CN114500533B - Unmanned aerial vehicle auxiliary mobile edge computing system optimization method based on user cooperation - Google Patents

Abstract

The invention discloses an unmanned aerial vehicle auxiliary mobile edge computing system optimizing method based on user cooperation, which comprises an unmanned aerial vehicle computing node and a plurality of ground users, wherein the ground users comprise near users and far users, and the ground users have a certain amount of computing tasks to be processed. Due to the limitation of unmanned aerial vehicle battery energy, ground user cooperation offloads computing tasks to the unmanned aerial vehicle in order to reduce unmanned aerial vehicle flight energy consumption. The to-be-processed computing task of the far user is relayed to the near user, and the near user then unloads the to-be-processed computing task of the far user to the unmanned aerial vehicle for computing. The invention aims at minimizing the weighted total energy consumption of the unmanned aerial vehicle and the ground user on the basis of meeting the service quality of the user, and realizes the joint optimization of the flight track of the unmanned aerial vehicle, the data quantity of the task to be processed, which is relayed and unloaded by the ground user, and the CPU calculation frequency of the ground user and the unmanned aerial vehicle based on a two-step iterative algorithm.

Description

Unmanned aerial vehicle auxiliary mobile edge computing system optimization method based on user cooperation

Technical Field

The invention relates to an unmanned aerial vehicle auxiliary mobile edge computing system optimization method based on user cooperation, and belongs to the technical field of wireless communication and Internet of things.

Background

With the rapid development of the internet of things and 5G communication technologies, applications such as smart home, face recognition and augmented reality have appeared in our lives. This has also prompted the mobile communication devices to move toward low latency, low power consumption, high reliability, and high density connections. However, existing communication systems are challenging in performing computationally intensive and delay sensitive tasks due to limitations in the battery and computing power of their mobile devices. To address this problem, mobile edge computing (Mobile Edge Computing System, MEC) is an effective technique, meaning that servers deployed at the edge of the network can provide services to users. Since edge servers are located near the user, they can provide computing services to the user at lower energy costs and low latency. However, the deployment of edge servers at fixed locations suffers from inflexible and difficult service coverage adjustment, subject to complex terrain and uncertainty requirements.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provide an unmanned aerial vehicle auxiliary mobile edge computing system optimization method based on user cooperation.

In order to achieve the above object, the present invention provides an unmanned aerial vehicle assisted mobile edge computing system optimization method based on user cooperation, comprising:

among the ground users, the far user m _i Pairing with near user i, m _i ∈[1,M],i∈R，R＝{1,…,I}，M＝I；

Based on unmanned aerial vehicle auxiliary mobile edge computing system model, optimal number of remote users m _i Relaying the calculation task to be processed to a near user i;

based on unmanned aerial vehicle auxiliary mobile edge computing system model, optimal number of remote users m _i And unloading the to-be-processed calculation tasks relayed to the near user i and the optimal number of to-be-processed calculation tasks of the near user i to the unmanned aerial vehicle for calculation.

Preferentially, 1) constructing an initial model of an unmanned aerial vehicle auxiliary mobile edge computing system of user cooperation;

2) Setting iteration number variable ζ=1, and ending the iteration

Giving a flight path U of the unmanned aerial vehicle;

3) Under the condition of a given unmanned aerial vehicle flight trajectory, optimizing an initial model of an unmanned aerial vehicle auxiliary mobile edge computing system until a sub-problem meets a convergence condition, stopping iteration, and obtaining an optimal ground user, a CPU computing frequency F of the unmanned aerial vehicle and the total number L of to-be-processed computing tasks relayed and unloaded by the ground user;

4) Inputting and optimizing an initial model of an unmanned aerial vehicle auxiliary moving edge computing system by using the optimal ground user, the CPU computing frequency F of the unmanned aerial vehicle and the total number L of the to-be-processed computing tasks relayed and unloaded by the ground user until the set convergence condition is met, and outputting an optimal unmanned aerial vehicle flight track U;

5) Inputting the optimal F, L and U obtained in the step 3) and the step 4) into a minimum objective function of an initial model of an unmanned aerial vehicle auxiliary moving edge computing system to obtain a value E _ζ ；

6) Updating the iteration number variable ζ=ζ+1;

7) Optimal unmanned aerial vehicle flight railSubstituting the trace U into the step 3), and repeating the steps 3) to 6) until the initial model of the unmanned aerial vehicle auxiliary mobile edge computing system meets the convergence condition

And (5) ending the iteration to obtain the unmanned aerial vehicle auxiliary mobile edge computing system model.

Preferably, during the task completion time T, the drone flies at a fixed altitude H, the far user m _i And the height of the near user i is 0 m, the position is fixed, and the coordinate of the near user i is s _i ＝(x _i ,y _i ) Far user m _i Is the coordinates of (a)

The near user and the unmanned aerial vehicle communicate in a time division multiplexing mode;

dividing task completion time T into N slots, N e {1, …, N }, width δ=t/N of each slot;

within the task completion time T, the unmanned aerial vehicle starts from the starting point u _I Fly to destination u _F ；

The speed of the unmanned plane in the nth time slot is as follows:

v[n]＝||u[n]-u[n-1]||/δ；

the initial model of the unmanned aerial vehicle auxiliary mobile edge computing system is as follows:

s.t.C1:

C2:

C3:

C4:

C5

C6:

C7:

C8:u[0]＝u _I ,u[N]＝u _F

C9:

C10:

C11:

C12:

in the minimum objective function, u= { U [ n ]]Is the unmanned plane from the initial point u _I Fly to the end point u _F Is a flying trace of (a);

the total number of tasks to be processed is calculated for the ground user to relay and offload,

For far user m _i The number of calculation tasks to be processed relayed to the near user i in the nth time slot,/-, is->

For near user i to far user m in nth time slot _i The number of pending calculation tasks offloaded to the unmanned plane u +.>

Unloading the number of the self to-be-processed calculation tasks for the near user i in the nth time slot;

calculating frequency for ground users and CPU of unmanned aerial vehicle, f _i [n]Calculating the frequency for the CPU of the near user, +.>

For far user m _i CPU calculation frequency f _i,u [n]CPU calculation frequency for processing the to-be-processed calculation task of the near user i for the unmanned aerial vehicle,/for the unmanned aerial vehicle>

Handling near user i offloaded from far user m for drone _i CPU computing frequency of the computing task to be processed;

θ ₁ weighting coefficient, θ, for total energy consumption of ground users ₂ A weighting coefficient for the total energy consumption of the unmanned aerial vehicle;

wherein, c ₁ And c ₂ Two preset parameters related to the weight, wing area, wingspan efficiency and air density of the unmanned aerial vehicle; e (E) _i [n]Calculating the number L of the completed calculation tasks to be processed in the nth time slot for the near user i _i [n]The required energy consumption:

