Multi-unmanned aerial vehicle cooperative mobile edge computing network computing unloading method and system under fixed trajectory

Abstract

The invention belongs to the technical field of communication, and relates to a multi-unmanned aerial vehicle cooperative mobile edge computing network computing unloading method and system under a fixed trajectory, and the method comprises the steps: initializing user position information, channel power gain and bandwidth; calculating local time delay, transmission time delay and processing time delay by combining local and unmanned aerial vehicle end calculation resources; calculating the waiting time delay of the user equipment according to the flight path of the unmanned aerial vehicle; and calculating and outputting an optimal unloading strategy by using a simulated annealing differential evolution fusion algorithm. Effective communication with the ground terminal can be maintained in a harsh environment, and the task processing delay of the ground equipment is reduced.

Description

Technical Field

[0001] The present invention belongs to the field of communication technologies, and designs a drone-assisted mobile edge computing technology, specifically a method and system for computing offloading in a multi-drone cooperative mobile edge computing network with a fixed trajectory. Background Art

[0002] With the development of 5G networks, emerging Internet applications such as AR / VR, vehicle-to-everything (V2X), and remote medical surgery have rapidly spread, generating a large amount of task data with high computational latency requirements at user terminals. However, the computing capabilities of Internet of Things (IoT) devices are relatively limited, making it difficult to process these tasks in a short time. Mobile edge computing (MEC), as a solution, reduces energy consumption and latency by offloading computing tasks to servers at the network edge. However, terrestrial MEC is affected by the multipath effect of non-line-of-sight paths, resulting in limited communication quality and insufficient service flexibility due to its fixed location.

[0003] Considering the flexible mobility and easy deployment characteristics of drones (UAVs), they have become a new carrier for MEC. Based on this, to further improve the reliability and efficiency of task offloading, especially to ensure the continuous connection between users and the network under extreme conditions, the present invention introduces an innovative drone-assisted mobile edge computing method and system to overcome the above problems existing in the prior art. Summary of the Invention

[0004] To solve the above problems existing in the prior art, the present invention designs a method and system for computing offloading in a multi-drone cooperative mobile edge computing network with a fixed trajectory. The present invention can maintain effective communication with ground terminals in a harsh environment, aiming to reduce the task processing latency of user equipment.

[0005] The application scenario of the present invention is as follows: User equipment (UE) is fixedly distributed on a circle with a radius of R. Each drone is equipped with a mobile edge computing (MEC) server, which can be applied to task offloading in remote areas, complex terrains, and working environments with a wide and scattered distribution of devices. Among them, U drones cruise around a circle with a radius of R at a speed of v, and the coverage range of the MEC server installed on each drone is a circle with a radius of r.

[0006] The present invention adopts the following technical solutions:

[0007] The steps of a method for computing offloading in a multi-drone cooperative mobile edge computing network with a fixed trajectory are as follows:

[0008] S1: Initialize the system model: Initialize user location information, channel power gain, bandwidth, etc.;

[0009] S2: Calculate the offloading delay: Combine the local and drone-side computing resources to calculate the local delay, transmission delay, and processing delay;

[0010] S3: Calculate the waiting delay: Calculate the waiting delay of the user equipment according to the UAV flight trajectory;

[0011] S4: Calculate the optimal strategy: Use the simulated annealing differential evolution fusion algorithm to calculate and output the optimal offloading strategy.

[0012] Preferably, in step S1, initialize the system model as follows:

[0013] The system objective is to minimize the computing delay T, and the optimization problem is modeled as:

[0014] T = max{T L , max{T UAV + T TX , a(T W + T MAX + T UAV' + T TX' )}}

[0015] The constraints include:

[0016] β = β1 + aβ2

[0017] (T UAV + T TX )| a＝1 < 2T MAX + T W

[0018] (T UAV' + T TX' )| a＝1 < T MAX

[0019] (T UAV + T TX )| a＝0 < T mcom

[0020] Where max{a, b} represents the larger value between a and b, T L is the local offloading delay, T UAV , T TX are the processing delay and transmission delay of the current UAV respectively, T UAV' , T TX' are the processing delay and transmission delay of the next UAV respectively, T W represents the longest waiting time of the UE, T MAXDenote the upper limit of the communication time between the UE and the UAV. Let a be a binary variable, where a = 0 means the UE only offloads the task to the current UAV, and a = 1 means the UE offloads the task to the current and the next UAV. Let β be the total offloading ratio of the UE, β1 be the offloading ratio to the current UAV, and β2 be the offloading ratio to the next UAV.

