Content distribution method in multi-drone assisted edge computing system

Abstract

The present invention relates to a content distribution method in a multi-drone assisted edge computing system. First, a multi-drone assisted edge computing system model is established, and the joint optimization problem of drone deployment and content caching is defined. Then, with the minimum average response delay of each user requesting content in the system as the goal, the joint optimization problem is solved by combining a genetic algorithm and a greedy strategy to obtain the optimal drone deployment position and content caching solution. The method is conducive to optimizing drone deployment and content caching, and reducing the average response delay of each user requesting content in the system.

Description

Technical Field

[0001] The present invention relates to the field of drones, and in particular to a content distribution method in a multi-drone assisted edge computing system. Background Art

[0002] In order to meet the challenges brought by the rapid growth of new mobile application needs and the number of mobile connected devices, a large number of new architectures and technologies have emerged, and edge computing is a product that conforms to this development trend. Edge computing sinks the cache and computing functions in the cloud computing architecture to the edge of the network, so that the user's computing and transmission needs are largely met at the edge of the network close to the user, making up for the defects of insufficient user computing power, limited energy supply, and high service latency and user energy consumption caused by long-distance transmission of large amounts of data, so that the user's latency experience is qualitatively improved.

[0003] In recent years, drones have been widely used in many industries due to their small size, low price and strong flexibility. The characteristics of drones enable them to effectively solve the problems of high deployment cost and poor adaptability to special scenarios in traditional communications. Drones can be deployed as aerial base stations (BS) to assist conventional cellular networks. The main application scenarios of drone communications include hotspot high-speed coverage, information transmission, emergency communications, etc. The data traffic requested by mobile users will increase dramatically in future mobile networks, of which 71% of the data traffic will be used for content distribution. Content caching at the edge of the network is considered to be a key technology for content-centric cellular networks to alleviate network traffic load. Popular content is placed at the BS and user terminals close to the user to reduce content acquisition latency and backhaul link load. In order to meet the data volume requirements of multimedia content distribution and alleviate the traffic pressure of ground BSs in cellular networks, caching-enabled drones are deployed at some hotspots to reduce traffic during peak hours. It provides a low-cost and fast deployment solution for high data rate and low-latency content distribution applications. In high-speed cache drone-assisted cellular networks, how to cache the most appropriate content within limited cache space and perform high-speed content distribution under limited wireless resources is a major problem that needs to be solved. Therefore, the quality of content distribution technology is crucial to the user's quality of experience. Summary of the invention

[0004] The purpose of the present invention is to provide a content distribution method in a multi-drone assisted edge computing system, which is conducive to optimizing drone deployment and content caching, and reducing the average response delay of content requests from each user in the system.

[0005] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is: a content distribution method in a multi-drone assisted edge computing system, firstly, a multi-drone assisted edge computing system model is established, and the joint optimization problem of drone deployment and content caching is defined, and then the average response delay of each user requesting content in the system is minimized as the goal, and the joint optimization problem is solved by combining genetic algorithm and greedy strategy to obtain the optimal drone deployment location and content caching solution.

[0006] Furthermore, in the caching-supported UAV-assisted edge computing system, the set of ground user mobile devices is defined as K = {1,2,...,k}; a quasi-static environment is considered, the height of the positions of all user mobile devices in the system is 0, and the positions do not change in the considered time slot; M UAVs are deployed in the air close to the user, that is, UAVs are used as communication relays and edge servers, and the set of UAVs is represented as M = {1,2,...,m}, and each UAV can cache content of size Sbits; consistent with the user mobile devices, the height of each UAV is set to H, and the deployment position of the UAV does not change in the considered time slot; a three-dimensional Euclidean coordinate system is used to represent the deployment positions of user mobile devices and UAVs, where the position coordinates of user mobile device k are defined as Accordingly, the position coordinates of UAVm are set to Define a finite content library as F = {1,2,...,f}, where the size of each content is fixed to sbits, and the content requested by the user is included in the content library; define a binary variable a m,k represents the connection relationship between user equipment k and UAVm, where a m,k =1 indicates that user equipment k is associated with UAVm, otherwise a m,k = 0; a user device can only be associated with one UAV, but a UAV can be associated with multiple user devices; in addition, define a binary variable q k,f represents f requests for content from user device k, where q k,f =1 indicates that user device k requests content f, otherwise q k,f =0; let u m,f =1 means that there is content f cached in UAVm, otherwise u m,f =0;

