An Energy Saving Method and System for NOMA-based UAV-assisted MEC System

Abstract

The present invention relates to the technical field of UAV-assisted edge computing, and is an energy-saving method and system for a NOMA-based UAV-assisted MEC system. First, a NOMA-based UAV-assisted MEC system is constructed. Given the computing tasks and task deadlines, a joint optimization scheme for computing resource allocation, transmission power, and UAV trajectory is proposed to minimize the system's weighted energy consumption. The total energy consumption of the system within a preset time is expressed as the weighted sum of communication energy consumption, computing energy consumption, and propulsion energy consumption; the total energy consumption is modeled as a formulated optimization problem through a mathematical model; the optimization problem is decoupled into sub-problems of transmission power and computing resource allocation, and a sub-problem of UAV trajectory scheduling; an iterative optimization algorithm is proposed to alternately solve each sub-problem until the algorithm converges. The present invention solves the problem of energy limitation faced in existing NOMA-based UAV-assisted MEC systems, and can support the access of future large-scale users and meet the requirements of emergency computing offloading services.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV)-assisted edge computing, and particularly to an energy-saving method and system for a UAV-assisted multi-access edge computing (MEC) system based on non-orthogonal multiple access (NOMA). Background Art

[0002] With the development of 5G, emerging applications represented by virtual reality / augmented reality, vehicle-to-everything, natural language processing, etc. have emerged continuously. These applications are usually some computationally intensive and latency-sensitive tasks with large amounts of data and high computational power requirements, resulting in an explosive growth trend in mobile data traffic. Given the limited computational power and battery life of mobile terminals (MTs), they cannot provide satisfactory service capabilities. MEC is considered a promising method to solve this problem because it supports MTs with limited computational power to offload computational tasks to computational servers near the network edge to provide computing power assistance for MTs. Considering the problems of traditional terrestrial network communication facilities being restricted by geographical environment and poor flexibility, it is difficult to meet the communication and computing needs in hot spots or areas with network paralysis. Fortunately, as an important part of the air-ground integrated network, the UAV communication system provides reliable and on-demand computing support for MTs by carrying computational servers due to its advantages such as high mobility and flexible deployment. Therefore, the UAV-assisted air-ground collaborative MEC system framework has attracted much attention since it was proposed.

[0003] Different from the ground base station MEC system that does not need to consider the energy consumption problem, in the UAV-assisted air-ground cooperative MEC system, the UAV not only needs to consider the limited onboard energy and computing capabilities of MTs and the UAV, but also needs to consider the additional propulsion energy consumption of the UAV to support its own flight. Therefore, the system architecture, energy efficiency management, resource allocation, and trajectory optimization in the UAV-assisted air-ground cooperative MEC system have become important research directions. Considering that arbitrarily distributed MTs may face interference from the same or adjacent channels, the UAV-assisted air-ground cooperative MEC system still faces many challenges such as high energy consumption, low spectral efficiency, and difficult cooperative communication. On the one hand, due to the multi-dimensional resource high coupling faced by the UAV-assisted air-ground cooperative MEC system, the existing UAV trajectory scheduling and resource allocation schemes for single communication and computing dimensions are difficult to apply. Therefore, in order to improve the communication cooperation performance of the system and avoid potential UAV collisions, a key challenge is to jointly optimize the multi-dimensional resource (communication and computing resources) allocation and elaborate UAV trajectory design. On the other hand, from the perspective of the transmission protocol, the existing research models are generally based on time division multiple access or frequency division multiple access technologies, and do not fully consider the large-scale access requirements in the UAV-assisted MEC system. To cope with the challenges of complex communication scenarios and the sharp increase in the number of terminals, and to make full use of the channel capacity from each MT to the server, researchers have begun to try to apply non-orthogonal multiple access (NOMA) technology to the UAV-assisted air-ground cooperative MEC system. Different from traditional orthogonal transmission technologies, NOMA allows multiple users to superimpose and send information in the same time-frequency domain, and uses serial interference cancellation technology at the receiving end to achieve correct demodulation.

[0004] The current research on the NOMA-based UAV-assisted air-ground cooperative MEC system is still in its infancy. Considering the tight coupling of communication and computing resources, the traditional performance indicators focusing on a single communication or computing dimension are difficult to meet the joint performance optimization requirements of communication and computing resources. Therefore, it is necessary to jointly establish an evaluation system for UAV trajectories, communication, and computing resources. When the task data volume of MTs is relatively large, the introduction of the NOMA protocol will bring interference and affect the total system energy consumption. And the allocation of computing resources directly affects the computing delay, and the uncertainty of the UAV's movement trajectory determines the link channel gain between the UAV and the MT and the sorting of successive interference cancellation, thus affecting the total amount of transmitted data and the system energy consumption, making the optimization problem difficult to solve. Therefore, how to efficiently utilize the limited communication and computing resources and the energy supply of the unmanned aerial vehicle, design the UAV trajectory planning and the coordinated allocation scheme of computing power and network resources to reduce the system weighted energy consumption is the key to improving the performance of the air-ground cooperative MEC system. Summary of the Invention

[0005] The object of the present invention is to provide an energy-saving method for a NOMA-based UAV-assisted MEC system, so as to solve the problem of limited system energy efficiency in an air-ground cooperative MEC network based on NOMA and UAV assistance, and to support the access of a large number of future users and meet the requirements of emergency computing offloading services.