[0021] Preferably, in step S2, calculate the offloading delay as follows: Calculate the local delay, transmission delay, and processing delay by combining the local and UAV-side computing resources.

[0022] The tuple {D, F, β, a} represents the key parameters of the task offloading strategy. Among them, D is the task volume, F is the number of CPU cycles required to process a single bit, β represents the offloading ratio, and a represents the offloading strategy. (x UE , y UE ) represents the planar coordinates of the UE, and the three-dimensional coordinates of each UAV are denoted as (X UAV , Y UAV , H UAV );

[0023] l represents the planar distance between the UE and the UAV:

[0024]

[0025] g represents the channel gain of the task offloading between the UE and the UAV:

[0026]

[0027] Among them, H UAV represents the UAV flight altitude, β0 represents the channel power gain at a reference distance of 1 m; r represents the uplink transmission rate from the UE to the UAV:

[0028]

[0029] Among them, p represents the transmission power from the UE to the UAV in watts; B represents the channel bandwidth in hertz, and the bandwidth between the UAV and the UEs within its coverage is equal. N0 represents the noise power spectral density in watts per hertz.

[0030] The local delay T L is expressed as:

[0031]

[0032] Among them, β is the offloading ratio that the UE selects to offload the task to the UAV, D represents the task data size, F L is the number of CPU cycles required for the user equipment to calculate one-bit task, f LIt is the computing power of the UE's CPU.

[0033] When calculating the processing delay and transmission delay, the offloading strategy a needs to be considered;

[0034] a) When a = 0, the processing delay T of the current UAV UAV is expressed as:

[0035]

[0036] where β1 represents the offloading ratio offloaded to the current UAV, and F UAV represents the number of CPU cycles required for the current UAV to calculate one bit, and f UAV represents the computing power of the current UAV;

[0037] The transmission delay T between the UE and the current UAV TX is expressed as:

[0038]

[0039] b) When a = 1, the delay T generated by the next UAV processing this task UAV' can be expressed as:

[0040]

[0041] where β2 represents the offloading ratio offloaded by the UE to the next UAV, and F UAV' represents the number of CPU cycles required for the next UAV to calculate one bit, and f UAV' represents the computing power of the next UAV;

[0042] The transmission delay T between the UE and the next UAV TX' is:

[0043]

[0044] Preferably, in step S3, calculate the waiting delay, specifically as follows: According to the UAV flight trajectory, calculate the waiting delay of the user equipment;

[0045] T MAX represents the communication time limit between the UE and the UAV:

[0046]

[0047] where r represents the coverage radius of the UAV, R is the flight radius of the UAV, and v represents the flight speed of the UAV;

[0048] T W represents the longest waiting time of the UE:

[0049]

[0050] Among them, U is the number of UAVs, R is the flight radius of the UAV, and r is the communication radius of the UAV;

[0051] Preferably, in step S4, calculating the optimal strategy is specifically as follows: using the simulated annealing differential evolution fusion algorithm to calculate and output the optimal offloading strategy;

[0052] In step S41, randomly generate a vector within the feasible region as the initial population. Set parameters including the maximum number of iterations G m , the population size Np, the feasible region Z, the mutation rate F0, the crossover probability CR, the initial temperature T0, the cooling rate ρ, the termination temperature T end , and the delay function f(x). In the feasible region, randomly generate a vector set X containing 3N dimensions as the initial population;

[0053] In step S42, calculate the mutation rate F corresponding to the current number of iterations;

[0054]

[0055] F = F0·2 λ

[0056] where G is the current number of iterations;

[0057] In step S43, calculate the mutant y corresponding to the i-th individual according to the following formula for the mutation rate i :

[0058] y i = x i - F·(x j - x k )

[0059] where x j , x k are two different individuals randomly selected from the population;

[0060] In step S44, cross-mutate y i with the parent x i ;