[0007] In the UAV-assisted edge computing system, the UAV is equipped with a cache storage device with limited capacity; if the content requested by the user has been cached in its associated UAV, the content will be directly transmitted to the user through the UAV; otherwise, the requested content will first be obtained from the core network by its associated UAV through the ground base station, and then transmitted to the user through the UAV; the content cached by the UAV is recorded in the content library, which consists of a limited number of different contents; each content can be cached in different UAVs, but can only be cached once in the same UAV;

[0008] The distance d between drone m and user device k m,k It is expressed as:

[0009]

[0010] The free space path loss model is used to simulate the wireless channel between UAVm and user equipment k. The wireless channel link power gain is calculated by the FSPL model as follows:

[0011]

[0012] Among them, d m,k represents the spatial distance between UAV m and user equipment k, g0 represents the channel power gain per unit spatial distance;

[0013] In the UAV-assisted edge computing system model, the transmission delay is divided into two parts, namely, the transmission delay between the UAV and the user equipment and the transmission delay between the UAV and the ground base station;

[0014] When calculating the transmission rate between UAV and user equipment, in order to minimize the transmission delay, UAVm always uses its maximum transmission power P for data transmission to improve the signal-to-noise ratio of the communication link; therefore, the transmission rate r between UAVm and user equipment k is m,k It is expressed as:

[0015]

[0016] Among them, B and σ 2 represent the channel bandwidth and white Gaussian noise (WGN) power in the communication link respectively;

[0017] The transmission delay from UAVm to user equipment k is further expressed as:

[0018]

[0019] When calculating the transmission rate between the UAV and the ground base station, since the distance between the ground base station and the UAV is much greater than the distance between the UAV and the user equipment, the transmission rate between the UAV and the ground base station is set to a fixed value Rs; the transmission delay from the ground base station to the UAVm is expressed as:

[0020]

[0021] If the content requested by user device k is already cached in the associated UAV m, the user device directly obtains the requested content from the associated UAV, so the transmission delay between the base station and the UAV is not considered. m,f =1,

[0022] The transmission delay required for the user to request the content is recorded as:

[0023]

[0024] Furthermore, in the UAV-assisted edge computing system, by determining the deployment position of the UAV {P UAV} and the content caching scheme {u} to minimize the average transmission delay of all user requests in the system; the joint optimization problem is formally expressed as:

[0025]

[0026] st

[0027]

[0028] Among them, constraints C1 and C2 restrict the range of locations where the UAV can be deployed; C3 represents the cache capacity constraint of the UAV.

[0029] Furthermore, the joint optimization problem of drone deployment and content caching is a complex mixed integer programming problem. The content distribution method is based on a genetic algorithm to deploy drones and a greedy strategy to cache content, specifically:

[0030] First, encode the individuals, initialize the population, and initialize the relevant variables;

[0031] After initialization, for each individual, drone content caching is performed based on a greedy strategy according to the user device location, requested content, and drone deployment location;

[0032] Then calculate the fitness function value and find the individual with the best fitness function value; in each iteration, compare whether the best fitness value is the same as the last one. sameWhen the algorithm reaches the maximum number of iterations, it is considered that the algorithm has converged and the iteration ends. If the algorithm has not yet converged, genetic operations are performed, including selection, crossover, and mutation operations, until the algorithm converges or reaches the maximum number of iterations.