[0006] Another object of the present invention is to provide an energy-saving system for a NOMA-based UAV-assisted MEC system.

[0007] To achieve the above object, the present invention provides an energy-saving method for a NOMA-based UAV-assisted MEC system, including the following steps:

[0008] S1. Construct a NOMA-based UAV-assisted air-ground cooperative MEC network system, including m mobile terminals MT and u UAVs carrying MEC servers;

[0009] S2. On the premise of given computing tasks and task deadlines, propose a joint optimization scheme for computing resource allocation, transmission power, and UAV trajectories to minimize the system weighted energy consumption;

[0010] Construct a communication channel model, a computing model, and an energy consumption model. The communication channel model is used to calculate path loss and channel gain, and the computing model is used to calculate the CPU frequencies of each mobile terminal and UAV in each time slot; based on the calculation results of the computing model, the energy consumption model further calculates the energy consumption, including local computing energy consumption, edge computing energy consumption, communication transmission energy consumption, and flight propulsion energy consumption;

[0011] Express the total energy consumption of the system within a preset time as a weighted sum of communication energy consumption, computing energy consumption, and propulsion energy consumption;

[0012] S3. Model the total energy consumption as a formulated optimization problem through a mathematical model;

[0013] S4. Decouple the optimization problem into two tractable sub-problems, namely, the sub-problem of transmission power and computing resource allocation, and the sub-problem of UAV trajectory scheduling;

[0014] S5. Propose an iterative optimization algorithm to alternately solve each sub-problem until the algorithm converges;

[0015] During the iteration process, first, under the condition of the given UAV trajectory, perform equivalent transformation on the objective function and constraint conditions, and use the method of successive convex approximation to solve the allocation of transmission power and computing resources; then, under the given computing resources and allocated power, introduce a slack variable and the method of first-order Taylor series approximation to solve the optimal trajectory planning of the UAV.

[0016] The present invention provides an energy-saving system for a NOMA-based unmanned aerial vehicle (UAV)-assisted mobile edge computing (MEC) system, including the following modules:

[0017] A system construction module, configured to construct a NOMA-based UAV-assisted air-ground collaborative MEC network system, including m mobile terminals (MTs) and u UAVs equipped with MEC servers;

[0018] A total energy consumption calculation module, configured to propose a joint optimization scheme for computing resource allocation, transmission power, and UAV trajectory under the premise of given computing tasks and task deadlines, so as to minimize the weighted energy consumption of the system;

[0019] It is also configured to construct a communication channel model, a computing model, and an energy consumption model. The communication channel model is used to calculate path loss and channel gain. The computing model is used to calculate the CPU frequencies of each mobile terminal and UAV in each time slot. Based on the calculation results of the computing model, the energy consumption model further calculates the energy consumption, including local computing energy consumption, edge computing energy consumption, communication transmission energy consumption, and flight propulsion energy consumption;

[0020] Express the total energy consumption of the system within a preset time as a weighted sum of communication energy consumption, computing energy consumption, and propulsion energy consumption;

[0021] A modeling module, which models the total energy consumption as a formulated optimization problem through a mathematical model;

[0022] A transformation module, configured to decouple the optimization problem into two tractable sub-problems, namely, the sub-problem of transmission power and computing resource allocation, and the sub-problem of UAV trajectory scheduling;

[0023] An iteration module, configured to propose an iterative optimization algorithm to alternately solve each sub-problem until the algorithm converges;

[0024] During the iteration process, first, under the condition of the given UAV trajectory, perform equivalent transformation on the objective function and constraint conditions, and use the method of successive convex approximation to solve the allocation of transmission power and computing resources; then, under the given computing resources and allocated power, introduce a slack variable and the method of first-order Taylor series approximation to solve the optimal trajectory planning of the UAV.

[0025] Compared with the prior art, the beneficial effects of the present invention include:

[0026] 1. The present invention establishes an air - ground collaborative system based on NOMA and multiple UAVs, constructs a communication channel model, a computing model, and an energy consumption model based on the air - ground collaborative system; based on the communication channel, computing, and energy consumption models, a joint optimization scheme for computing resources, transmission power allocation, and UAV trajectories is proposed to minimize the system - weighted energy consumption under the premise of given computing tasks and task deadlines. Due to the coupling between multiple variables, the optimization problem is non - convex. To address this non - convexity problem, the present invention decomposes the weighted energy consumption minimization problem into two sub - problems and designs an efficient iterative optimization algorithm, alternately optimizing and solving each sub - problem through successive convex approximation, introducing slack variables, and Taylor - series approximation methods until the algorithm converges.