E _i [n]＝δκ _i f _i ³ [n]，

wherein, kappa _i An effective capacitance coefficient for near user i;

for far user m _i Calculating the number of completed calculation tasks to be processed in the nth time slot +.>

The required energy consumption:

in the method, in the process of the invention,

Representing far user m _i Effective capacitance coefficient of (a);

to accomplish->

The required energy consumption:

in the far user m _i When the pending computing task of (a) is relayed to near user iWidth of space, near user i will far user m _i The time width of unloading the calculation task to be processed to the unmanned plane u and the time width of unloading the calculation task to be processed to the unmanned plane u by the near user i are delta ₀ ＝δ/3I；

For far user m _i Transmit power of>

For the received noise power of near user i, +.>

For near user i and far user m _i The channel gain of the link between the two, B is the communication bandwidth;

to accomplish->

The required energy consumption:

in the method, in the process of the invention,

will come from far user m for near user i _i Is offloaded to the transmit power of unmanned plane u, +.>

For receiving noise power of the unmanned aerial vehicle, a free space path loss model is adopted for a channel between the unmanned aerial vehicle and a near user, and the unmanned aerial vehicle is in a +.>

For the channel gain between the near user i and the unmanned plane u, d _i [n]Gamma is the distance between the near user i and the unmanned plane u ₀ For reference distance d ₀ Channel gain at=1m;

to accomplish->

The required energy consumption:

in the method, in the process of the invention,

unloading the self calculation task to be processed to the transmitting power of the unmanned plane u for the near user i;

E _i,u [n]the number L of completed tasks to be processed from the near user i is calculated for the unmanned aerial vehicle in the nth time slot _i,u [n]The required energy consumption:

wherein, kappa _u Is the effective capacitance coefficient of the unmanned aerial vehicle;

E _mi,u [n]offloaded from far user m for near user i completed for drone computation in nth time slot _i Number of computing tasks to be processed

The required energy consumption:

in the method, in the process of the invention,

handling far user m for unmanned aerial vehicle _i The CPU calculation frequency of (2);

equation C1 ensures that near user i completely offloads from far user m during each time slot _i N, N ₁ = {1,..n-1 }; equation C2 is an information causal randomness limit, the drone can only calculate the pending calculation task that has been received in the last time slot, and the processing delay is one time slot,

for the near user i to offload its own number of computational tasks to be processed in the kth time slot, k= {1, …, n-1}, L _i,u [k]For the unmanned u to calculate the number of completed self-pending calculation tasks offloaded from the near user i in the kth time slot, N ₂ ＝{2,...,N}；

Equation C3 is the information causal random limit,

for near user i to far user m in kth time slot _i The number of pending calculation tasks offloaded to the unmanned plane u +.>

Offloaded from far user m for near user i completed by drone computation in kth slot _i The number of computing tasks to be processed;

Equation C4 ensures that the calculation tasks to be processed of the near user i self unloaded in the nth time slot can be all calculated by the unmanned plane, L _i,u [n]Calculating the number of completed tasks to be processed from the near user i per se for the unmanned plane u in the nth time slot;

equation C5 ensures that near user i is offloaded to drone u from far user m in the nth time slot _i Can be calculated and completed by the unmanned aerial vehicle,

indicating that the unmanned plane u is counted in the nth time slotOffloaded from far user m for near user i after completion of calculation _i The number of computing tasks to be processed;

equation C6 ensures that the unmanned aerial vehicle calculates the distance user m from the distance user who finishes the unloading of the near user i within the task completion time T _i Is used for processing the calculation task to be processed,

for far user m _i The number of computational tasks to be processed for local computation at the nth time slot,

for far user m _i Is used for calculating task demands;

formula C7 ensures that the unmanned plane completes the task to be processed and calculated unloaded by the near user i in the task completion time T, L _i [n]The number of calculation tasks to be processed for local calculation of near user I in the nth time slot, I _i Representing the computing task requirements of a near user i;

equation C8 limits the unmanned aerial vehicle from the initial point u within the task completion time T _I Fly to the end point u _F ，u[0]＝u _I ，u[N]＝u _F ；

Equation C9 limits the maximum speed of the unmanned aerial vehicle, u [ n ]]＝u(δn)＝(x[n],y[n]) For the projected coordinates of the unmanned u on the horizontal plane in the nth time slot, u [ n-1 ]]For the projected coordinates of the drone u on the horizontal plane in the n-1 th time slot, V _max Is the maximum speed of the unmanned aerial vehicle;

the formula C10 ensures that the ground user does not relay or unload the calculation task to be processed in the last time slot, and ensures that the unmanned aerial vehicle does not process the calculation task to be processed in the first time slot;

equation C11 limits the non-negative or maximum value, f _i,max For the maximum calculated frequency of far user i, f _mi,max For far user m _i Maximum calculated frequency of f _u,max The maximum calculation frequency of the unmanned plane is calculated;

c12 is the maximum limit for near user transmit power, the maximum limit for far user transmit power;

P _max for the maximum of the transmit power of all terrestrial users,

for near user i to far user m in nth time slot _i Is offloaded to the transmit power of unmanned plane u, +.>

For the next user i to unload its own transmit power of the calculation task to be processed in the nth time slot,/>

For far user m _i And relaying the calculation task to be processed to the transmitting power of the near user i in the nth time slot.

Preferably, step 3) comprises:

3.1 Under the condition of the given unmanned aerial vehicle flight track, the initial model of the unmanned aerial vehicle auxiliary movement edge computing system to be optimized is as follows:

s.t.C1-C7,C10-C12,

3.2 Given the dual variable β= { β corresponding to the inequality constraint C1-C3 _i,n }、λ＝{λ _i,n Sum η= { η _i,n A value of };

3.3 Determining a binary variable ζ= { ζ corresponding to the equation constraint C4-C7 based on a binary search method according to values of the binary variables β, λ, and η _i }、ω＝{ω _i }、μ＝{μ _i Sum ρ= { ρ _i A value of };

3.4 Using Karush-Kuhn-Tucker conditions to solve the neutron problem of the initial model of the unmanned aerial vehicle auxiliary mobile edge computing system, wherein the solution is as follows:

wherein C is _i The computational resources required to compute a one-bit input bit for near user i,

for far user m _i Computing resources required to compute a one-bit input bit;

setting up

Far user m _i The maximum number of computational tasks to be processed relayed to near user i in the nth time slot, near user i unloading from far user m in the nth time slot _i The number of the maximum to-be-processed computing tasks of the near user i unloading itself in the nth time slot is respectively as follows:

[x] ⁺ ＝max{x,0}，

3.5 Iterative updating of the dual variables β, λ and η) based on the algorithm of the secondary gradient:

where j is the iteration index variable of the secondary gradient algorithm iteration,

And->

Represents the jth iteration step, respectively, of obtaining the dual variables β, λ and η, +.>

f _i,u,j-1 [n]And->

Respectively is

f _i,u [n]And->

Variable values of the corresponding j-1 th iteration;

3.6 Repeating the steps 3.2) to 3.5) until the objective function value of the sub-problem meets the set iteration termination precision, and obtaining the optimal F and L after the iteration is terminated.

