove that a function is a potential function, we need to satisfy

[0108] The present invention proposes a potential function

[0109]

[0110] It is proved that this function satisfies the definition of potential function, so the game has a Nash equilibrium solution. After solving the potential game problem, a specific allocation algorithm is given:

[0111] 1) Initialize time slice t, user i sends a task to the corresponding provider e, all competing providers publish the task information sent by the user, and calculate the intermediary fee for each task. The task quotation sent by user i of provider e is Calculate the agency fee for publishing tasks as

[0112] 2) All providers request to be allocated, and the winner will be given priority to obtain the allocation. According to the reputation selection algorithm, for each provider, first update its reputation score based on user factors, and then calculate the total reputation score of all providers. Then generate a random value l, ranging from 0 to the total reputation score. Traverse all providers and accumulate their reputation scores. If the accumulated value reaches or exceeds l, the current provider is selected as the winner. The higher the reputation value, the greater the probability of being called the winner.

[0113] 3) The winner then calculates his optimal allocation method π e , solved by mixed integer linear programming method, and the optimal solution is obtained using the branch and bound algorithm with the help of standard solvers such as CVX or MOSEK.

[0114] 4) The blockchain assigns the task to the winner and records the assignment result and task information. At the same time, the winner is removed from the competitor set and the assigned task is removed from the task set. The above process is repeated until all competitors or tasks are assigned.

[0115] Step 4: Deploy the allocation method on the blockchain smart contract. The smart contract automatically executes resource transactions and task allocation, and records all transaction and allocation information. According to the method proposed by the present invention, the task is scheduled, and the drone or ground server receives and executes the task. The system monitors the progress of the task in real time to ensure that it is completed on time and records relevant data.

[0116] Specifically, the task allocation method is deployed on the blockchain, using the consortium chain Hyperledger Fabric, and smart contracts are written to realize the automation of resource transactions and task allocation. All transaction and allocation information are recorded transparently and tamper-proof on the blockchain to ensure the openness and credibility of the system. In the task scheduling process, according to the method proposed by the present invention, the drone or ground server receives and executes the tasks assigned to them. The smart contract calculates and allocates the optimal task allocation strategy. At the same time, the smart contract monitors the progress of the task in real time, and uses the preset feedback mechanism and reputation scoring system to dynamically adjust the task allocation strategy to ensure that the task can be completed on time.

[0117] Step 5: After the task is completed, the reputation of the server that performed the task is evaluated based on the task completion status, user feedback and historical performance. The reputation score will affect the task allocation process, and providers with higher reputation scores will be given priority. The score is updated immediately after the evaluation and affects future competition. The evaluation results are recorded through the blockchain. Through the reputation smart contract, the service quality of the provider is evaluated in real time, and the evaluation results are fed back to the task allocation to optimize resource allocation decisions.

[0118] Considering the execution of the current task, for each completed task, a basic credit score s is set basw , and additional credit points are awarded based on task performance bonus , task performance refers to the time it takes to complete a task. Therefore, when a server successfully completes a task, the reputation score it obtains can be expressed as s = s base +s bonus The number of interactions between the server and the user is m, and the long-term stability factor is calculated s bonus The specific calculation is as follows:

[0119]

[0120] Among them, θ is the reward parameter, t act is the actual execution time of the task. The shorter the actual execution time, the more extra rewards you will get.

[0121] Next, consider the user evaluation. In order to prevent malicious users from reducing the reputation of servers with good performance, the present invention incorporates the user's credibility into the evaluation. Assume that after ξ-1 rounds of evaluation, the scores of user i on the interactive servers (the tasks sent by the user are unloaded to the corresponding servers after allocation, that is, the servers interacted with) form a set M i , whose average score is calculated as make represents the score of each round, and the standard deviation is defined as:

[0122]

[0123] Then the accuracy of user ratings is expressed as The credit score obtained by the user in the first round of evaluation can be obtained as Considering the performance and user evaluation of server j, the total reputation evaluation of server j in round m is:

[0124]

[0125] The reputation score is constructed by referring to the exponential moving average technique (see the paper: Q. He, J. Yan, H. Jin, and Y. Yang, "Service Trust: Supporting reputation-oriented service selection," in Service-Oriented Comput. Berlin, Germany: Springer, 2009, pp. 269–284.):

[0126]

[0127] Among them, α+β=1, α∈[0,1] represents the weight of past reputation scores. Increasing the weight of α means that recent transactions have a greater impact on the reputation score, ensuring that only continued good performance can maintain a high reputation score. By considering both task performance and user evaluation, the overall reputation score of provider e can be calculated as

[0128]

[0129] The present invention focuses on maximizing the utility of each provider while meeting the user's communication and computing time requirements. This includes the sum of the utilities of each provider's internal servers and the intermediary fees obtained by providing tasks to other providers. The utility of server j of provider e is defined as:

[0130]

[0131] where γ is the cost per unit of energy.