[0027] 2. The present invention is effectively applicable to NOMA - and UAV - assisted MEC networks with requirements such as low latency, high reliability, and high efficiency, can support the access of a large number of future users, and meet the demand for emergency computing offloading services. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] To more clearly illustrate the technical solutions of the present invention, the accompanying drawings required for implementation will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0029] Figure 1 It is a schematic flowchart of an energy - saving method for a NOMA - based UAV - assisted MEC system provided by an embodiment of the present invention;

[0030] Figure 2 It is a schematic diagram of an air - ground collaborative MEC system based on NOMA and UAV assistance provided by an embodiment of the present invention;

[0031] Figure 3 It is a schematic diagram of the time - slot framework of the task completion time T in the NOMA mode provided by an embodiment of the present invention;

[0032] Figure 4 It is a flight - trajectory simulation diagram for different numbers of MTs in the NOMA mode provided by an embodiment of the present invention; among them, (a) is the flight - trajectory simulation diagram when deploying one UAV, and (b) is the flight - trajectory simulation diagram when deploying multiple UAVs;

[0033] Figure 5 It is a flight - trajectory simulation diagram for different numbers of MTs in the OMA mode provided by an embodiment of the present invention; among them, (a) is the flight - trajectory simulation diagram when deploying one UAV, and (b) is the flight - trajectory simulation diagram when deploying multiple UAVs;

[0034] Figure 6It is the simulation diagram of energy consumption composition under NOMA and OMA modes provided by the embodiments of the present invention;

[0035] Figure 7 It is the simulation diagram of the system weighted energy consumption and task processing time provided by the embodiments of the present invention;

[0036] Figure 8 It is the simulation diagram of the system weighted energy consumption and the number of MTs provided by the embodiments of the present invention. Detailed implementation manners

[0037] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0038] Embodiment 1

[0039] This embodiment discloses an energy-saving method for a NOMA-based UAV-assisted MEC system, as Figure 1 shown, including the following steps:

[0040] S1. Construct a multi-UAV-assisted air-ground collaborative MEC network system based on NOMA.

[0041] In this step, NOMA technology is introduced into the MEC system to construct an air-ground collaborative system based on NOMA and multi-UAV assistance. The constructed MEC system includes m mobile terminals (MTs) and u UAVs equipped with MEC servers. The UAVs provide computing services for the mobile terminals, which are respectively represented as M = {1, 2,... m} and U = {1, 2,... u}; the position vector of MT m at the nth time slot is represented as The flight altitude of the UAV is h, then the position vector of UAV u at the nth time slot can be represented as

[0042] In this step, a computing server is installed on the UAV to provide computing services for the MT to support the MTs with limited computing capabilities to complete task offloading. Both the UAVs and the mobile terminals MTs are equipped with separate antennas. The MTs are randomly and evenly deployed in the specified area and collect sensing data from the surrounding environment in real time. The task of the UAVs is to fly from the designated initial position to the end point and assist the MTs to complete the computing task offloading.

[0043] S2. Given a computing task and a task deadline, a joint optimization scheme for computing resource allocation, transmission power, and UAV trajectory is proposed to minimize the weighted energy consumption of the system.

[0044] In this step, a communication channel model, a computing model, and an energy consumption model are constructed to solve the problems of the existing spectrum resources being unable to support the access of a large number of users and the system energy consumption. Among them, the communication channel model is used to calculate the path loss and channel gain, and the computing model is used to calculate the CPU frequencies of each mobile terminal and UAV in each time slot; based on the calculation results of the computing model, the energy consumption model further calculates the energy consumption, including local computing energy consumption, edge computing energy consumption, communication transmission energy consumption, and flight propulsion energy consumption.

[0045] Specifically, the above communication channel model is a communication channel model that is the probability average of the line-of-sight link and the non-line-of-sight link. The LOS and NLOS probabilities between UAVs and MTs calculated by the communication channel model are:

[0046]

[0047] Among them, the parameters a and b are constant values depending on the environment. ∈ m,u [n] is the elevation angle between the user and the UAV, that is:

[0048]

[0049] Among them, d is the horizontal distance between MT m and UAV u, and h is the flight altitude of the UAV.

[0050] In the UAV-assisted air-to-ground channel, the LOS and NLOS path losses between MT m and UAV u can be expressed as:

[0051]

[0052] Among them, f c is the carrier frequency, η LoS and η NLoS are additional path losses, and c is the speed of light.

[0053] Therefore, the average path loss can be expressed as:

[0054]

[0055] Considering small-scale fading, the channel gain between MT m and UAV u at time slot n can be expressed as:

[0056]

[0057] g m,u [n] is the channel gain.

[0058] When constructing the energy consumption model, attention is paid to the technical indicators of system energy consumption, which specifically include the following three parts: 1) The computing energy consumption of tasks, which includes the local computing energy consumption of tasks and the edge computing energy consumption of some tasks on UAVs; 2) The communication transmission energy consumption of tasks unloaded to UAVs; 3) The flight propulsion energy consumption of UAVs. It should be noted that UAVs do not perform task computing in the first time slot and the last time slot, and task offloading does not occur for MTs in the last time slot.