[0061]

[0062] where rand is a random number uniformly distributed in the interval [0,1]; CR represents the crossover probability;

[0063] In step S45, select the optimal individual according to the following formula:

[0064]

[0065] Step S46, generate a new candidate solution within the neighboring solution space m of the current solution:

[0066]

[0067] where m represents the radius of the neighboring solution space, which is used to control the generation range of the new candidate solution near the current solution;

[0068] Step S47, determine whether to accept the new solution according to the current temperature;

[0069] Δ i represents the difference between the current solution and the new solution:

[0070] Δ i = f(h i ) - f(k i )

[0071] γ represents the acceptance probability:

[0072]

[0073] Step S48, reduce the temperature proportionally and update the current temperature:

[0074] T = ρ · T

[0075] where ρ is the temperature reduction coefficient;

[0076] Step S49, determine whether the current temperature T reaches the minimum value. If not, return to Step S46; otherwise, proceed to the next step;

[0077] Step S410, increment the iteration count by one;

[0078] Step S411, determine whether the iteration count reaches the maximum value. If not, return to Step S42; otherwise, proceed to the next step;

[0079] Step S412, output the optimal strategy.

[0080] The present invention also discloses a multi-UAV collaborative mobile edge computing network computing offloading system under a fixed trajectory for executing the above method, including the following modules:

[0081] System model initialization module: Initialize the user location information, channel power gain, and bandwidth;

[0082] Offloading delay calculation module: Combine the local and UAV-side computing resources to calculate the local delay, transmission delay, and processing delay;

[0083] Waiting delay calculation module: Calculate the waiting delay of the user equipment according to the UAV flight trajectory;

[0084] Optimal Strategy Calculation Module: Using the simulated annealing differential evolution fusion algorithm, calculate and output the optimal offloading strategy.

[0085] The present invention has the following remarkable technical effects:

[0086] The present invention reduces the task processing delay: By dynamically optimizing the data offloading strategy of the user equipment (UE), allowing tasks to be offloaded to the current drone or the drone that is about to enter the communication range, and combining the collaborative computing resources of multiple drones, the overall computing efficiency is improved, thereby reducing the total system delay. Description of the Drawings

[0087] Figure 1 It is a flowchart of the computing offloading method for a multi-drone collaborative mobile edge computing network under a fixed trajectory according to a preferred embodiment of the present invention.

[0088] Figure 2 It is a three-dimensional schematic diagram of a drone-assisted mobile edge computing system according to a preferred embodiment of the present invention.

[0089] Figure 3 It is a top view of a drone-assisted mobile edge computing system according to an embodiment of the present invention.

[0090] Figure 4 It is a flowchart of the simulated annealing differential evolution fusion algorithm according to a preferred embodiment of the present invention.

[0091] Figure 5 It is a graph showing the relationship between the number of iterations and the optimal value of each generation according to a preferred embodiment of the present invention.

[0092] Figure 6 It is a block diagram of a computing offloading system for a multi-drone collaborative mobile edge computing network under a fixed trajectory according to a preferred embodiment of the present invention. Detailed Embodiments

[0093] The following is a detailed description of the preferred embodiments of the present invention.

[0094] Embodiment 1:

[0095] As Figure 1 shown, this embodiment discloses a computing offloading method for a multi-drone collaborative mobile edge computing network under a fixed trajectory, including the following:

[0096] Step S101, initialize the system model: Initialize the user location information, channel power gain, and bandwidth;

[0097] The system model is as Figures 2-3As shown in the figure, the drone-assisted network provides MEC services to support UEs in executing computationally intensive and latency-critical tasks. Among them, U drones cruise around a circle with a radius of R at a speed of v, and the coverage range of the MEC server installed on each drone is a circle with a radius of r. The UE is allowed to flexibly choose to offload data to the current drone or consider both the current and the upcoming next drone as data offloading targets. When the UE chooses the latter, the task results processed by the current drone can be transmitted to the subsequent drone via a data transmission link, and then the latter will send the task results back to the UE. This scheme aims to effectively utilize the time period when the UE waits for the subsequent drone, thereby avoiding unnecessary waste of resources. When the drone establishes a communication connection with the user equipment, the UE can choose to offload all or part of the task to the current or the current and the next drone, which is represented by the tuple {D, F, β, a}. Among them, D is the task volume, F is the number of CPU cycles required to process a single bit, β represents the offloading ratio, and a is a variable from 0 to 1. a = 0 means the device offloads the task only to the current drone, and a = 1 means the device offloads the task to the current and the next UAV. (x UE , y UE ) represents the planar coordinates of the UE, and the three-dimensional coordinates of each drone are denoted as (X UAV , Y UAV , H UAV );