Preferably, step 4) comprises:

4.1 Inputting the optimal ground user and the CPU calculation frequency F of the unmanned aerial vehicle and the total number L of the to-be-processed calculation tasks relayed and unloaded by the ground user into an initial model of the unmanned aerial vehicle auxiliary mobile edge calculation system:

s.t.C8,C9andC12；

4.2 The flight energy consumption of the unmanned aerial vehicle is as follows:

due to u [ n ]]Is non-convex with respect to the equation, so introduces a relaxation variable

To obtain E ^fly [n]Upper bound of (2):

the added constraint translates into the following form:

wherein u is ^(r) [n]Is a local point of the flight path of the unmanned aerial vehicle;

in step 4.1), updating an initial model of the unmanned aerial vehicle auxiliary mobile edge computing system as follows:

s.t.C8,C9andC12，

C13:

4.3 Using an initial unmanned aerial vehicle flight trajectory u [ n ]]As a local point, updating the iterative unmanned aerial vehicle track u [ n ] based on the interior point method]Until the convergence accuracy of the interior point method meets the set convergence accuracy

And obtaining the optimal unmanned aerial vehicle flight path U.

Preferably, step 3.3), comprises:

3.3.1 The solution of the neutron problem is constrained by equations C4-C7 and step 3.4), and is obtained

And->

Different expressions relating to values;

and (3) with

The different expressions relating to the values are:

and (3) with

The different expressions relating to the values are:

and (3) with

The different expressions relating to the values are:

3.3.2 Given lambda, combine

Scope of->

The range of the obtained dual variable ρ is:

when ρ is _i Given, the range of binary search for the dual variable ζ is:

wherein ρ is _i,min As a dual variable ρ _i Minimum value ρ of _i,max As a dual variable ρ _i Is set at the maximum value of (c),

n=1 +.>

A value;

when the binary search is terminated, the result is at a given ρ _i Corresponding xi under the condition of (2) _i Values such that

Continuously binary search ρ _i And iterate the corresponding ζ _i Values such that

When the binary search method converges, a dual variable rho is obtained _i And xi _i Is the optimum value of (2);

3.3.3 Given beta), binding

Scope of->

The range of the binary variable mu binary search is:

when the binary search converges, a dual variable mu is obtained _i Is the optimum value of (3):

3.3.4 Given η and β), combined

Scope of->

The range of obtaining the binary variable omega binary search is as follows:

when the binary search method converges, the dual variable omega is obtained _i Is the optimum value of (3):

the invention has the beneficial effects that:

1) The invention relates to an unmanned aerial vehicle auxiliary mobile edge computing system optimizing method based on unmanned aerial vehicle and ground user weighting and energy minimization user cooperation. Due to the limitation of unmanned aerial vehicle battery energy, ground user cooperation offloads computing tasks to the unmanned aerial vehicle in order to reduce unmanned aerial vehicle flight energy consumption. The to-be-processed computing task of the far user is relayed to the near user, and the near user then unloads the to-be-processed computing task of the far user to the unmanned aerial vehicle for computing. The invention aims at minimizing the weighted total energy consumption of the unmanned aerial vehicle and the ground user on the basis of meeting the service quality of the user, and realizes the joint optimization of the flight track of the unmanned aerial vehicle, the data quantity of the task to be processed, which is relayed and unloaded by the ground user, and the CPU calculation frequency of the ground user and the unmanned aerial vehicle based on a two-step iterative algorithm.

2) The invention provides a novel unmanned aerial vehicle auxiliary mobile edge computing system frame based on user cooperation, and the unmanned aerial vehicle auxiliary mobile edge computing system frame provided in the past does not consider the situation of relay cooperation of ground users. The invention considers the time division multiplexing communication mode, aims at optimizing the weighted total energy consumption of the unmanned aerial vehicle and the user on the basis of meeting the service quality of the user, and performs joint optimization on the flight track of the unmanned aerial vehicle, the data quantity relayed or unloaded by the ground user and the CPU calculation frequency of the ground user and the unmanned aerial vehicle.

3) Since the proposed optimization problem is non-convex and difficult to solve, the present invention solves this problem by iteratively solving two sub-problems. The data quantity relayed or unloaded by the ground user and the CPU calculation frequency of the ground user and the unmanned aerial vehicle are obtained in a closed form through a Lagrange dual method, lagrange multipliers related to inequality constraint can be obtained through a secondary gradient method, and Lagrange multipliers related to equality constraint can be obtained through binary search. The interior point method can effectively solve the sub-problems related to unmanned aerial vehicle track optimization.

4) The convergence of the algorithm is guaranteed, requiring less complexity. Simulation results show that the UAV's optimized trajectories in different scenarios and with the proposed algorithm significantly improves performance compared to existing solutions (e.g., solutions that do not take into account ground user cooperation, solutions with preset UAV trajectories, etc.), and furthermore, the algorithm can provide more stable performance to accommodate changes in the operating environment and its advantages will be more pronounced when processing computationally intensive and delay critical tasks.

Drawings

FIG. 1 is a system model diagram of the present invention;

FIG. 2 is a basic flow chart of the present invention;

FIG. 3 is a line graph of the flight path of the unmanned aerial vehicle at different mission completion times according to the present invention;

FIG. 4 is a graph of the total energy consumption of a ground user and a drone as a function of the amount of tasks under conditions of the present invention without user cooperation and with remote user consideration of drone cooperation, drone straight flight and without user cooperation and without user consideration of drone cooperation;

FIG. 5 is a graph of the total energy consumption of a ground user and a drone over time under conditions of the present invention without user cooperation and with remote user consideration of drone cooperation, drone straight flight and without user cooperation and without user consideration of drone cooperation;

FIG. 6 is a schematic diagram showing the convergence of the present invention under different ground user task volume requirements.

Detailed Description

The following examples are only for more clearly illustrating the technical aspects of the present invention, and are not intended to limit the scope of the present invention.

An Unmanned Aerial Vehicle (UAV) enabled MEC has flexible and fast deployment capabilities that are considered suitable for servicing specific areas and special computing needs. The edge computing server based on the unmanned aerial vehicle has high-speed mobility, can freely approach a mobile user and provide computing services, and can remarkably improve network performance. Furthermore, air-to-ground communications in a drone-supported MEC network may provide higher link capacity than ground communications due to line-of-sight link transmission between the drone and the user. Therefore, unmanned assisted MEC networks are a hotspot in current MEC research.