[0059] In the computing model, each MT and UAV adopts the dynamic frequency scaling (DFS) technology to adjust the CPU frequencies of MT m and UAV u in each time slot. and is generally related to the number of bits of the computing tasks executed by MT m and UAV u in the nth time slot, that is: It is related to the number of bits of the computing tasks executed by MT m and UAV u in the nth time slot, that is:

[0060]

[0061] where ρ k is the task computing density.

[0062] Then the local computing energy consumption of MTs in all time slots can be expressed as:

[0063]

[0064] The edge computing energy consumption of UAVs in all time slots can be expressed as:

[0065]

[0066] where is the effective switching capacitance coefficient of MT m and UAV u, which depends on the chip structure of the processor.

[0067] MT m uses the uplink NOMA technology to offload computing tasks to UAVs through the wireless link. To avoid interference when MTs offload tasks to multiple UAVs, this embodiment assumes that different UAVs occupy a frequency band through frequency division multiple access, and different MTs share the same frequency band when offloading data to the same UAV. Therefore, only MTs offloaded to the same UAV will cause interference, and different UAVs will not cause interference because they operate on different frequency bands. Without loss of generality, in this embodiment, the MTs sharing the same frequency band in each time slot are sorted according to the channel gain, that is where j k,u[n] ∈ U represents the channel gain index value of the k-th smallest MT unloaded to UAV u in the n-th time slot. The UAV receiver adopts SIC technology to eliminate multi-user interference. The basic idea of SIC is to first sort the users according to the received signal strength, then preferentially decode the signal with the largest signal strength, regard other signals as interference signals, and finally delete the decoded signal from the mixed superimposed signal. The subsequent decoded signals will not be interfered by this signal, and so on. In the n-th time slot, MTj k,u The data transmission rate of [n] unloaded to UAV m can be expressed as:

[0068]

[0069] where B represents the channel bandwidth allocated to each UAV, represents the transmit power of MTj k,u [n], and σ 2 is the Gaussian white noise power.

[0070] To ensure the successful offloading of computing tasks, the data transmission rate should satisfy:

[0071]

[0072] Therefore, the communication transmission energy consumption of the system over a period of time is expressed as:

[0073]

[0074] In the UAV-assisted air-ground cooperative MEC system, the propulsion energy consumption of the UAV in time T can be:

[0075]

[0076] where P 0 is the profile power of the rotor in the hovering state of the UAV, and P 1 is the induced power of the UAV in the hovering state, is the tip speed of the rotor, v O is the average induced speed of the rotor, d O is the fuselage drag ratio, s is the rotor stability, ρ is the air density, and A is the rotor disk area.

[0077] In summary, the total energy consumption E of the system within the preset time T sum can be expressed as the weighted sum of communication transmission energy consumption, computing energy consumption, and propulsion energy consumption, and can be expressed as:

[0078]

[0079] where is the local computing energy consumption, is the edge computing energy consumption, is the communication transmission energy consumption, is the propulsion energy consumption; ω mt ∈1 and ω uav ∈1 are the energy consumption weight factors of the MT and UAV respectively, and satisfy ω mt +ω uav = 1. The energy consumption factors can be adjusted according to the actual system and preferences to balance the energy consumption of the UAV and MTs. η is the flight energy consumption coefficient of the UAV, which is used to reduce the difference in the energy consumption levels of the UAV and MTs in the actual environment. θ is the length of each time slot, is the effective switching capacitance coefficient of the mobile terminal, is the effective switching capacitance coefficient of the UAV, is the CPU frequency of the mobile terminal MT m at the nth time slot, is the CPU frequency of the UAV UAV u at the nth time slot.

[0080] S3. Model the above total energy consumption as a formulated optimization problem through a mathematical model.

[0081] In this step, by jointly optimizing the transmission power p of the task, the computing resources f m on the MTs, the computing resources f u on the UAVs, and the flight trajectory q u of the UAVs, the system weighted energy consumption within the preset time T is minimized. The mathematical model is described as follows:

[0082] The objective function is problem P1:

[0083] The constraints are:

[0084] s.t. C1:

[0085] C2:

[0086] C3:

[0087] C4:

[0088] C5:

[0089] C6:

[0090] C7:

[0091] C8:

[0092] C9:

[0093] C10:

[0094] where ω mt ∈1 and ω uav ∈1 are the energy consumption weight factors of the MT and UAV respectively, and satisfy ω mt + ω uav = 1. The energy consumption factors can be adjusted according to the actual system and preferences to balance the energy consumption of the UAV and MTs. η is the flight energy consumption coefficient of the UAV, which is used to reduce the difference in the energy consumption levels of the UAV and MTs in the actual environment. The constraint condition C1 represents the limit of the uplink transmission power of the task. C2 - C3 are the CPU cycle frequency limits of the UAVs and MTs. C4 - C6 represent the causal relationship between the task communication transmission time and the offloading calculation. C7 - C10 represent the flight speed, position coordinates and battery capacity limits of the UAV.