[0098] Step S102: Combine the local and drone-side computing resources to calculate the local latency, transmission latency, and processing latency. The channel gain of task offloading cannot be regarded as a constant, and it is affected by the distance between the UE and the drone. The planar distance between the UE and the UAV can be calculated by the following formula:

[0099]

[0100] The channel gain of task offloading between the UE and the UAV is expressed as:

[0101]

[0102] Among them, β0 represents the channel power gain at a reference distance of 1m. The uplink transmission rate from the UE to the UAV is expressed as:

[0103]

[0104] Among them, B represents the channel bandwidth in Hz, and the bandwidth between the UAV and the UEs within its coverage range is equal. N0 represents the noise power spectral density in W / Hz. p represents the transmission power from the UE to the UAV in W;

[0105] The local latency T L can be expressed as:

[0106]

[0107] Among them, β is the offloading ratio of the UE to select to offload the task to the UAV, D represents the task data size, and F L is the number of CPU cycles required for the user equipment to calculate one-bit task, and f L is the computing power of the UE's CPU.

[0108] The offloading strategy a needs to be considered for calculating the processing delay and transmission delay.

[0109] a) When a = 0, that is, when the UE selects to offload only to the current UAV, the processing delay T of the current UAV UAV is expressed as:

[0110]

[0111] Among them, β1 represents the offloading ratio offloaded to the current UAV, and F UAV represents the number of CPU cycles required for the current UAV to calculate one bit, and f UAV represents the computing power of the current UAV.

[0112] Use T TX to represent the transmission delay between the UE and the current UAV, which can be expressed as:

[0113]

[0114] b) When a = 1, that is, when the UE decides to offload its task to the current UAV and the next upcoming UAV, the delay T generated by the next UAV processing the task UAV' can be expressed as:

[0115]

[0116] β2 represents the offloading ratio offloaded by the UE to the next UAV, and F UAV' represents the number of CPU cycles required for the next UAV to calculate one bit, and f UAV' represents the computing power of the next UAV.

[0117] The transmission delay T between the UE and the next UAV TX' is:

[0118]

[0119] Step S103, calculate the waiting delay of the user equipment according to the UAV flight trajectory:

[0120] Due to the mobility of the UAV, it cannot be guaranteed that the UE is always within the coverage area of the UAV. The communication time between the UE and the UAV has an upper limit, denoted by T MAX which represents the upper limit of the communication time between the UE and a UAV, and can be expressed as:

[0121]

[0122] where r represents the coverage radius of the UAV, R is the flight radius of the UAV, and v represents the flight speed of the drone.

[0123] The UAV flies around a circle with a radius of R at a fixed interval distance. Assuming the UAV flies clockwise, the maximum waiting time for the UE until the UAV can communicate with it is:

[0124]

[0125] where U is the number of drones, R is the flight radius of the UAV, and r is the communication radius of the drone;

[0126] Therefore, by optimizing the offloading drone selection a, the offloading ratio β, and the allocation of the offloading ratios of two drones β1 and β2 to minimize the system computation latency, this problem can be formulated as:

[0127] T = max{T L , max{T UAV + T TX , a(T W + T MAX + T UAV' + T TX' )}} (11)

[0128] The four constraint conditions included in the optimization model are respectively:

[0129] β = β1 + aβ2 (12)

[0130] (T UAV + T TX )| a＝1 <2T MAX + T W (13)

[0131] (T UAV' + T TX' )| a＝1 <T MAX (14)

[0132] (T UAV + T TX )| a＝0 <T mcom (15)

[0133] Among them, the constraint condition (12) ensures that the sum of the task offloading ratios borne by the current UAV and the next upcoming UAV is consistent with the total offloading ratio determined by the UE. The constraint conditions (13) and (14) indicate that when the UE chooses to offload tasks to two UAVs for processing, it must ensure that the computing delays generated by each of these two UAVs are less than the allowed maximum communication time, so as to ensure that the calculation results can be transmitted back to the UE in a timely manner. The constraint condition (15) stipulates that when the UE decides to communicate only with the current UAV, the computing delay of this UAV must be less than the preset maximum communication time threshold to ensure that the results of the computing tasks can be effectively transmitted back to the UE.