When the position distribution of the ground users is uneven and the distribution positions are wider, the unmanned aerial vehicle needs to fly in a large range to better execute the unloading task of each ground user, but the traditional unloading scheme cannot well meet the energy consumption requirement of the unmanned aerial vehicle due to the limitation of the battery energy of the unmanned aerial vehicle.

The invention provides an unmanned aerial vehicle auxiliary mobile edge computing system optimization method based on user cooperation, which comprises the following steps:

the unmanned aerial vehicle comprises an unmanned aerial vehicle computing node u and ground users, wherein the ground users comprise far users and near users, the far users and the near users have a certain amount of to-be-processed computing tasks, the far users and the near users adopt a mode of unloading part of to-be-processed computing tasks, namely, part of to-be-processed computing tasks of the far users and part of to-be-processed computing tasks of the near users are locally computed, and the rest of to-be-processed computing tasks of the far users and the near users are unloaded to the unmanned aerial vehicle for computation.

Specifically, among the ground users, the far user m _i Pairing with unique near user i, m _i ∈[1,M],i∈R，R＝{1,…,I}，M＝I；

Based on unmanned aerial vehicle auxiliary mobile edge computing system model, optimal number of remote users m _i Relaying the calculation task to be processed to a near user i;

based on unmanned aerial vehicle auxiliary mobile edge computing system model, optimal number of remote users m _i And unloading the to-be-processed calculation tasks relayed to the near user i and the optimal number of to-be-processed calculation tasks of the near user i to the unmanned aerial vehicle for calculation.

Further, 1) constructing an initial model of an unmanned aerial vehicle auxiliary mobile edge computing system with user cooperation;

2) Setting iteration number variable ζ=1, and ending the iteration

Giving a flight path U of the unmanned aerial vehicle;

3) Under the condition of a given unmanned aerial vehicle flight trajectory, optimizing an initial model of an unmanned aerial vehicle auxiliary mobile edge computing system until a sub-problem meets a convergence condition, stopping iteration, and obtaining an optimal ground user, a CPU computing frequency F of the unmanned aerial vehicle and the total number L of to-be-processed computing tasks relayed and unloaded by the ground user;

4) Inputting and optimizing an initial model of an unmanned aerial vehicle auxiliary moving edge computing system by using the optimal ground user, the CPU computing frequency F of the unmanned aerial vehicle and the total number L of the to-be-processed computing tasks relayed and unloaded by the ground user until the set convergence condition is met, and outputting an optimal unmanned aerial vehicle flight track U;

5) Inputting the optimal F, L and U obtained in the step 3) and the step 4) into an initial model of an unmanned aerial vehicle auxiliary mobile edge computing systemIs the minimum objective function of (1) to obtain the value E _ζ ；

6) Updating the iteration number variable ζ=ζ+1;

7) Substituting the optimal unmanned aerial vehicle flight track U into the step 3), and repeating the steps 3) to 6) until the initial model of the unmanned aerial vehicle auxiliary movement edge computing system meets the convergence condition

And (5) ending the iteration to obtain the unmanned aerial vehicle auxiliary mobile edge computing system model.

Further, in this embodiment, during the task completion time T, the unmanned aerial vehicle flies at a fixed height H, and the remote user m _i And the height of the near user i is 0 m, the position is fixed, and the coordinate of the near user i is s _i ＝(x _i ,y _i ) Far user m _i Is the coordinates of (a)

The near user and the unmanned aerial vehicle communicate in a time division multiplexing mode; assuming that the system communication adopts a time division multiplexing technology, a ground user can simultaneously carry out local calculation and unloading of a calculation task to be processed without interference, and meanwhile, the time for returning the calculation result of the ground user by the unmanned aerial vehicle is ignored;

dividing task completion time T into N slots, N e {1, …, N }, width δ=t/N of each slot;

each time slot delta is further divided into I time segments, each time segment comprising a far user m _i Time when the pending computing task of (a) is relayed to near user i, near user i will far user m _i The time for unloading the to-be-processed computing task of the near user i to the unmanned aerial vehicle u and the time for unloading the to-be-processed computing task of the near user i to the unmanned aerial vehicle u, the far user m _i The time width of relaying the task to be processed to the near user i, the near user i relays the far user m _i The time width of unloading the to-be-processed calculation task of the near user i to the unmanned aerial vehicle u and the time width of unloading the to-be-processed calculation task of the near user i to the unmanned aerial vehicle u are delta ₀ ＝δ/3I；

Each remote user m _i Is expressed as a pending computational taskPositive tuple

The computational task to be processed near user I is represented as a positive tuple [ I _i ,C _i ]；

Within the task completion time T, the unmanned aerial vehicle starts from the starting point u _I Fly to destination u _F ；

The speed of the unmanned plane in the nth time slot is as follows:

v[n]＝||u[n]-u[n-1]||/δ；

the initial model of the unmanned aerial vehicle auxiliary mobile edge computing system is as follows:

s.t.C1:

C2:

C3:

C4

C5:

/>

C6:

C7:

C8:u[0]＝u _I ,u[N]＝u _F

C9:

C10:

C11:

C12:

in the minimum objective function, u= { U [ n ]]Is the unmanned plane from the initial point u _I Fly to the end point u _F Is a flying trace of (a);

the total number of tasks to be processed is calculated for the ground user to relay and offload,

for far user m _i The number of calculation tasks to be processed relayed to the near user i in the nth time slot,/-, is->

For near user i to far user m in nth time slot _i The number of pending calculation tasks offloaded to the unmanned plane u +.>

Unloading the number of the self to-be-processed calculation tasks for the near user i in the nth time slot;

calculating frequency for ground users and CPU of unmanned aerial vehicle, f _i [n]Calculating the frequency for the CPU of the near user, +. >

For far user m _i CPU calculation frequency f _i,u [n]CPU calculation frequency for processing the to-be-processed calculation task of the near user i for the unmanned aerial vehicle,/for the unmanned aerial vehicle>

Handling near user i offloaded from far user m for drone _i CPU computing frequency of the computing task to be processed;

θ ₁ weighting coefficient, θ, for total energy consumption of ground users ₂ A weighting coefficient for the total energy consumption of the unmanned aerial vehicle;

wherein, c ₁ And c ₂ Two preset parameters related to the weight, wing area, wingspan efficiency and air density of the unmanned aerial vehicle; e (E) _i [n]Calculating the number L of the completed calculation tasks to be processed in the nth time slot for the near user i _i [n]The required energy consumption:

E _i [n]＝δκ _i f _i ³ [n]，

wherein, kappa _i An effective capacitance coefficient for near user i;

for far user m _i Calculating the number of completed calculation tasks to be processed in the nth time slot +.>