[0033] The method for caching drone content based on greedy strategy is:

[0034] When the distance between the UAV and the user device is closer, the power gain of the wireless channel link is greater, and the transmission delay of the user's requested content is smaller; however, the cache size on the UAV is limited, and it is impossible to cache all content. Therefore, more appropriate content is selectively cached in the limited capacity: first, each user device calculates the distance between the M deployed UAVs, thereby obtaining the UAV closest to it and associating the user device with the UAV; then the UAV records the type of content requested by the user and its number of times; finally, the UAV obtains and caches the content from the ground base station in the order of the number of requests from most to least based on the list of requested content it has recorded, if the cache capacity is not full.

[0035] Furthermore, in the process of UAV deployment based on genetic algorithm, the specific method of individual coding is as follows:

[0036] The solution required by the problem is the deployment location plan of M UAVs. Considering that the flight altitudes of multiple UAVs are consistent, the deployment plan of UAVs is represented by a set of two-dimensional coordinate sequences; therefore, a solution P UAV It is expressed as:

[0037]

[0038] in, is the deployment position of UAVj, expressed as:

[0039]

[0040] The solution to the problem is encoded as an individual in the genetic algorithm. Since the height of each UAV is set to a fixed value, the individual encoding only requires the horizontal and vertical coordinate information of the UAV position. In order to ensure that the encoded solution is meaningful and still within the feasible domain after the subsequent mutation operation, the real number encoding method is adopted to encode the x value of each UAV. j and j The value is used as the chromosome information of the individual.

[0041] Furthermore, in the process of UAV deployment based on genetic algorithm, the inverse of the average transmission delay is used as the fitness function.

[0042] Compared with the prior art, the present invention has the following beneficial effects: a content distribution method in a multi-UAV assisted edge system is proposed, which optimizes the deployment of UAVs based on a genetic algorithm, and optimizes the content cache of UAVs using a greedy strategy, thereby obtaining a better UAV deployment plan and content cache plan. The UAV deployment plan and content cache plan obtained by this method can reduce the average transmission delay of user request content, so that mobile users have a better experience quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of the method implementation of the embodiment of the present invention.

[0044] Figure 2 is the average transmission delay under different numbers of user equipments in the embodiment of the present invention.

[0045] Figure 3 is the average transmission delay under different numbers of drones in the embodiment of the present invention.

[0046] Figure 4 is the average transmission delay under different UAV cache capacities in the embodiment of the present invention. DETAILED DESCRIPTION

[0047] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0048] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0049] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0050] This embodiment provides a content distribution method in a multi-drone assisted edge computing system. First, a multi-drone assisted edge computing system model is established, and the joint optimization problem of drone deployment and content caching is defined. Then, with the goal of minimizing the average response delay of content requests from each user in the system, a genetic algorithm and a greedy strategy are combined to solve the joint optimization problem to obtain the optimal drone deployment location and content caching solution.

[0051] 1 System Model

[0052] In the caching-supported drone-assisted edge computing system, the set of ground user mobile devices is defined as K = {1, 2, ..., k}. In order to simplify the model, the present invention considers a quasi-static environment, where the height of all user mobile devices in the system is 0, and the position does not change within the considered time slot. M drones are deployed in the air close to the user, that is, UAVs serve as communication relays and edge servers, and the set of drones is represented as M = {1, 2, ..., m}, and each UAV can cache content of size Sbits. Consistent with the user mobile device, the height of each UAV is set to H, and the deployment position of the UAV does not change within the considered time slot. For ease of analysis, the present invention uses a three-dimensional Euclidean coordinate system to represent the deployment positions of user mobile devices and drones, where the position coordinates of user mobile device k are defined as Corresponding UAV m The position coordinates are set to Define a finite content library as F = {1, 2, ..., f}, where the size of each content is fixed to sbits, and the content requested by the user is included in the content library. Define a binary variable a m,k represents the connection relationship between user equipment k and UAVm, where a m,k =1 indicates that user equipment k is associated with UAVm, otherwise a m,k = 0. A user device can only be associated with one UAV, but a UAV can be associated with multiple user devices. In addition, define a binary variable q k,f represents f requests for content from user device k, where q k,f =1 indicates that user device k requests content f, otherwise q k,f =0. Let u m,f =1 means that there is content f cached in UAVm, otherwise u m,f =0.