[0134] Step S104, use the simulated annealing differential evolution fusion algorithm to calculate and output the optimal offloading strategy;

[0135] As Figure 4 shown, the specific simulated annealing differential evolution fusion algorithm specifically includes:

[0136] First, perform the differential evolution algorithm. The differential evolution algorithm is mainly divided into four steps: initialization, mutation, crossover, and selection.

[0137] Step S1041, initialize the parameters and randomly generate vectors within the feasible region as the initial population;

[0138] The initialized parameters include the maximum number of iterations G m , population size Np, feasible region Z, mutation rate F0, crossover probability CR, initial temperature T0, cooling rate ρ, termination temperature T end , and the delay function f(x). In the feasible region, a vector set X containing 3N dimensions is randomly generated as the initial population.

[0139] Step S1042, calculate the mutation rate corresponding to the current iteration number;

[0140] First, obtain the mutation operator λ based on the current iteration number to determine the mutation rate F:

[0141]

[0142] F = F0·2 λ (17)

[0143] Among them, G is the current iteration number, and the range of the adaptive mutation rate is F0 to 2F0. The existence of the adaptive mutation rate avoids premature convergence of the algorithm.

[0144] Step S1043, calculate the mutant y corresponding to the i-th individual according to the following formula for the mutation rate i :

[0145] yi = x i - F·(x j - x k ) (18)

[0146] Step S1044, cross-mutate y i with the parent x i ;

[0147] Determine whether to select the mutant or its parent as the candidate solution in the subsequent genetic algorithm operations through the crossover rate CR:

[0148]

[0149] Step S1045, select the optimal individual according to the following formula:

[0150]

[0151] where f(x i ) is the time delay generated when executing the unloading strategy x i .

[0152] Since the differential evolution algorithm is prone to falling into local optima, simulated annealing is used to perturb the population to avoid premature convergence of the algorithm. The simulated annealing algorithm consists of two main loop structures: the outer loop and the inner loop. The outer loop simulates the annealing process, that is, starting from a higher initial temperature, gradually reducing the temperature according to a predetermined cooling coefficient until reaching the preset lowest temperature and then stopping. The inner loop iterates at each specific temperature level and decides whether to accept the newly generated solution according to a certain acceptance probability. Simply put, the simulated annealing algorithm can be divided into three steps: generating a new solution, accepting the new solution, and cooling down.

[0153] Step S1046, under the condition of maintaining the same temperature, explore and generate a new candidate solution within the neighborhood solution space m of the current solution according to the following formula:

[0154]

[0155] Step S1047, judge whether to accept the new solution according to the current temperature;

[0156] First, calculate the difference between the current solution and the new solution:

[0157] Δ i = f(h i ) - f(k i ) (22)

[0158] Then calculate the acceptance probability according to the current temperature and the difference value:

[0159]

[0160] If the new solution performs better, it is directly adopted. Otherwise, the following criteria are used to decide whether to accept the new solution:

[0161]

[0162] Step S1048, reduce the temperature proportionally and update the current temperature:

[0163] T = ρ·T(25)

[0164] where ρ is the temperature reduction coefficient.

[0165] Step S1049, determine whether the current temperature T has reached the minimum value. If not, return to step S46; otherwise, proceed to the next step.

[0166] Step S10410, increment the iteration count by one.

[0167] Step S10411, determine whether the iteration count has reached the maximum value. If not, return to step S42; otherwise, proceed to the next step.

[0168] Step S10412, output the optimal strategy.