[0132] The total utility obtained by each provider can be expressed as:

[0133]

[0134] The second term on the right side of the equation represents the intermediary fee charged for providing tasks to other providers, and the third term represents the intermediary fee paid for selecting tasks from other providers.

[0135] The present invention uses π e Denote the decision of provider e to allocate tasks to its own servers, denoted by π -e represents the task allocation decision of all other providers except provider e. The multi-provider task offloading optimization problem is formulated as follows:

[0136]

[0137] Constraint (12) states that each task can only be sent to one server.

[0138] Constraint (13) indicates whether to offload task i of provider e to server j.

[0139] Constraint (14) ensures that the delay must be less than the deadline of each task.

[0140] Constraint (15) ensures that the number of tasks executed by each server does not exceed the upper limit of the number of cores of the server.

[0141] Use potential game theory to solve it, specifically in the problem of multi-edge computing task offloading, by building a potential game model and solving the potential game problem, to achieve efficient task allocation and optimal resource utilization. The specific method is:

[0142] First, a provider game model is established. Each provider Method π e Defined as a collection of tasks The subset of tasks selected in That is, select the task to be uninstalled. Method space Π e For task set All possible combinations of , the overall method space Π is the Cartesian product of all provider methods, that is, Π = Π 1 ×Π 2 ×…×Π E . Let the utility function of each provider be u e,t (π e ), the overall utility set is U = {u 1,t ,u 2,t ,…,u E,t}.

[0143] Nash equilibrium refers to a stable state in which no single provider can improve its utility by changing its own method. Next, we have:

[0144]

[0145] Then construct the potential function and define the potential function φ(π e ,π -e ) is the sum of all provider utilities minus the sum of all transaction intermediary fees, that is:

[0146]

[0147] Finally, the correctness of the potential function is verified. By proving that when any provider changes its method π e to π e′ When , the change of its utility is equivalent to the change of the potential function, that is:

[0148] ue,t (π e′ ,π -e )-u e,t (π e ,π -e )=φ(π e′ ,π -e )-φ(π e ,π -e )

[0149] Thus, it is verified that this game model is a potential game. According to the properties of potential games, it is ensured that there are one or more Nash equilibria. It is verified that the potential function holds and the game is a potential game with a Nash equilibrium solution.

[0150] Next, we build the allocation algorithm and embed it into the blockchain smart contract. The specific steps of the algorithm are as follows:

[0151] 1) At the beginning of each time slice, all competing providers first publish their task requests and calculate the agency fee for each task. The agency fee is the bid for each task multiplied by a certain ratio.

[0152] 2) Providers request to update their allocation method on the blockchain. The blockchain selects the winner based on the reputation selection contract.

[0153] 3) The winner then calculates his optimal allocation method π e , and is solved by a mixed integer linear programming approach, using a branch and bound algorithm with the help of standard solvers such as CVX or MOSEK to obtain the optimal solution.

[0154] 4) The blockchain records the relevant information and assigns the task to the winner. At the same time, the winner is removed from the competitor set and the assigned task is removed from the task set.

[0155] Repeat the above process until all competitors or tasks have been assigned.

[0156] Next, we build a reputation update algorithm, which is embedded in the blockchain and used to dynamically update the reputation score of each provider. For each provider, we first update its reputation change based on user factors, and then calculate the total reputation score of all providers. Then we generate a random value l, ranging from 0 to the total reputation score. We traverse all providers and accumulate their reputation scores. If the accumulated value reaches or exceeds l, we select the current provider as the winner. We record the winner information on the blockchain and return the winner.

[0157] Through the above-mentioned trusted task offloading scheme based on potential game, it is possible to effectively achieve fair distribution of tasks and optimal utilization of resources in a multi-provider competition environment. The potential game model ensures the stability and balance of the system, and the reputation update mechanism ensures the integrity and service quality of the participating providers, thereby improving the credibility and efficiency of the overall system.