[0053] In the UAV-assisted edge computing system, the UAV is equipped with a cache storage device with limited capacity. If the content requested by the user has been cached in its associated UAV, the content will be transmitted directly to the user via the UAV. Otherwise, the requested content will first be obtained from the core network by its associated UAV through the ground base station, and then transmitted to the user by the UAV. The content cached by the UAV is recorded in the content library, which consists of a limited number of different contents. Each content can be cached in different UAVs, but can only be cached once in the same UAV.

[0054] The distance d between drone m and user device k m,k It can be expressed as:

[0055]

[0056] In the UAV-assisted edge computing system model, since the UAV is deployed at a higher altitude than the user device, the channel loss of the communication link between the UAV and the user device becomes more significant. Therefore, the Free Space Path Loss (FSPL) model is used to simulate the wireless channel between UAVm and user device k. The wireless channel link power gain is calculated by the FSPL model and expressed as:

[0057]

[0058] Among them, d m,k represents the spatial distance between UAV m and user equipment k, and g0 represents the channel power gain per unit spatial distance.

[0059] In the UAV-assisted edge computing system model, the transmission delay is divided into two parts: the transmission delay between the UAV and the user equipment and the transmission delay between the UAV and the ground base station.

[0060] When calculating the transmission rate between UAV and user equipment, in order to minimize the transmission delay, UAVm always uses its maximum transmission power P for data transmission to improve the signal-to-noise ratio of the communication link. Therefore, the transmission rate r between UAVm and user equipment k is m,k It is expressed as:

[0061]

[0062] Among them, B and σ 2 represent the channel bandwidth and white Gaussian noise (WGN) power in the communication link respectively.

[0063] The transmission delay from UAVm to user equipment k is further expressed as:

[0064]

[0065] When calculating the transmission rate between the UAV and the ground base station, since the distance between the ground base station and the UAV is much greater than the distance between the UAV and the user equipment, in order to simplify the model, the present invention sets the transmission rate between the UAV and the ground base station to a fixed value Rs. The transmission delay from the ground base station to the UAVm is expressed as:

[0066]

[0067] It should be noted that if the content requested by user device k has been cached in the associated UAV m, the user device can directly obtain the requested content from the associated UAV, so there is no need to consider the transmission delay between the base station and the UAV. m,f =1,

[0068] Based on the above analysis, the transmission delay required for user request content is recorded as:

[0069]

[0070] 2 Problem Definition

[0071] In the UAV-assisted edge computing system, by determining the deployment position of the UAV {P UAV The joint optimization of {u} and the content caching solution {u} minimizes the average transmission delay of all user requests in the system. The joint optimization problem can be formally expressed as:

[0072]

[0073] st

[0074]

[0075] Among them, constraints C1 and C2 restrict the range of locations where the UAV can be deployed. C3 represents the cache capacity constraint of the UAV.

[0076] 3 Method design

[0077] 3.1 Individual coding

[0078] The problem solved by the present invention requires a solution to the deployment location of M UAVs. Considering that the flight altitudes of the multiple UAVs are consistent, the deployment plan of the UAVs can be represented by a set of two-dimensional coordinate sequences. UAV It is expressed as:

[0079]

[0080] in, is the deployment position of UAVj, expressed as:

[0081]

[0082] The solution to the problem is encoded as an individual in the genetic algorithm. Since the height of each UAV is set to a fixed value, the individual encoding only requires the horizontal and vertical coordinate information of the UAV position. In order to ensure that the encoded solution is meaningful and still within the feasible domain after subsequent mutation operations, real number encoding is adopted to encode the x value of each UAV. j and j The value is used as the chromosome information of the individual.