[0169] The experimental results are as Figure 5 shown Figure 5 which shows the relationship between the number of algorithm iterations and the optimal value calculated for each generation within a specific area with a radius of 1000 meters when five drones are deployed. From this figure, it can be clearly observed that as the number of algorithm iterations gradually increases, the minimum latency of the offloading tasks calculated for each generation shows a decreasing trend. This decreasing process reflects the continuous optimization of the algorithm in computing and the search for a more efficient offloading strategy. Eventually, the algorithm reaches a convergence state, that is, the optimal value calculated for each generation tends to be stable and no longer changes significantly. This process fully demonstrates the effectiveness and convergence of the algorithm in solving the problems of drone deployment and task offloading.

[0170] Embodiment 2:

[0171] As Figure 6 shown, this embodiment discloses a multi - drone collaborative mobile edge computing network computing offloading system under a fixed trajectory for implementing the above - mentioned method, including the following modules:

[0172] System model initialization module: Initialize the user location information, channel power gain, and bandwidth;

[0173] Offloading latency calculation module: Combine the local and drone - side computing resources to calculate the local latency, transmission latency, and processing latency;

[0174] Waiting latency calculation module: calculate the waiting latency of the user equipment according to the flight trajectory of the UAV;

[0175] Optimal strategy calculation module: calculate and output the optimal offloading strategy by using the simulated annealing differential evolution fusion algorithm.

[0176] For other contents of this embodiment, reference can be made to the above embodiments.

[0177] In summary, the present invention designs a multi-UAV collaborative mobile edge computing network computing offloading method and system under a fixed trajectory, which can maintain effective communication with ground terminals in a harsh environment, aiming to reduce the task processing delay of user equipment.

[0178] The above description only details the preferred embodiments and principles of the present invention. For those of ordinary skill in the art, according to the idea provided by the present invention, there will be changes in the specific implementation manners, and these changes should also be regarded as the protection scope of the present invention.

Claims

1. A method for computing offloading in a multi-UAV collaborative mobile edge computing network under a fixed trajectory, characterized in that The specific steps are as follows: S1. Initialize the system model: Initialize the user location information, channel power gain, and bandwidth; S2. Calculate the offloading delay: Combine the local and UAV-side computing resources to calculate the local delay, transmission delay, and processing delay; S3. Calculate the waiting delay: Calculate the waiting delay of the user equipment UE according to the UAV flight trajectory; S4. Calculate the optimal strategy: Use the simulated annealing differential evolution fusion algorithm to calculate and output the optimal offloading strategy.

2. The multi-UAV collaborative mobile edge computing network computing offloading method under a fixed trajectory as claimed in claim 1, wherein, For step S1, specifically as follows: The system objective is to minimize the computing delay T, and the optimization problem is modeled as: T = max{T L , max{T UAV + T TX , a(T W + T MAX + T UAV' + T TX' )}} Constraints include: β＝β1+aβ2 (T UAV +T TX )| a＝1 <2T MAX +T W (T UAV' +T TX' )| a＝1 <T MAX (T UAV +T TX )| a＝0 <T mcom Among them, max{a, b} represents the larger value between a and b, and T L is the local offloading delay, and T UAV , T TX are the processing delay and transmission delay of the current UAV respectively, and T UAV' , T TX' are the processing delay and transmission delay of the next UAV respectively, and T W represents the maximum waiting time of the UE, and T MAX represents the upper limit of the communication time between the UE and the UAV. a is a 0-1 variable. a = 0 means the UE only offloads the task to the current UAV, and a = 1 means the UE offloads the task to the current and the next UAV. β is the total offloading ratio of the UE, β1 is the offloading ratio to the current UAV, and β2 is the offloading ratio to the next UAV.