The required energy consumption:

in the method, in the process of the invention,

representing far user m _i Effective capacitance coefficient of (a); />

To accomplish->

The required energy consumption:

in the far user m _i The time width of relaying the task to be processed to the near user i, the near user i relays the far user m _i The time width of unloading the calculation task to be processed to the unmanned plane u and the time width of unloading the calculation task to be processed to the unmanned plane u by the near user i are delta ₀ ＝δ/3I；

For far user m _i Transmit power of >

The received noise power for near user i, h _mi [n]For near user i and far user m _i Channel gain for inter-linkB is the communication bandwidth;

to accomplish->

The required energy consumption:

in the method, in the process of the invention,

will come from far user m for near user i _i Is offloaded to the transmit power of unmanned plane u, +.>

For receiving noise power of the unmanned aerial vehicle, a free space path loss model is adopted for a channel between the unmanned aerial vehicle and a near user, and the unmanned aerial vehicle is in a +.>

For the channel gain between the near user i and the unmanned plane u, d _i [n]Gamma is the distance between the near user i and the unmanned plane u ₀ For reference distance d ₀ Channel gain at=1m;

to accomplish->

The required energy consumption:

in the method, in the process of the invention,

unloading the self calculation task to be processed to the transmitting power of the unmanned plane u for the near user i;

E _i,u [n]for the unmanned aerial vehicle to calculate the number of completed pending calculation tasks from the near user i itself in the nth time slot

L _i,u [n]The required energy consumption:

wherein, kappa _u Is the effective capacitance coefficient of the unmanned aerial vehicle;

offloaded from far user m for near user i completed for drone computation in nth time slot _i The number of computing tasks to be processed +.>

The required energy consumption:

in the method, in the process of the invention,

handling far user m for unmanned aerial vehicle _i The CPU calculation frequency of (2); />

Equation C1 ensures that near user i completely offloads from far user m during each time slot _i N, N ₁ ＝{1,...,N-1}；

Equation C2 is an information causal randomness limit, the drone can only calculate the pending calculation task that has been received in the last time slot, and the processing delay is one time slot,

for the near user i to offload its own number of computational tasks to be processed in the kth time slot, k= {1, …, n-1}, L _i,u [k]For the unmanned u to calculate the number of completed self-pending calculation tasks offloaded from the near user i in the kth time slot, N ₂ ＝{2,...,N}；

Equation C3 is the information causal random limit,

for near user i to far user m in kth time slot _i The number of pending calculation tasks offloaded to the unmanned plane u +.>

Offloaded from far user m for near user i completed by drone computation in kth slot _i The number of computing tasks to be processed;

equation C4 ensures that the calculation tasks to be processed of the near user i self unloaded in the nth time slot can be all calculated by the unmanned plane, L _i,u [n]Calculating the number of completed tasks to be processed from the near user i per se for the unmanned plane u in the nth time slot;

equation C5 ensures that near user i is offloaded to drone u from far user m in the nth time slot _i Can be calculated and completed by the unmanned aerial vehicle,

Near user i offloaded from far user m indicating completion of drone u computation in nth time slot _i The number of computing tasks to be processed;

equation C6 ensures that the unmanned aerial vehicle calculates the distance user m from the distance user who finishes the unloading of the near user i within the task completion time T _i Is used for processing the calculation task to be processed,

for far user m _i The number of computational tasks to be processed for local computation at the nth time slot,

for far user m _i Is used for calculating task demands;

formula C7 ensures that the unmanned plane completes the task to be processed and calculated unloaded by the near user i in the task completion time T, L _i [n]The number of calculation tasks to be processed for local calculation of near user I in the nth time slot, I _i Representing the computing task requirements of a near user i;

equation C8 limits the unmanned aerial vehicle from the initial point u within the task completion time T _I Fly to the end point u _F ，u[0]＝u _I ，u[N]＝u _F ；

Equation C9 limits the maximum speed of the unmanned aerial vehicle, u [ n ]]＝u(δn)＝(x[n],y[n]) For the projected coordinates of the unmanned u on the horizontal plane in the nth time slot, u [ n-1 ]]For the projected coordinates of the drone u on the horizontal plane in the n-1 th time slot, V _max Is the maximum speed of the unmanned aerial vehicle;

the formula C10 ensures that the ground user does not relay or unload the calculation task to be processed in the last time slot, and ensures that the unmanned aerial vehicle does not process the calculation task to be processed in the first time slot;

Equation C11 limits the non-negative or maximum value, f _i,max For the maximum calculated frequency of the far user i,

for far user m _i Maximum calculated frequency of f _u,max The maximum calculation frequency of the unmanned plane is calculated;

c12 is the maximum limit for near user transmit power, the maximum limit for far user transmit power;

P _max for the maximum of the transmit power of all terrestrial users,

for near user i to far user m in nth time slot _i Is offloaded to the transmit power of unmanned plane u, +.>

For the next user i to unload its own transmit power of the calculation task to be processed in the nth time slot,/>

For far user m _i And relaying the calculation task to be processed to the transmitting power of the near user i in the nth time slot.

Further, step 3) in this embodiment includes:

3.1 Under the condition of the given unmanned aerial vehicle flight track, the initial model of the unmanned aerial vehicle auxiliary movement edge computing system to be optimized is as follows:

s.t.C1-C7,C10-C12,

3.2 Given the dual variable β= { β corresponding to the inequality constraint C1-C3 _i,n }、λ＝{λ _i,n Sum η= { η _i,n A value of };

3.3 Determining a binary variable ζ= { ζ corresponding to the equation constraint C4-C7 based on a binary search method according to values of the binary variables β, λ, and η _i }、ω＝{ω _i }、μ＝{μ _i Sum ρ= { ρ _i A value of };

3.4 Using Karush-Kuhn-Tucker conditions to solve the neutron problem of the initial model of the unmanned aerial vehicle auxiliary mobile edge computing system, wherein the solution is as follows:

/>

Wherein C is _i The computational resources required to compute a one-bit input bit for near user i,

for far user m _i Computing resources required to compute a one-bit input bit;

setting up

Far user m _i The maximum number of computational tasks to be processed relayed to near user i in the nth time slot, near user i unloading from far user m in the nth time slot _i The number of the maximum to-be-processed computing tasks of the near user i unloading itself in the nth time slot is respectively as follows:

[x] ⁺ ＝max{x,0}，

3.5 Iterative updating of the dual variables β, λ and η) based on the algorithm of the secondary gradient:

where j is the iteration index variable of the secondary gradient algorithm iteration,

and->

Represents the jth iteration step, respectively, of obtaining the dual variables β, λ and η, +.>

f _i,u,j-1 [n]And->

Respectively is

f _i,u [n]And->

Variable values of the corresponding j-1 th iteration;

3.6 Repeating the steps 3.2) to 3.5) until the objective function value of the sub-problem meets the set iteration termination precision, and obtaining the optimal F and L after the iteration is terminated.