[0083] 3.2 Fitness Function

[0084] The fitness function needs to accurately evaluate the quality of individuals. Equation (2-1) represents the optimization goal of solving the problem in the present invention, that is, to minimize the average transmission delay of all user requests within the system. According to the rule of survival of the fittest in the genetic algorithm, the smaller the average transmission delay of a user request, the larger the fitness value of the individual should be. For the convenience of implementing the subsequent selection operation method, the present invention selects the reciprocal of the average transmission delay as the fitness function.

[0085] 3.3 Selection operation

[0086] The selection operation is to screen out excellent individuals in the population, but it does not mean that it is completely based on the size of the fitness. If only the size of the fitness value is used to screen individuals, it may lead to the rapid convergence of the algorithm and the premature phenomenon of falling into the local optimum. In this algorithm, the roulette wheel method is used for individual selection, and the selection probability of an individual is positively correlated with the numerical value of its fitness function. Calculate the fitness function value f(x i ) of each individual, and calculate the sum of the fitness function values of the population which is the probability that the individual is selected. At the same time, the elite retention strategy is used to retain the individual with the best fitness value in the existing population in each selection.

[0087] 3.4 Crossover operation

[0088] The crossover operation can maintain the diversity of the population and has a certain probability of generating better individuals. Traverse the individuals in the population, take it as the first parent individual P1, and then randomly select an individual from the population as the second parent individual P2, and then decide whether to perform the crossover operation according to a certain probability. If the crossover is performed, the two-point crossover strategy is adopted, and two random numbers x1 and x2 are selected within a given range, where 0 << x1 < x2 < L, and L represents the length of the individual chromosome. Exchange the chromosome part of the P2 individual between the random numbers to the corresponding position of the P1 individual to obtain a new individual. Calculate the fitness function value of the newly generated individual. If its value is better than P1, then P1 will be replaced by the new individual.

[0089] 3.5 Mutation operation

[0090] Mutation operations can also enrich the diversity of the population and help genetic algorithms jump out of the scope of local search and better achieve global search. Like the crossover operation, the mutation operation does not necessarily occur, but there is a certain probability to choose whether to mutate. Traverse the individuals of the population one by one, and through the strategy of single-point mutation, randomly generate a number x as the individual chromosome mutation position within a given range, where 0≤x<L. The value of the chromosome after mutation at position x is also randomly generated. Compare the fitness value of the new individual with that of the parent individual, and the individual with a high fitness value will be retained.

[0091] 3.6 Method flow

[0092] The problem solved by this method is a joint optimization problem involving UAV deployment and content caching, which is a complex mixed integer programming problem. In order to better solve the problem, the present invention decomposes it into two sub-problems and proposes corresponding algorithms, namely, UAV deployment based on genetic algorithm and content caching based on greedy strategy:

[0093] (1) UAV deployment algorithm based on genetic algorithm;

[0094] (2) Content caching algorithm based on greedy strategy.

[0095] The implementation process of this method is as follows Figure 1 Among them, the UAV deployment algorithm based on genetic algorithm is shown in Algorithm 1.

[0096]

[0097]

[0098] First, according to the input content, the population is initialized through the above-mentioned individual encoding method, and the relevant variables are initialized (line 1). After the initialization is completed, for each individual, it is necessary to cache the drone content according to the greedy strategy based on the user device location, request content and drone deployment location, that is, Algorithm 2. Then the fitness function value can be calculated (lines 2-6). At this time, it is necessary to find the individual with the best fitness function value. In each iteration, it is necessary to compare whether the best fitness value is the same as the previous one. When the number of identical times reaches Iteration, same When the algorithm reaches the maximum number of iterations, it is considered that the algorithm has converged and the iteration ends (lines 7-16). If the algorithm has not converged, genetic operations are required, including selection, crossover, and mutation operations, until the algorithm converges or reaches the maximum number of iterations (lines 17-20).

[0099] In order to obtain a complete solution to Algorithm 1, it is also necessary to arrange the content cache of the UAV reasonably. In order to quickly solve this problem, the present invention proposes a low-complexity greedy strategy to optimize the UAV content cache.