[0158] like Figure 3 As shown, this embodiment discloses a trusted resource allocation system in a drone-assisted edge computing scenario, which is used to execute the above method and includes the following modules:

[0159] Model building module: Construct the user layer, server layer, and blockchain layer structure, and establish a multi-provider assisted edge computing model based on this structure; users in the user layer generate computing tasks and send task information to the contracted providers; the server layer contains drones and ground servers to process the sent tasks, and each provider connects drones and ground servers; the blockchain layer coordinates the resource allocation process between the user layer and the server layer, and records information;

[0160] Optimization target building module: Providers collect task requests from their users and publish tasks to form a queue of pending tasks; Providers publish the received task requests to all other providers; Providers consider the revenue, latency, and energy consumption of each server to build optimization targets;

[0161] Problem conversion module: converts the optimization problem into a potential game problem, solves the potential game problem and gives an allocation algorithm;

[0162] Task scheduling and execution module: Introducing blockchain technology, deploying the allocation algorithm on smart contracts, which automatically execute resource transactions and task allocation, and record all transaction and allocation information; scheduling tasks according to the allocation algorithm, and drones or ground servers receive and execute tasks;

[0163] Reputation Assessment Module: Each provider is given a reputation score that takes into account task completion, user feedback, and historical performance. The reputation score is stored in the blockchain.

[0164] For other contents of this embodiment, please refer to the above method embodiment.

[0165] Although the above content has described the embodiments of the present invention in detail, the present invention is not limited to the application fields in the specification and specific embodiments. The present invention can be widely applied to various related fields, and those skilled in the art can make various modifications without departing from the scope of the claims of the present invention and their equivalents. Therefore, the present invention is not limited to the specific details and embodiments shown and described herein.

Claims

1. A trusted resource allocation method in a drone-assisted edge computing scenario, characterized in that: The following steps are involved: Step 1: Build the user layer, server layer, and blockchain layer structure, and establish a multi-provider assisted edge computing model based on this structure; users in the user layer generate computing tasks and send task information to the contracted providers; the server layer includes drones and ground servers to process the sent tasks, and each provider connects drones and ground servers; the blockchain layer coordinates the resource allocation process between the user layer and the server layer, and records information; Step 2: The provider collects the task requests of its users and publishes the tasks to form a queue of pending tasks; the provider publishes the received task requests to all other providers; the provider considers the revenue, latency, and energy consumption of each server and builds the optimization goal; Step 3: Convert the optimization problem into a potential game problem, solve the potential game problem and give an allocation algorithm; Step 4: Introduce blockchain technology and deploy the allocation algorithm on the smart contract. The smart contract automatically executes resource transactions and task allocation and records all transaction and allocation information.

2. The trusted resource allocation method in a drone-assisted edge computing scenario as claimed in claim 1, characterized in that: In step 1, at the user level, use represents the set of mobile users whose provider e regularly sends task requests to the edge server; at the server layer, there are E service providers, using the set To represent; any service provider e Unicom's ground server is represented by Unicom’s drones are represented as Any edge server The computing power possessed is expressed as in, Indicates the number of CPU cores, and assumes that each core provides the same computing resources. j ; Time is evenly divided into T time slots, expressed as Any user periodically sends computing tasks to the server; definition is a binary variable, where It means that user i of provider e successfully sends a task to server j of provider e' at time slot t. Indicates that the user has not sent a task to the corresponding server; any user can only choose one server of a provider to offload at time slot t; at time slot t, the task posted by user i of provider e Information is expressed as Respectively represent the amount of data, the amount of calculation, and the calculation time limit; each task has a bid It is used to pay the provider who provides the computing service for the task. The agency fee for each task is expressed as 3. A trusted resource allocation method in a drone-assisted edge computing scenario as claimed in claim 2, characterized in that: Step 2 is as follows: Assume that all tasks need to be offloaded to the edge server for calculation, and the positions of the user and the ground server remain unchanged; vector Indicates that user i is at the horizontal coordinate position, vector represents the location of the ground server g, and the distance from user i to the ground server g in time slot t is Consider the channel gain from user i to ground server g, expressed as Among them, β0 represents the received power; represents the horizontal position of the UAV of provider e at time slot t; assuming that the UAV flies at a fixed height H, the distance from user i to the UAV is The channel gain from user i to drone f is expressed as By using OFDMA to get user i to the server The transmission rate is expressed as: in, It is represented as the channel bandwidth from user i to server j, represents the transmission power of user i, is the channel gain, σ 2 Potential noise power; When the task When it is transmitted to a server for calculation, the data transmission delay is expressed as The task computation delay is expressed as in, represents the set of drones and ground servers; therefore, the latency of the entire task being executed is as follows: The computing energy consumed by the server to perform this task is: Among them, κ j is the chip correlation coefficient of server j; The utility of server j in each provider is calculated as: Where γ is the cost per unit of energy; The total utility obtained by each provider is expressed as: The second term on the right side of the equation represents the intermediary fee charged for providing tasks to other providers, and the third term on the right side of the equation represents the intermediary fee paid for selecting tasks from other providers; Use π e Denote the decision of provider e to allocate tasks to its own servers, denoted by π -e represents the task allocation decision of all other providers except provider e; let represents the utility function of provider e at time slot t; the multi-provider task offloading optimization problem is expressed as follows:

4. The trusted resource allocation method in a drone-assisted edge computing scenario as claimed in claim 3, characterized in that: Step 3 is as follows: First, we need to clarify the three basic elements of the game: participants, method space, and utility function, which can be expressed as in represents the set of participants, i.e., the set of providers, Π represents the set of method spaces for each provider, and U represents the set of utility functions for each provider; the method space Π e It is a collection of tasks All task combinations in the method space Π are the method combinations of all providers, that is, Π = Π1×…×Π E ,in The utility function of each player is u e,t (π e ), so U={u 1,t ,u 2,t ,…,u E,t }; Propose the potential function: After solving, the specific allocation algorithm is given: 1) Initialize time slice t, user i sends a task to the corresponding provider e, all competing providers publish the task information sent by the user to the blockchain, and calculate the intermediary fee for each task. The task quotation sent by user i of provider e is Calculate the agency fee for publishing tasks as 2) All providers request to be allocated, and the winner will have priority in obtaining the allocation rights; for each provider, the blockchain first updates its reputation score based on user factors, and then calculates the total reputation score of all providers; then generates a random value l ranging from 0 to the total reputation score; traverses all providers and accumulates the reputation scores. If the accumulated value reaches or exceeds l, the current provider is selected as the winner; the winner information is recorded on the blockchain and the winner is returned; 3) The winner calculates his optimal allocation method π e , and obtain the optimal solution by solving it using the mixed integer linear programming method; 4) The blockchain assigns the task to the winner and records the assignment result and task information. At the same time, the winner is removed from the competitor set and the assigned task is removed from the task set. Return to step 1) until all competitors or tasks are assigned.

5. A trusted resource allocation method in a drone-assisted edge computing scenario as described in any one of claims 1 to 4, characterized in that: After step 4, proceed to step 5 to assign a reputation score to each provider. The reputation score takes into account task completion, user feedback, and historical performance. The reputation score is stored in the blockchain.

6. A trusted resource allocation method in a drone-assisted edge computing scenario as claimed in claim 5, characterized in that: In step 5, the credit score is as follows: First consider the execution of the current task, and set a basic credit score s for each completed task base , and additional credit points are awarded based on task performance bonus , task performance refers to the time it takes to complete a task; therefore, when a server successfully completes a task, the reputation score obtained is expressed as s = s base +s bonus ; A factor m is introduced to represent the number of interactions between the server and the user. The long-term stability factor is defined as The more tasks the server completes, the more rewards it will receive. bonus The specific calculation is as follows: Among them, θ is the reward parameter, t act is the actual execution time of the task.

7. A trusted resource allocation method in a drone-assisted edge computing scenario as claimed in claim 6, characterized in that: In step 5, the user feedback is as follows: Assume that after ξ-1 rounds of evaluation, user i’s ratings of the interactive servers form a set M i , whose average score is calculated as make represents the score of each round, and the standard deviation is defined as: The accuracy of user ratings is expressed as The credit score obtained by the user in the first round of evaluation is The total reputation evaluation of server j in round m is: Building a reputation score: Among them, α+β=1, α∈[0,1] represents the weight of the past reputation score; the overall reputation score of provider e is calculated as:

8. A trusted resource allocation system in a drone-assisted edge computing scenario, used to execute the method according to any one of claims 1 to 4, characterized in that: Includes the following modules: Model building module: Construct the user layer, server layer, and blockchain layer structure, and establish a multi-provider assisted edge computing model based on this structure; users in the user layer generate computing tasks and send task information to the contracted providers; the server layer contains drones and ground servers to process the sent tasks, and each provider connects drones and ground servers; the blockchain layer coordinates the resource allocation process between the user layer and the server layer, and records information; Optimization target building module: Providers collect task requests from their users and publish tasks to form a queue of pending tasks; Providers publish the received task requests to all other providers; Providers consider the revenue, latency, and energy consumption of each server and build optimization targets; Problem conversion module: converts the optimization problem into a potential game problem, solves the potential game problem and gives an allocation algorithm; Task scheduling and execution module: Introduce blockchain technology and deploy the allocation algorithm on smart contracts. Smart contracts automatically execute resource transactions and task allocation, and record all transaction and allocation information.

9. A trusted resource allocation system in a drone-assisted edge computing scenario as claimed in claim 8, characterized in that: It also includes a reputation assessment module: a reputation score is assigned to each provider, and the reputation score takes into account task completion, user feedback, and historical performance, and the reputation score is stored in the blockchain.

Images