Further, step 4) in this embodiment includes:

4.1 Inputting the optimal ground user and the CPU calculation frequency F of the unmanned aerial vehicle and the total number L of the to-be-processed calculation tasks relayed and unloaded by the ground user into an initial model of the unmanned aerial vehicle auxiliary mobile edge calculation system:

s.t.C8,C9andC12；

4.2 The flight energy consumption of the unmanned aerial vehicle is as follows:

due to u [ n ]]Is non-convex with respect to the equation, so introduces a relaxation variable

To obtain E ^fly [n]Upper bound of (2):

the added constraint translates into the following form:

wherein u is ^(r) [n]Is a local point of the flight path of the unmanned aerial vehicle;

in step 4.1), updating an initial model of the unmanned aerial vehicle auxiliary mobile edge computing system as follows:

s.t.C8,C9andC12，

C13:

4.3 Using an initial unmanned aerial vehicle flight trajectory u [ n ]]As a local point, updating the iterative unmanned aerial vehicle track u [ n ] based on the interior point method]Until the convergence accuracy of the interior point method meets the set convergence accuracy

And obtaining the optimal unmanned aerial vehicle flight path U.

Further, step 3.3) in the present embodiment includes:

3.3.1 The solution of the neutron problem is constrained by equations C4-C7 and step 3.4), and is obtained

And->

Different expressions relating to values;

and (3) with

The different expressions relating to the values are:

/>

and (3) with

The different expressions relating to the values are:

and (3) with

The different expressions relating to the values are:

3.3.2 Given lambda, combine

Scope of->

The range of the obtained dual variable ρ is:

when ρ is _i Given, the range of binary search for the dual variable ζ is:

wherein ρ is _i,min As a dual variable ρ _i Minimum value ρ of _i,max As a dual variable ρ _i Is set at the maximum value of (c),

N=1 +.>

A value;

when the binary search is terminated, the result is at a given ρ _i Corresponding xi under the condition of (2) _i Values such that

Continuously binary search ρ _i And iterate the corresponding ζ _i Values such that

When the binary search method converges, a dual variable rho is obtained _i And xi _i Is the optimum value of (2);

3.3.3 Given beta), binding

Scope of->

The range of the binary variable mu binary search is:

when the binary search converges, a dual variable mu is obtained _i Is the optimum value of (3):

3.3.4 Given η and β), combined

Scope of->

The range of obtaining the binary variable omega binary search is as follows:

when the binary search method converges, the dual variable omega is obtained _i Is the optimum value of (3):

the far user relays the data to be offloaded to the near user, and the near user offloads the data to be processed to the unmanned plane. Because the near user is closer to the unmanned aerial vehicle, the communication stability of the ground user and the unmanned aerial vehicle can be improved, the flight range of the unmanned aerial vehicle for executing the ground unloading task is reduced, and the flight energy consumption of the unmanned aerial vehicle is correspondingly reduced.

The following describes the embodiment of the present invention in detail with reference to a specific example. The specific embodiment was simulated using MATLAB software. The specific parameters are set as shown in table 1.

TABLE 1

In addition to the above table, the total task amounts of the far user and the near user are set to 45MBit, respectively, and the coordinates of the far user are set to { s } ₁ ,s ₂ ,s ₃ ,s ₄ }＝{(20,20),(20,-20),(-20,-20),(-20,20)}，The coordinates of the near user are set to { s } _m1 ，s _m2 ，s _m3 ，s _m4 }＝{(70，70)，(70，-70)，(-70，-70)，(-70，70)}。

As can be seen from a comparison of the trajectories of the unmanned aerial vehicle at different task completion times in fig. 3, the UAV uses its mobility to find the best position in each slot to complete the computational offload as the task completion time increases, with the requirement of a fixed ground user's task completion amount. It can be observed that as the task completion time T increases, the trajectory tends to approach the near user s ₁ Sum s ₄ . This enhances the communication link and the curve tends to be smooth, indicating that in a short time the drone is a compromise in order to be able to complete the flight and perform the unloading.

Fig. 4 shows a graph of energy consumption versus various schemes at a fixed task completion time, with different ground user task completion demands. Compared with other schemes, the scheme can always achieve the best performance, and as the task requirement of each ground user increases, the scheme provided by the invention has more obvious advantages. Compared with other schemes, the scheme provided by the invention has obvious advantages that no user cooperation exists, and the remote user considers the unmanned plane cooperation scheme and the preset unmanned plane track scheme, because all users have the cooperation participation of the users or the unmanned plane, the task quantity of local and unloading or relay can be optimally distributed. It can be seen from the figure that no user cooperates and the remote user considers that the unmanned plane cooperation scheme has obvious difference with the increase of the task amount of the ground user, because the local calculation and the unloading amount cannot be optimally distributed with the increase of the task amount after the maximum transmitting power of the remote user is reached due to the limitation of the transmitting power of the ground user.

In fig. 5, the relation of the system weighting energy consumption to the time delay T is depicted. It can be seen that the weighted energy consumption of all schemes decreases with increasing T, which coincides with the intuition that there is a trade-off between energy consumption and time consumption to accomplish the same task, and that energy consumption will decrease with increasing time consumption. Notably, the proposed solutions are superior to other benchmarks and the performance improvement is more pronounced under severe time constraints (small T), which further demonstrates that the algorithm can handle delay-critical computational tasks well and can achieve energy-to-delay tradeoff.

Fig. 6 shows the convergence of the algorithm at different task completion times, which indicates that the algorithm provided by the invention has good convergence.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

The foregoing is merely a preferred embodiment of the present invention, and it should be noted that modifications and variations could be made by those skilled in the art without departing from the technical principles of the present invention, and such modifications and variations should also be regarded as being within the scope of the invention.