[0100] The greedy strategy is: always make the best choice in the current state. When the distance between the UAV and the user device is closer, the power gain of the wireless channel link will be greater, and the transmission delay of the user's requested content will be smaller. However, the cache size on the UAV is limited, and it is impossible to cache all content, so it is necessary to selectively cache more appropriate content in the limited capacity. To address the above problems, a content caching algorithm based on the greedy strategy is proposed.

[0101] The UAV content caching algorithm based on the greedy strategy is shown in Algorithm 2.

[0102]

[0103] First, each user device needs to calculate the distance between the deployed UAVs, and then find the UAV closest to it, and associate the user device with the UAV. Then the UAV needs to record the type of content requested by the user and its number of times (lines 1-13). Finally, the UAV obtains and caches the content from the ground base station in the order of the number of requests from most to least based on the requested content list it has recorded, if the cache capacity is not full (lines 14-19).

[0104] 4 Method evaluation

[0105] 4.1 Experimental Environment

[0106] In the simulation experiment, a UAV-assisted mobile edge computing system supporting high-speed caching was built, and the simulation environment was set to an area of ​​1000m×1000m, in which a certain number of user devices and UAVs were distributed. The location of the user device was randomly generated, and the user's request content was randomly obtained from the content library. The specific parameter settings are shown in Table 1.

[0107] Table 1 Simulation experiment parameter settings

[0108]

[0109] 4.2 Comparison of algorithm settings

[0110] In order to verify the feasibility of the present method, this embodiment rewrites the following three benchmark algorithms to adapt them to the problems raised by the present invention, and designs sufficient comparative experiments.

[0111] Random algorithm: One of the most commonly used comparison algorithms, which uses a random method to implement UAV deployment and adopts the greedy strategy proposed in this invention to implement content caching.

[0112] K-Means algorithm: a typical clustering algorithm, which is one of the commonly used UAV deployment optimization algorithms. The K-Means algorithm is used to implement UAV deployment, and the greedy strategy proposed in the present invention is used to implement content caching.

[0113] PSO algorithm: a classic swarm intelligence algorithm, one of the commonly used server placement algorithms. The PSO algorithm is used to implement UAV deployment, and the greedy strategy proposed in the present invention is used to implement content caching.

[0114] 4.3 Experimental Results Analysis

[0115] In order to better compare the proposed algorithm with the other three benchmark algorithms, the present invention adopts the method of controlling variables to conduct control experiments on variables such as the number of user devices, the number of drones, and the cache capacity of drones. Multiple repeated experiments were conducted for each case, and the average value was taken to improve the stability of the experimental results.

[0116] Figure 2 The impact of different numbers of user devices on the algorithm performance is considered. The number of user devices varies between 60 and 140, and other parameters remain the same. The number of drones is set to 10, and the drone cache capacity is 100Mbits. Figure 2 As shown in the figure, when the number of user devices continues to increase, the average transmission delay of user request content obtained by the method proposed in the present invention and the three benchmark algorithms will also increase. This is because when the number of user devices increases, the corresponding number of request content will also increase, and the cache space of the drone is limited, so there are more request content that cannot be cached. At this time, there are more content transmission delays composed of two parts, namely the transmission delay between the drone and the user device and the transmission delay between the drone and the ground base station, so the average transmission delay also increases.

[0117] Figure 3 The impact of different numbers of drones on algorithm performance is considered. The number of drones varies between 8 and 12, and other parameters remain the same. The number of user devices is set to 100, and the drone cache capacity is set to 100 Mbits. Figure 3 As shown in the figure, when the number of drones increases, the average transmission delay of user request content of the four algorithms will decrease. This is because when the number of drones increases and the number of user devices remains unchanged, the number of user devices associated with each drone decreases, the user request content is more likely to be cached by the drone, and the distance between the drone and the user device will also decrease, so the average transmission delay will decrease.