[0095] S4. Decouple the optimization problem formulated in step S3 into two tractable sub - problems.

[0096] Since the constructed communication, computing and energy consumption models are proven to be non - convex optimization problems. Considering that the computing offloading involves the coupling of communication and computing processes, and the multi - dimensional resources are highly correlated and interact with each other, resulting in the objective optimization problem P1 having highly coupled non - convex constraints. To solve this non - convex optimization problem with multiple constraints, first decouple the original problem into two tractable sub - problems, namely the sub - problem of transmission power and computing resource allocation, and the sub - problem of UAV trajectory scheduling.

[0097] Specifically, since the flight trajectory q u of the UAV is coupled with other variables, it is difficult to directly solve the mixed - integer non - linear programming problem P1. Fortunately, the Hessian matrix of E sum may not be positive definite, but the second - order derivative sum with respect to each parameter of E is non - negative. Considering that C5 is highly correlated with the UAV's flight trajectory and C6 is a concave set with respect to the UAV's flight trajectory, P1 becomes a non - convex problem. To solve problem P1, first decouple problem P1 into two tractable sub - problems: (i) Solve the optimal solution of the computing resource and power allocation scheme when the UAV trajectory is fixed; (ii) Find the optimal UAV trajectory under the optimal computing resource and power allocation scheme.

[0098] S5. Propose an efficient iterative optimization algorithm to alternately solve each sub - problem decoupled in step S4 until the algorithm converges.

[0099] During the iteration process, first, under the trajectory conditions of the given UAV, the objective function and constraint conditions are equivalently transformed, and the continuous convex approximation method is used to solve the allocation of transmission power and computing resources. Then, under the given computing resources and allocated power, slack variables, first-order Taylor series approximation, and SVA method are introduced to alternate each sub-problem until the increase in the objective function is lower than the given threshold or the number of iterations is large enough, that is, until the algorithm converges, and the optimal trajectory planning of the UAV is obtained by solving.

[0100] The specific steps of the iteration are as follows:

[0101] S51. Under the trajectory of the given UAV, the optimization problem of the allocation of transmission power and computing resources is simplified to:

[0102] The objective function is Problem P1.1:

[0103] The objective constraint conditions are: C1 to C6

[0104] Since the trajectory of the UAV is predetermined, the propulsion energy consumption of the UAV (i.e., the last term in Problem P1.1) is a constant and can be considered negligible in Problem P1.1. Starting from the cumulative transmission power sum of all MTs at time slot n and from the perspective of recurrence, Problem P1.1 is transformed into a convex function for solution. For the convenience of calculation, the constant term and the denominator in the calculation formula of the data transmission rate are defined as follows:

[0105]

[0106] In addition, according to the calculation formula of the total energy consumption E sum it can also be directly observed that E sum is convex with respect to f m and f u However, the convexity and concavity of the total energy consumption E sum with respect to the variable p need to be verified according to the equation transformation.

[0107] First, the calculation formula of the data transmission rate is converted into an exponential form, expressed as:

[0108]

[0109] Given the cumulative transmission power sum of the MTs and its recurrence, Problem P1.1 can be transformed into an equivalent form of a convex function. For the convenience of calculation, the constant term and the denominator can be defined as N = ln2 / Bθ, Using the recurrence of can obtain

[0110] In addition, define the following can be obtained:

[0111]

[0112] The sum of the transmission powers of all MTs is as follows:

[0113]

[0114] Considering Therefore, the coefficient of each exponential function in the above formula is non - negative. Define Then the sum of the transmission powers of all MTs can be re - defined as:

[0115]

[0116] So far, the verification and solution of the convexity of problem P1.1 with respect to the variable p are completed.

[0117] In addition, this embodiment also defines w (u,n,m) as the sorting for MT m to select offloading to UAV u at time slot n. Then the data transmission volume at this time is:

[0118]

[0119] Redefine the objective function in the optimization problem P1.1 as σ:

[0120] The objective function is problem P1.2:

[0121] Objective constraint conditions: C2 - C5

[0122] C1′:

[0123] C6′:

[0124] Among them, C1′ and C6′ are respectively the equivalent transformation forms of C1 and C6 in the optimization problem P1.1. It can be observed from problem P1.2 that C2, C3, C4, C5, C6′ are all affine transformations with respect to the variables and are still convex constraints. However, the numerator in C1′ is the weighted sum of multiple exponential functions, which cannot guarantee that C1′ is convex. Therefore, this embodiment plans to use the SCA and first - order Taylor series approximation methods to transform C1′ into a convex constraint, thereby transforming problem P1.2 into a standard convex problem. Finally, use the traditional convex optimization algorithm (interior - point method) for solution, and obtain an approximate optimal solution by continuously updating the iteration factor when the convergence condition is satisfied.