Claims (2)

1. The unmanned aerial vehicle auxiliary mobile edge computing system optimizing method based on user cooperation is characterized in that,

among the ground users, the far user m _i Pairing with near user i, m _i ∈[1,M],i∈R，R＝{1,…,I}，M＝I；

Based on unmanned aerial vehicle auxiliary mobile edge computing system model, optimal number of remote users m _i Relaying the calculation task to be processed to a near user i; the unmanned aerial vehicle auxiliary mobile edge computing system model is obtained through the following steps:

1) Constructing an initial model of an unmanned aerial vehicle auxiliary mobile edge computing system of user cooperation; during the task completion time T, the unmanned aerial vehicle flies at a fixed height H, and the remote user m _i And the height of the near user i is 0 m, the position is fixed, and the coordinate of the near user i is s _i ＝(x _i ,y _i ) Far user m _i Is the coordinates of (a)

The near user and the unmanned aerial vehicle communicate in a time division multiplexing mode;

dividing task completion time T into N slots, N e {1, …, N }, width δ=t/N of each slot;

within the task completion time T, the unmanned aerial vehicle starts from the starting point u _I Fly to destination u _F ；

The speed of the unmanned plane in the nth time slot is as follows:

v[n]＝||u[n]-u[n-1]||/δ；

the initial model of the unmanned aerial vehicle auxiliary mobile edge computing system is as follows:

s.t.C1:

C2:

C3:

C4:

C5:

C6:

C7:

C8:u[0]＝u _I ,u[N]＝u _F

C9:

C10:

C11:

C12:

in the minimum objective function, u= { U [ n ]]Is the unmanned plane from the initial point u _I Fly to the end point u _F Is a flying trace of (a);

Total number of tasks to be processed, relayed and offloaded for the ground user, +.>

For far user m _i The number of calculation tasks to be processed relayed to the near user i in the nth time slot,/-, is->

For near user i to far user m in nth time slot _i The number of pending calculation tasks offloaded to the unmanned plane u +.>

Unloading the number of the self to-be-processed calculation tasks for the near user i in the nth time slot;

calculating frequency for ground users and CPU of unmanned aerial vehicle, f _i [n]Calculating the frequency for the CPU of the near user, +.>

For far user m _i CPU calculation frequency f _i,u [n]CPU calculation frequency for processing the to-be-processed calculation task of the near user i for the unmanned aerial vehicle,/for the unmanned aerial vehicle>

Handling near user i offloaded from far user m for drone _i CPU computing frequency of the computing task to be processed;

θ ₁ weighting coefficient, θ, for total energy consumption of ground users ₂ A weighting coefficient for the total energy consumption of the unmanned aerial vehicle;

wherein, c ₁ And c ₂ Two preset parameters related to the weight, wing area, wingspan efficiency and air density of the unmanned aerial vehicle;

E _i [n]calculating the number L of the completed calculation tasks to be processed in the nth time slot for the near user i _i [n]The required energy consumption:

E _i [n]＝δκ _i f _i ³ [n]，

wherein, kappa _i An effective capacitance coefficient for near user i;

for far user m _i Calculating the number of completed calculation tasks to be processed in the nth time slot +. >

The required energy consumption:

in the method, in the process of the invention,

representing far user m _i Effective capacitance coefficient of (a);

to accomplish->

The required energy consumption:

in the far user m _i The time width of relaying the task to be processed to the near user i, the near user i relays the far user m _i The time width of unloading the calculation task to be processed to the unmanned plane u and the time width of unloading the calculation task to be processed to the unmanned plane u by the near user i are delta ₀ ＝δ/3I；

For far user m _i Transmit power of>

For the received noise power of near user i, +.>

For near user i and far user m _i The channel gain of the link between the two, B is the communication bandwidth;

to accomplish->

The required energy consumption:

in the method, in the process of the invention,

will come from far user m for near user i _i Any of the pending calculations of (a)Transmitting power of service offloading to unmanned plane u, < ->

For receiving noise power of the unmanned aerial vehicle, a free space path loss model is adopted for a channel between the unmanned aerial vehicle and a near user, and the unmanned aerial vehicle is in a +.>

For the channel gain between the near user i and the unmanned plane u, d _i [n]Gamma is the distance between the near user i and the unmanned plane u ₀ For reference distance d ₀ Channel gain at=1m;

to accomplish->

The required energy consumption:

in the method, in the process of the invention,

unloading the self calculation task to be processed to the transmitting power of the unmanned plane u for the near user i;

E _i,u [n]for the unmanned aerial vehicle to calculate the number of completed pending calculation tasks from the near user i itself in the nth time slot

L _i,u [n]The required energy consumption:

wherein, kappa _u Is the effective capacitance coefficient of the unmanned aerial vehicle;

offloaded from far user m for near user i completed for drone computation in nth time slot _i The number of computing tasks to be processed +.>

The required energy consumption:

in the method, in the process of the invention,

handling far user m for unmanned aerial vehicle _i The CPU calculation frequency of (2);

equation C1 ensures that near user i completely offloads from far user m during each time slot _i N, N ₁ = {1,..n-1 }; equation C2 is an information causal randomness limit, the drone can only calculate the pending calculation task that has been received in the last time slot, and the processing delay is one time slot,

for the near user i to offload its own number of computational tasks to be processed in the kth time slot, k= {1, …, n-1}, L _i,u [k]For the unmanned u to calculate the number of completed self-pending calculation tasks offloaded from the near user i in the kth time slot, N ₂ ＝{2,...,N}；

Equation C3 is the information causal random limit,

for near user i to far user m in kth time slot _i The number L of the calculation tasks to be processed is unloaded to the unmanned plane u _mi,u [k]Offloaded from far user m for near user i completed by drone computation in kth slot _i The number of computing tasks to be processed;

Equation C4 ensures that all the near user i's own tasks to be processed, which are offloaded in the nth time slot, can be unmannedCompletion of computer calculation, L _i,u [n]Calculating the number of completed tasks to be processed from the near user i per se for the unmanned plane u in the nth time slot;

equation C5 ensures that near user i is offloaded to drone u from far user m in the nth time slot _i Can be calculated and completed by the unmanned aerial vehicle,

near user i offloaded from far user m indicating completion of drone u computation in nth time slot _i The number of computing tasks to be processed;

equation C6 ensures that the unmanned aerial vehicle calculates the distance user m from the distance user who finishes the unloading of the near user i within the task completion time T _i Is used for processing the calculation task to be processed,

for far user m _i The number of calculation tasks to be processed for local calculation in the nth time slot, +.>

For far user m _i Is used for calculating task demands;

formula C7 ensures that the unmanned plane completes the task to be processed and calculated unloaded by the near user i in the task completion time T, L _i [n]The number of calculation tasks to be processed for local calculation of near user I in the nth time slot, I _i Representing the computing task requirements of a near user i;

equation C8 limits the unmanned aerial vehicle from the initial point u within the task completion time T _I Fly to the end point u _F ，u[0]＝u _I ，u[N]＝u _F The method comprises the steps of carrying out a first treatment on the surface of the Equation C9 limits the maximum speed of the unmanned aerial vehicle, u [ n ]]＝u(δn)＝(x[n],y[n]) For the projected coordinates of the unmanned u on the horizontal plane in the nth time slot, u [ n-1 ]]For the projected coordinates of the drone u on the horizontal plane in the n-1 th time slot, V _max Is the maximum speed of the unmanned aerial vehicle;

the formula C10 ensures that the ground user does not relay or unload the calculation task to be processed in the last time slot, and ensures that the unmanned aerial vehicle does not process the calculation task to be processed in the first time slot;

equation C11 limits the non-negative or maximum value, f _i,max For the maximum calculated frequency of far user i, f _mi,max For far user m _i Maximum calculated frequency of f _u,max The maximum calculation frequency of the unmanned plane is calculated;

c12 is the maximum limit for near user transmit power, the maximum limit for far user transmit power;