3. The multi-UAV collaborative mobile edge computing network computing offloading method under a fixed trajectory according to claim 2, wherein, For step S2, specifically as follows: The tuple {D, F, β, a} represents the key parameters of the task offloading strategy; where D is the task volume, F is the CPU cycles required to process a single bit, β represents the offloading ratio, and a represents the offloading strategy; (x UE , y UE ) represents the planar coordinates of the UE, and the three-dimensional coordinates of each UAV are denoted as (X UAV , Y UAV , H UAV ); l represents the planar distance between the UE and the UAV; g represents the channel gain of the task offloading between the UE and the UAV; Among them, H UAV represents the UAV flight altitude, and β0 represents the channel power gain at a reference distance of 1 m; r represents the uplink transmission rate from the UE to the UAV; Among them, p represents the transmission power from the UE to the UAV in watts; B represents the channel bandwidth in Hz, and the bandwidth between the UAV and the UEs within its coverage is equal; N0 represents the noise power spectral density in W / Hz; Local delay T L Expressed as: where β is the offloading ratio of the UE to offload tasks to the UAV, D represents the task data size, F L is the number of CPU cycles required for the UE to compute one bit of the task, and f L is the computing power of the UE's CPU; When calculating the processing delay and transmission delay, consider the offloading strategy a; a) When a = 0, the processing delay T of the current UAV UAV is expressed as: where β1 represents the offloading ratio offloaded to the current UAV, F UAV represents the number of CPU cycles required for the current UAV to compute one bit, f UAV represents the computing power of the current UAV; The transmission delay T between the UE and the current UAV TX Expressed as: b) When a = 1, the time delay T generated by the next UAV to process this task UAV' is expressed as: Among them, β2 represents the offloading ratio of the UE to the next UAV, and F UAV' represents the number of CPU cycles required for the next UAV to calculate one bit, and f UAV' represents the computing power of the next UAV; The transmission delay T between the UE and the next drone TX' is as follows:

4. The multi-UAV collaborative mobile edge computing network computing offloading method under a fixed trajectory according to claim 3, characterized in that For step S3, specifically as follows: T MAX Indicates the upper limit of the communication time between the UE and the UAV: Among them, r represents the coverage radius of the UAV, R is the flight radius of the UAV, and v represents the flight speed of the UAV; T W Indicates the longest waiting delay of the UE: Among them, U is the number of UAVs, R is the flight radius of the UAV, and r is the communication radius of the UAV.

5. The multi-UAV collaborative mobile edge computing network computing offloading method under a fixed trajectory according to claim 4, characterized in that, For step S4, specifically as follows: Step S41, randomly generate vectors within the feasible region as the initial population; set parameters including the maximum number of iterations G m , population size Np, feasible region Z, mutation rate F0, crossover probability CR, initial temperature T0, cooling rate ρ, termination temperature T end and delay function f(x); within the feasible region, randomly generate a vector set X containing 3N dimensions as the initial population; Step S42. Calculate the mutation rate F corresponding to the current iteration number; F = F0·2 λ Among them, G is the current iteration number; Step S43, calculate the mutant y corresponding to the i-th individual according to the following formula for the mutation rate i :[[]]END]] y i = x i - F·(x j - x k ) where x j and x k are two different individuals randomly selected from the population; Step S44, crossover and mutation y i and the parent x i ; Among them, rand is a random number uniformly distributed in the interval [0,1]; CR represents the crossover probability; Step S45. Select the optimal individual according to the following formula: Step S46. Generate a new candidate solution within the neighboring solution space m of the current solution: Among them, m represents the radius of the neighboring solution space; Step S47. According to the current temperature, determine whether to accept the new solution; Δ i Indicates the difference between the current solution and the new solution: Δ i = f(h i ) - f(k i ) γ represents the acceptance probability: Step S48. Cool down proportionally and update the current temperature: T = ρ·T Among them, ρ is the cooling coefficient; Step S49. Determine whether the current temperature T has reached the minimum value. If not, return to step S46; otherwise, proceed to step S410; Step S410. Increment the iteration number by 1; Step S411. Determine whether the iteration number has reached the maximum value. If not, return to step S42; otherwise, proceed to step S412; Step S412. Output the optimal strategy.

6. A multi-UAV collaborative mobile edge computing network computing offloading system under a fixed trajectory, which is used to execute the method according to any one of claims 1-5, and is characterized in that, It includes the following modules: System model initialization module: Initialize the user location information, channel power gain, and bandwidth; Offloading delay calculation module: Combine the local and UAV-side computing resources to calculate the local delay, transmission delay, and processing delay; Waiting delay calculation module: Calculate the UE waiting delay according to the UAV flight trajectory; Optimal strategy calculation module: Use the simulated annealing differential evolution fusion algorithm to calculate and output the optimal offloading strategy.

Images