[0125] Let f(z)=eJz , it can be observed that f(z) is a convex function with respect to z. Since the global lower bound of a convex function is its first-order Taylor expansion, we can obtain That is:

[0126]

[0127] Substitute the above equation into the constraint C1′, and rewrite the objective function optimization problem P1.2 as follows:

[0128] The objective function is problem P1.3:

[0129] The objective constraint conditions are: C2~C5, C6’

[0130] C1"":

[0131] where is the value of x k,u [n] at the (λ - 1)-th iteration. The constraint C1″ is expressed as the difference between an exponential function and an affine function, so it is guaranteed to be convex, thus ensuring the convexity of the optimization problem.

[0132] So far, problem P1.3 has been transformed into a standard convex problem. This embodiment can find a solution by using classical convex optimization tools (such as the CVX toolbox).

[0133] S52. The problem of optimizing the UAV trajectory under given power and computing resources.

[0134] In this step, consider optimizing the UAV trajectory under a given transmission power and computing resource allocation scheme. The optimization problem P1 can be formulated as:

[0135] The objective function is problem P2:

[0136] The objective constraint conditions are: C7~C10

[0137] Considering that the transmission power and computing resource allocation scheme are fixed, there is no need to consider the computing energy consumption and offloading transmission energy consumption in problem P2, and only the propulsion energy consumption of UAVs needs to be concerned. Since the second term in the component term is which is non-convex, resulting in problem P2 being a non-convex function; therefore, in this embodiment, a slack variable f[n] is introduced, that is:

[0138]

[0139] to solve P2. Substitute the slack variable into problem P2, and we get:

[0140] The objective function is Problem P2.1:

[0141] The constraint conditions are: C11:

[0142] C12:

[0143] C7 - C10

[0144] When the optimal solution of Problem P2.1 is obtained, the equal sign in the constraint condition C12 holds. It should be noted that for the first constraint condition, the left side is a joint convex function of f[n] and q u [n], but the right side is a concave function of q u . Using the SCA method, the right side of the constraint condition C11 is expressed by its first-order Taylor expansion as:

[0145]

[0146] It can be observed that the above first-order Taylor expansion is an affine function of f[n] and q u [n]. Among them, f[n] (l) and q u [n] (l) are the l-th iteration values of f[n] and q u [n] respectively. So far, the propulsion energy consumption of the UAV can be approximately expressed as a convex function. Combining with the derived lower bound function, the optimization problem P2.1 can be redefined as

[0147] The objective function is Problem P2.2:

[0148] The objective constraint conditions are:

[0149] C10':

[0150] C7 - C9, C11, C12

[0151] So far, Problem P2.2 has been transformed into a convex optimization problem and is equivalent to P2.1, which can be solved by using tools such as CVX. Based on the CVX tool, the local optimal solutions of each sub-problem are alternately iteratively solved to obtain the approximate optimal solution of the overall system algorithm.

[0152] The pseudo-code of the efficient iterative optimization algorithm designed in this embodiment is shown in Table 1 below:

[0153]

[0154] Table 1 Pseudo-code

[0155] This embodiment uses MATLAB to simulate the scenario, and through simulation analysis, the performance of the proposed NOMA-based multi-UAV-assisted air-ground cooperative resource allocation and trajectory optimization strategy is verified. It is considered that multiple MTs are randomly distributed within a 1*1 km 2 area, and each MT needs to complete a computationally intensive task of 150 Mb within the specified time T. The UAV can assist the MT in performing calculations simultaneously. The fixed flight altitude h of the UAV is 20 m, and its initial position and final position are fixed, which are r I,m =[0,0], r F,m =[1000,0], [0,1000], [1000,1000]. The maximum flight speed v max =15 m / s. According to the principle that the closer the weight factor is, the smaller the system energy consumption is, the weight coefficients are set as ω mt =0.6, ω uav =0.4.

[0156] The simulation parameters are set as shown in Table 2 below:

[0157]

[0158] Table 2 Simulation Parameter Settings

[0159] From Figure 4 and Figure 5 it can be seen that the location of the MT has a great impact on the trajectory design of the UAV. Under the two access modes, the flight trajectories of the UAVs are very similar. The UAV will try to get as close as possible to the MT during flight to improve the channel conditions from the MT to the UAV, so as to assist the MT in completing more task offloading. Further analysis and comparison of Figure 4 and Figure 5 the optimal flight trajectories of the UAVs in Figure 4 subfigure (a) of Figure 5 and Figure 4 subfigure (b) of Figure 5 found more effective conclusions. When deploying a single UAV, the UAV will fly a longer path to traverse each MT in the network to complete task offloading, and the path in the OMA mode is longer, such as Figure 4 subfigure (a) of Figure 5 and Figure 4 subfigure (b) of Figure 5 . When deploying multiple UAVs, the multiple UAVs cooperate with each other through communication to jointly complete the offloading tasks within the area, such as Figure 4 subfigure (b) of Figure 5 and

[0160] Figure 6 Figure 6It shows that compared with the OMA mode, the technical solution proposed in this embodiment reduces the system weighted energy consumption by 18.91% and the offloading energy consumption by 40.54%, and the composition of the system weighted energy consumption is more balanced. This is because in the NOMA mode, computing tasks tend to be processed on MEC servers with stronger computing power. In addition, more MTs reuse the same frequency band, which enables MTs to use a larger bandwidth and a smaller transmission power to complete task offloading, thereby reducing the task transmission energy consumption and improving the system performance.