P _max for the maximum of the transmit power of all terrestrial users,

for near user i to far user m in nth time slot _i Is offloaded to the transmit power of unmanned plane u, +.>

For the next user i to unload its own transmit power of the calculation task to be processed in the nth time slot,/>

For far user m _i Relaying the computational task to be processed to the transmit power of near user i in the nth time slot

2) Setting iteration number variable ζ=1, and ending the iteration

Giving a flight path U of the unmanned aerial vehicle;

3) Under the condition of a given unmanned aerial vehicle flight trajectory, optimizing an initial model of an unmanned aerial vehicle auxiliary mobile edge computing system until a sub-problem meets a convergence condition, stopping iteration, and obtaining an optimal ground user, a CPU computing frequency F of the unmanned aerial vehicle and the total number L of to-be-processed computing tasks relayed and unloaded by the ground user; comprising the following steps:

3.1 Under the condition of the given unmanned aerial vehicle flight track, the initial model of the unmanned aerial vehicle auxiliary movement edge computing system to be optimized is as follows:

s.t.C1-C7,C10-C12,

3.2 Given the dual variable β= { β corresponding to the inequality constraint C1-C3 _i,n }、λ＝{λ _i,n Sum η= { η _i,n A value of };

3.3 Determining a binary variable ζ= { ζ corresponding to the equation constraint C4-C7 based on a binary search method according to values of the binary variables β, λ, and η _i }、ω＝{ω _i }、μ＝{μ _i Sum ρ= { ρ _i A value of };

3.4 Using Karush-Kuhn-Tucker conditions to solve the neutron problem of the initial model of the unmanned aerial vehicle auxiliary mobile edge computing system, wherein the solution is as follows:

wherein C is _i The computational resources required to compute a one-bit input bit for near user i,

for far user m _i Computing resources required to compute a one-bit input bit;

setting up

Far user m _i The maximum number of computational tasks to be processed relayed to near user i in the nth time slot, near user i unloading from far user m in the nth time slot _i The number of the maximum to-be-processed computing tasks of the near user i unloading itself in the nth time slot is respectively as follows:

[x] ⁺ ＝max{x,0}，

3.5 Iterative updating of the dual variables β, λ and η) based on the algorithm of the secondary gradient:

where j is the iteration index variable of the secondary gradient algorithm iteration,

and->

Represents the jth iteration step, respectively, of obtaining the dual variables β, λ and η, +.>

f _i,u,j-1 [n]And->

Respectively is

f _i,u [n]And->

Variable values of the corresponding j-1 th iteration;

3.6 Repeating the steps 3.2) to 3.5) until the objective function value of the sub-problem meets the set iteration termination precision, and obtaining optimal F and L after iteration termination;

4) Inputting and optimizing an initial model of an unmanned aerial vehicle auxiliary moving edge computing system by using the optimal ground user, the CPU computing frequency F of the unmanned aerial vehicle and the total number L of the to-be-processed computing tasks relayed and unloaded by the ground user until the set convergence condition is met, and outputting an optimal unmanned aerial vehicle flight track U; comprising the following steps:

4.1 Inputting the optimal ground user and the CPU calculation frequency F of the unmanned aerial vehicle and the total number L of the to-be-processed calculation tasks relayed and unloaded by the ground user into an initial model of the unmanned aerial vehicle auxiliary mobile edge calculation system:

s.t.C8,C9 and C12；

4.2 The flight energy consumption of the unmanned aerial vehicle is as follows:

due to u [ n ]]Is non-convex with respect to the equation, so introduces a relaxation variable

To obtain E ^fly [n]Upper bound of (2):

the added constraint translates into the following form:

wherein u is ^(r) [n]Is a local point of the flight path of the unmanned aerial vehicle;

in step 4.1), updating an initial model of the unmanned aerial vehicle auxiliary mobile edge computing system as follows:

s.t.C8,C9andC12，

C13:

4.3 Using an initial unmanned aerial vehicle flight trajectory u [ n ]]As a local point, updating the iterative unmanned aerial vehicle track u [ n ] based on the interior point method]Until the convergence accuracy of the interior point method meets the set convergence accuracy

Obtain the optimal unmanned plane flight track U

5) Inputting the optimal F, L and U obtained in the step 3) and the step 4) into a minimum objective function of an initial model of an unmanned aerial vehicle auxiliary moving edge computing system to obtain a value E _ζ ；

6) Updating the iteration number variable ζ=ζ+1;

7) Substituting the optimal unmanned aerial vehicle flight track U into the step 3), and repeating the steps 3) to 6) until the initial model of the unmanned aerial vehicle auxiliary movement edge computing system meets the convergence condition

The iteration is terminated, and an unmanned aerial vehicle auxiliary mobile edge computing system model is obtained;

based on unmanned aerial vehicle auxiliary mobile edge computing system model, optimal number of remote users m _i And unloading the to-be-processed calculation tasks relayed to the near user i and the optimal number of to-be-processed calculation tasks of the near user i to the unmanned aerial vehicle for calculation.

2. The unmanned aerial vehicle assisted mobile edge computing system optimization method based on user collaboration according to claim 1, wherein step 3.3) comprises:

3.3.1 The solution of the neutron problem is constrained by equations C4-C7 and step 3.4), and is obtained

And

different expressions relating to values;

and (3) with

The different expressions relating to the values are:

and (3) with

The different expressions relating to the values are:

and (3) with

The different expressions relating to the values are:

3.3.2 Given lambda, combine

Scope of->

The range of the obtained dual variable ρ is:

when ρ is _i Given, the range of binary search for the dual variable ζ is:

wherein ρ is _i,min As a dual variable ρ _i Minimum value ρ of _i,max As a dual variable ρ _i Is set at the maximum value of (c),

in the case of n=1

A value; when the binary search is terminated, the result is at a given ρ _i Corresponding xi under the condition of (2) _i Values such that

Continuously binary search ρ _i And iterate the corresponding ζ _i Values such that

When the binary search method converges, a dual variable rho is obtained _i And xi _i Is the optimum value of (2);

3.3.3 Given beta), binding

Scope of->

The range of the binary variable mu binary search is:

when the binary search converges, a dual variable mu is obtained _i Is the optimum value of (3):

3.3.4 Given η and β), combined

Scope of->

The range of obtaining the binary variable omega binary search is as follows:

when the binary search method converges, the dual variable omega is obtained _i Is the optimum value of (3):

Images