[0118] Figure 4The impact of different UAV cache capacities on the algorithm performance is considered. The UAV cache capacity varies between 60 and 140 Mbits, and other parameters remain the same. The number of user devices is set to 100 and the number of UAVs is set to 10. Figure 4 As shown in the figure, when the cache capacity of the drone increases, the average transmission delay of user request content of the four algorithms will decrease. This is because when the cache capacity of the drone increases, the drone can cache more types of content, and more user request content can be obtained directly from the drone edge server, so the average transmission delay will decrease.

[0119] according to Figure 2-4 It can be seen that under various conditions with different variables such as the number of user devices, the number of drones, and the cache capacity of drones, this method is superior to the three benchmark algorithms in optimizing the average transmission delay and has good performance. The random algorithm does not consider the distribution of users and the content of requests when deploying UAVs. It only caches content according to the greedy strategy after deployment. Therefore, the optimization effect of average transmission delay is not as good as the other three algorithms. The K-Means algorithm deploys UAVs according to the location of user devices, but it treats UAV deployment and content caching as two independent stages, ignoring the interaction between UAV deployment location and cache content. In addition, the K-Means algorithm may choose a discrete location for user device distribution as the UAV deployment location, resulting in uneven utilization of UAV resources in different areas. Although the PSO algorithm considers the interaction between UAV deployment and content caching, due to the distribution of user device locations, there may be multiple local areas suitable for drone deployment. At this time, the PSO algorithm is prone to fall into a local optimal state. Compared with the three benchmark algorithms, the genetic algorithm proposed in the present invention not only fully considers the interaction between UAV deployment and content caching, but also uses crossover and mutation operations to give the algorithm a strong global search capability. Therefore, it performs best in optimizing the average transmission delay, and its performance is improved by approximately 19.9%, 4.9%, and 13.1% compared with the random algorithm, K-Means algorithm, and PSO algorithm, respectively.