[0161] In addition, the solution proposed in this embodiment is compared with the UAV fixed trajectory solution (flying directly from the initial position to the end position) and the resource equivalent allocation solution. Figure 7 It shows that the system weighted energy consumption under each solution shows a downward trend as the task completion time T increases, which reveals the trade-off relationship between energy consumption and delay when processing tasks. Figure 8 It shows that the system weighted energy consumption under each solution gradually increases as the number of MTs increases, and the weighted energy consumption of the solution proposed in this embodiment has a significant performance improvement compared with the other three benchmark solutions. This verifies the proposed solution, that is, by jointly optimizing the task communication channel, computing resource allocation and UAV trajectory scheduling under the NOMA transmission protocol, it can support the access of more users, effectively process delay-sensitive tasks, and reduce the system weighted energy consumption.

[0162] Embodiment 2

[0163] Based on the same inventive concept as Embodiment 1, the energy-saving system of a NOMA-based UAV-assisted MEC system provided in this embodiment specifically includes the following modules:

[0164] The system construction module is used to construct a NOMA-based UAV-assisted air-ground collaborative MEC network system, including m mobile terminals MTs and u UAVs carrying MEC servers;

[0165] The total energy consumption calculation module is used to propose a joint optimization scheme for computing resource allocation, transmission power and UAV trajectory under the premise of given computing tasks and task deadlines, so as to minimize the system weighted energy consumption;

[0166] It is also used to construct a communication channel model, a computing model and an energy consumption model. The communication channel model is used to calculate path loss and channel gain, and the computing model is used to calculate the CPU frequencies of each mobile terminal and UAV in each time slot; based on the calculation results of the computing model, the energy consumption model further calculates the energy consumption, including local computing energy consumption, edge computing energy consumption, communication transmission energy consumption and flight propulsion energy consumption;

[0167] The total energy consumption of the system within a preset time is expressed as the weighted sum of communication energy consumption, computing energy consumption and propulsion energy consumption;

[0168] A modeling module that models the total energy consumption as a formulated optimization problem through a mathematical model;

[0169] A transformation module for decoupling the optimization problem into two tractable sub-problems, namely, the sub-problems of transmission power and computing resource allocation, and the sub-problem of UAV trajectory scheduling;

[0170] An iteration module for proposing an iterative optimization algorithm to alternately solve each sub-problem until the algorithm converges;

[0171] During the iteration process, first, under the condition of the given UAV trajectory, perform equivalent transformation on the objective function and the constraint conditions, and use the method of successive convex approximation to solve the allocation of transmission power and computing resources; then, under the given computing resources and allocated power, introduce the method of slack variables and first-order Taylor series approximation to solve the optimal trajectory planning of the UAV.

[0172] The above-mentioned modules in this embodiment are respectively used to implement the corresponding steps in Embodiment 1, and the detailed implementation process refers to Embodiment 1.

[0173] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can think of various deformations or modifications. However, all modifications and substitutions made without departing from the spirit of the present disclosure will fall within the protection scope of the present disclosure defined by the appended claims.

Claims

1. An energy-saving method for a NOMA-based UAV-assisted MEC system, characterized in that, it includes the following steps: S1. Construct a NOMA-based UAV-assisted air-ground cooperative MEC network system, including m mobile terminals MT and u UAVs carrying MEC servers; S2. On the premise of given computing tasks and task deadlines, propose a joint optimization scheme for computing resource allocation, transmission power, and UAV trajectories to minimize the system's weighted energy consumption; Construct a communication channel model, a computing model, and an energy consumption model. The communication channel model is used to calculate path loss and channel gain, and the computing model is used to calculate the CPU frequencies of each mobile terminal and UAV per time slot. Based on the calculation results of the computing model, the energy consumption model further calculates the energy consumption, including local computing energy consumption, edge computing energy consumption, communication transmission energy consumption, and flight propulsion energy consumption; Express the total energy consumption of the system within a preset time as a weighted sum of communication energy consumption, computing energy consumption, and propulsion energy consumption; S3. Model the total energy consumption as a formulated optimization problem through a mathematical model; S4. Decouple the optimization problem into two tractable sub-problems, namely the sub-problem of transmission power and computing resource allocation, and the sub-problem of UAV trajectory scheduling; S5. Propose an iterative optimization algorithm to alternately solve each sub-problem until the algorithm converges; During the iteration process, first, under the condition of the given UAV trajectory, perform equivalent transformation on the objective function and constraint conditions, and use the method of successive convex approximation to solve the allocation of transmission power and computing resources; then, under the given computing resources and allocated power, introduce a relaxation variable and the method of first-order Taylor series approximation to solve the optimal UAV trajectory planning; Total energy consumption E sum is as follows: Among them is the local computing energy consumption, is the edge computing energy consumption, is the communication transmission energy consumption, is the propulsion energy consumption; ω mt is the energy consumption weight factor of the mobile terminal, ω uav is the energy consumption weight factor of the UAV, and satisfies ω mt +ω uav = 1; η is the flight energy consumption coefficient of the UAV, is the effective switching capacitance coefficient of the mobile terminal, is the effective switching capacitance coefficient of the UAV, is the CPU frequency of the mobile terminal MT m at the nth time slot, is the CPU frequency of the UAV UAV u at the nth time slot, and θ is the length of each time slot; Step S3 jointly optimizes the transmission power p of the task, the computing resource f on the mobile terminal m , the computing resource f on the UAV u , and the flight trajectory q of the UAV u to minimize the system weighted energy consumption within the preset time T; the mathematical model is as follows: The objective function is for Problem P1: The constraint conditions are: Among them, constraint condition C1 represents the limitation of the uplink transmission power of the task; C2 - C3 are the CPU cycle frequency limitations of the UAV and mobile terminals; C4 - C6 represent the causal relationship between the task communication transmission time and offloading computing; C7 - C10 represent the flight speed, position coordinates, and battery capacity limitations of the UAV; Step S5 includes: S51. Under the given UAV trajectory, simplify the optimization problem of transmission power and computing resource allocation to: The objective function is Problem P1.1: The objective constraint conditions are: C1 - C6 Transform problem P1.1 into a convex function for solution; S52. Under the given power and computing resources, the problem of optimizing the UAV trajectory is: The objective function is Problem P2: The objective constraint conditions are: C7 - C10 Introduce a relaxation variable f[n], that is: Substitute the relaxation variable into problem P2 to obtain: The objective function is Problem P2.1: The constraints are as follows: Using the SCA method, the right side of the constraint condition C11 is expressed as its first-order Taylor expansion as follows: where f[n] (l) and q u [n] (l) are the l-th iteration values of f[n] and q u [n], respectively; The optimization problem P2.1 is reformulated as The objective function is Problem P2.2: The objective constraint conditions are: Solve problem P2.2 using a convex optimization tool to obtain an approximate optimal solution of the overall system algorithm.

2. The energy-saving method according to claim 1, characterized in that, the communication channel model is a probability-averaged communication channel model for line-of-sight and non-line-of-sight links.

3. The energy-saving method according to claim 1, characterized in that, the LOS and NLOS probabilities between the UAV and the mobile terminal calculated by the communication channel model are respectively: wherein, parameters a and b depend on the constant values of the environment; ∈ m,u [n] is the elevation angle between MTm and UAVu, i.e.: where d is the horizontal distance between the mobile terminal and the UAV, and h is the flight height of the UAV.

4. The energy-saving method according to claim 3, It is characterized in that In the UAV-assisted air-to-ground channel, the path losses of LOS and NLOS between the mobile terminal and the UAV are expressed as: where f c is the carrier frequency, η LoS and η NLoS are additional path losses, and c is the speed of light.

5. An energy-saving system for a NOMA-based UAV-assisted MEC system It is characterized in that It is implemented by the energy-saving method described in any one of claims 1-4. The energy-saving system includes the following modules: A system construction module, which is used to construct a NOMA-based UAV-assisted air-ground collaborative MEC network system, including m mobile terminals MT and u UAVs carrying MEC servers; A total energy consumption calculation module, which is used to propose a joint optimization calculation resource allocation, transmission power, and UAV trajectory scheme to minimize the system weighted energy consumption on the premise of a given calculation task and task deadline; It is also used to construct a communication channel model, a calculation model, and an energy consumption model. The communication channel model is used to calculate path loss and channel gain, and the calculation model is used to calculate the CPU frequencies of each mobile terminal and UAV in each time slot; based on the calculation results of the calculation model, the energy consumption model further calculates the energy consumption, including local calculation energy consumption, edge calculation energy consumption, communication transmission energy consumption, and flight propulsion energy consumption; Express the total energy consumption of the system within a preset time as a weighted sum of communication energy consumption, calculation energy consumption, and propulsion energy consumption; A modeling module, which models the total energy consumption as a formulated optimization problem through a mathematical model; A transformation module, which is used to decouple the optimization problem into two easy-to-handle sub-problems, namely, the sub-problem of transmission power and calculation resource allocation, and the sub-problem of UAV trajectory scheduling; An iteration module, which is used to propose an iterative optimization algorithm to alternately solve each sub-problem until the algorithm converges; During the iteration process, first, under the condition of the given UAV trajectory, perform equivalent transformation on the objective function and constraint conditions, and use the method of successive convex approximation to solve the allocation of transmission power and calculation resources; then, under the given calculation resources and allocated power, introduce a relaxation variable and the method of first-order Taylor series approximation to solve the optimal trajectory planning of the UAV.

Images