[0120] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0121] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0122] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0123] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0124] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A content distribution method in a multi-drone assisted edge computing system, characterized in that: Firstly, a multi-UAV-assisted edge computing system model is established, and the joint optimization problem of UAV deployment and content caching is defined. Then, with the goal of minimizing the average response delay of content requests from each user in the system, the joint optimization problem is solved by combining genetic algorithm and greedy strategy to obtain the optimal UAV deployment location and content caching solution. In the caching-supported drone-assisted edge computing system, the set of ground user mobile devices is defined as K = {1, 2, ..., k}; a quasi-static environment is considered, the height of the positions of all user mobile devices in the system is 0, and the positions do not change in the considered time slot; M drones are deployed in the air close to the user, that is, UAVs are used as communication relays and edge servers, and the drone set is represented as M = {1, 2, ..., m}, and each UAV can cache content of size Sbits; consistent with the user mobile device, the height of each UAV is set to H, and the UAV deployment position does not change in the considered time slot; a three-dimensional Euclidean coordinate system is used to represent the deployment positions of user mobile devices and drones, where the position coordinates of user mobile device k are defined as Correspondingly, the position coordinates of UAVm are set to Define a finite content library as F = {1, 2, ..., f}, where the size of each content is fixed to sbits, and the content requested by the user is included in the content library; define a binary variable a m,k represents the connection relationship between user equipment k and UAVm, where a m，k =1 indicates that user equipment k is associated with UAVm, otherwise a m，k = 0; a user device can only be associated with one UAV, but a UAV can be associated with multiple user devices; in addition, define a binary variable q k，f represents f requests for content from user device k, where q k，f =1 indicates that user device k requests content f, otherwise q k，f =0; let u m，f =1 means that there is content f cached in UAVm, otherwise u m，f =0; In the UAV-assisted edge computing system, the UAV is equipped with a cache storage device with limited capacity; If the content requested by the user has been cached in the associated UAV, the content will be directly transmitted to the user via the UAV; otherwise, the requested content will first be obtained from the core network by the associated UAV via the ground base station, and then transmitted to the user via the UAV; the content cached by the UAV is recorded in the content library, which consists of a limited number of different content; each content can be cached in different UAVs, but can only be cached once in the same UAV; The distance d between drone m and user device k m,k It is expressed as: The free space path loss model is used to simulate the wireless channel between UAVm and user equipment k. The wireless channel link power gain is calculated by the FSPL model as follows: Among them, d m,k represents the spatial distance between UAV m and user equipment k, g0 represents the channel power gain per unit spatial distance; In the UAV-assisted edge computing system model, the transmission delay is divided into two parts, namely, the transmission delay between the UAV and the user equipment and the transmission delay between the UAV and the ground base station; When calculating the transmission rate between UAV and user equipment, in order to minimize the transmission delay, UAVm always uses its maximum transmission power P for data transmission to improve the signal-to-noise ratio of the communication link; therefore, the transmission rate r between UAVm and user equipment k is m,k It is expressed as: Among them, B and σ 2 represent the channel bandwidth and white Gaussian noise (WGN) power in the communication link respectively; The transmission delay from UAVm to user equipment k is further expressed as: When calculating the transmission rate between the UAV and the ground base station, since the distance between the ground base station and the UAV is much greater than the distance between the UAV and the user equipment, the transmission rate between the UAV and the ground base station is set to a fixed value Rs; the transmission delay from the ground base station to the UAVm is expressed as: If the content requested by user equipment k is already cached in the associated UAV m, the user equipment directly obtains the requested content from the associated UAV, so the transmission delay between the base station and the UAV is not considered. m,f =1, The transmission delay required for the user to request the content is recorded as: In the UAV-assisted edge computing system, by determining the deployment position of the UAV {P UAV } and the content caching scheme {u} to minimize the average transmission delay of all user requests in the system; the joint optimization problem is formally expressed as: Among them, constraints C1 and C2 restrict the range of locations where the UAV can be deployed; C3 represents the cache capacity constraint of the UAV; The joint optimization problem of drone deployment and content caching is a complex mixed integer programming problem. The content distribution method is based on genetic algorithm for drone deployment and greedy strategy for content caching, specifically: First, encode the individuals, initialize the population, and initialize the relevant variables; After initialization, for each individual, drone content caching is performed based on a greedy strategy according to the user device location, requested content, and drone deployment location; Then calculate the fitness function value and find the individual with the best fitness function value; in each iteration, compare whether the best fitness value is the same as the last one. same When the algorithm reaches the maximum number of iterations, it is considered that the algorithm has converged and the iteration ends. If the algorithm has not yet converged, genetic operations are performed, including selection, crossover, and mutation operations, until the algorithm converges or reaches the maximum number of iterations. The method for caching drone content based on greedy strategy is: When the distance between the UAV and the user device is closer, the power gain of the wireless channel link is greater, and the transmission delay of the user's requested content is smaller; however, the cache size on the UAV is limited, and it is impossible to cache all content, so more appropriate content is selectively cached in the limited capacity: first, each user device calculates the distance between the M deployed UAVs, thereby obtaining the UAV closest to it, and associates the user device with the UAV; then the UAV records the type of content requested by the user and its number of times; finally, based on the requested content list recorded by itself, the UAV obtains and caches the content from the ground base station in the order of the number of requests from most to least, if the cache capacity is not full; In the process of UAV deployment based on genetic algorithm, the specific method of individual coding is as follows: The solution required by the problem is the deployment location plan of M UAVs. Considering that the flight altitudes of multiple UAVs are consistent, the deployment plan of UAVs is represented by a set of two-dimensional coordinate sequences; therefore, a solution P UAV It is expressed as: in, is the deployment position of UAVj, expressed as: The solution to the problem is encoded as an individual in the genetic algorithm. Since the height of each UAV is set to a fixed value, the individual encoding only requires the horizontal and vertical coordinate information of the UAV position. In order to ensure that the encoded solution is meaningful and still within the feasible domain after the subsequent mutation operation, the real number encoding method is adopted to encode the x value of each UAV. j and j The value is the chromosome information of the individual; In the process of UAV deployment based on genetic algorithm, the inverse of the average transmission delay is used as the fitness function.

Images