Method for maximizing the energy efficiency of backscatter RIS-assisted UAV-enabled MEC

Abstract

The present invention discloses a method for maximizing the energy efficiency of backscatter RIS-assisted UAV-enabled MEC, which mainly solves the problems of severe channel fading and limited energy of Internet of Things (IoT) devices in the prior art. The method includes: 1) building a MEC network system model based on backscatter RIS-assisted UAV-enabled; 2) adopting a time division multiple access protocol, where IoT devices obtain energy from the radio frequency broadcast signal of the UAV with the help of RIS and offload the parts of tasks that they cannot complete to the MEC server on the ground for calculation through backscattering; 3) finding the optimal backscatter reflection coefficient, the CPU frequency of local calculation of IoT devices, the duration of each sub-slot, the RIS phase shift, and the UAV trajectory by maximizing the system energy efficiency; 4) setting the system working parameters according to the optimization results to achieve optimization. The present invention can effectively improve the system throughput of the MEC network and significantly reduce the communication and computing energy consumption of IoT devices.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communication, and further relates to edge computing technology. Specifically, it is a method for maximizing the energy efficiency of mobile edge computing (MEC) enabled by backscatter reconfigurable intelligent surface (RIS) assisted unmanned aerial vehicle (UAV), which can be used for low-power communication in MEC systems and offloading tasks of Internet of Things (IoT) devices in scenarios where communication links are blocked. Background Art

[0002] MEC can alleviate the challenge of limited computing capabilities of devices by transferring heavy computing tasks from IoT devices to the edge. In addition, due to strict cost and size limitations, the battery capacity of IoT devices is limited. To mitigate the adverse effects of limited energy supply, backscatter communication has attracted extensive attention.

[0003] Backscatter communication can effectively reduce the energy consumption of IoT devices by reflecting external environmental radio frequency signals and can be powered by energy harvesting. Y. Ye, L. Shi, X. Chu et al. proposed an energy efficiency maximization scheme in a backscatter-assisted MEC network by jointly optimizing time allocation, communication resources, and local computing frequency in their published paper "Resource Allocation in Backscatter-Assisted Wireless Powered MEC Networks With Limited MEC Computation Capacity" (IEEE Transactions on Wireless Communications, 2022: 10678-10694). L. Shi, Y. Ye, X. Chu et al. considered the throughput maximization problem in a backscatter-assisted MEC network and jointly optimized communication and computing resources, backscatter reflection coefficient, and time allocation in their published paper "Computation Bits Maximization in a Backscatter Assisted Wirelessly Powered MEC Network" (IEEE Communications Letters, 2021: 528-532). The above-mentioned existing technical solutions all consider a fixed radio frequency energy source. However, the fixed radio frequency energy source and IoT devices are often blocked by ground obstacles, resulting in signal blockage and shadowing. Therefore, it is difficult to establish a reliable energy transmission link. In addition, the IoT devices and the base station are also blocked by ground obstacles, resulting in the inability of the IoT devices to establish effective ground wireless communication with the ground base station. Summary of the Invention

[0004] The object of the present invention is to address the deficiencies of the above-mentioned existing technologies. By leveraging the advantages of drones, backscattering, and RIS technologies, for the first time, an MEC energy efficiency maximization method based on backscattering RIS-assisted drone-enabled MEC is proposed by integrating MEC, backscattering communication, drone, and RIS technologies. By finding the optimal backscattering reflection coefficient, computing resources, time allocation, RIS phase shift, and drone trajectory, the system energy efficiency is maximized, thereby solving the problems of severe channel fading and limited energy of Internet of Things (IoT) devices in traditional MEC, effectively improving the system throughput of the MEC network, and significantly reducing the communication and computing energy consumption of IoT devices.

[0005] The basic idea for implementing the present invention is as follows: In the MEC scenario with backscattering RIS-assisted drone-enabled MEC, IoT devices adopt a time division multiple access protocol, obtain energy from the radio frequency broadcast signal of the drone with the help of RIS, and further partially offload the input data of tasks that they cannot complete themselves to the MEC server on the ground for calculation through backscattering, thereby saving their own energy consumption and improving computing efficiency.

[0006] To achieve the above object, the technical solution of the present invention includes the following steps:

[0007] (1) Build an edge computing network model:

[0008] Build an edge computing network model with a single-antenna base station equipped with an MEC server, a single-antenna rotary-wing drone, an RIS with M reflecting elements, and K single-antenna IoT devices; let and respectively represent the sets of IoT devices and RIS reflecting elements, where k and m represent the k-th IoT device and the m-th RIS reflecting element respectively;

[0009] (2) Divide the time slot structure of users and drones:

[0010] Discretize the limited task completion time T into N equal time slots, and let represent the set of N time slots, where n represents the n-th time slot; the duration of each time slot is τ = T / N, assuming that the position of the drone remains unchanged during each time slot; IoT devices adopt a time division multiple access protocol and divide each time slot into K' sub-time slots, where K' = K, and the duration of each sub-time slot is determined by the time allocation decision variable and satisfies

[0011] (3) Obtain the optimal optimization variables of the MEC network model with backscattering RIS-assisted drone-enabled MEC:

[0012] (3.1) Using the three-dimensional Euclidean coordinate system, assuming that the base station and all IoT devices are fixed on the ground at zero altitude, the horizontal positions of the base station and IoT device k are w A =(x A ,y A ) and w k =(x k ,y k ); Assuming that RIS is installed on the building facade, the horizontal position and height of RIS are w R =(x R ,y R ) and h R ; Let θ m [n] represents the phase shift of RIS reflection element m in the nth time slot, and the diagonal reflection coefficient matrix of RIS in the nth time slot is obtained as Assume that the UAV flies at a fixed altitude H within the mission completion time T, and the flight start and end points are set to q I and q F Based on the discrete path planning method, the horizontal position q[n] of the UAV in the nth time slot is obtained, where q[0] = q I , q[N]=q F Assuming that the links between the UAV and RIS, RIS and IoT devices, and RIS and base stations are dominated by line-of-sight links, the channel gains between the UAV and RIS, RIS and IoT device k, IoT device k and RIS, and RIS and base station in time slot n are h U,R [n]、h R,k 、g k,R and g R,A Assuming that the wireless channels between the drone and the IoT device and the IoT device and the base station are blocked and obey the Rayleigh fading channel model, the channel gains between the drone and the IoT device and the IoT device and the base station in time slot n are respectively obtained as h U,k [n] and g R,A , we further obtain the equivalent channel gain from the UAV to IoT device k and IoT device k to the base station as h k [n] and g k [n];

[0013] (3.2) Assume that each IoT device has a certain computing task to be performed, and the computing task is composed of a binary Indicates that I k Indicates the size of the input task bit, C k Indicates the computing resources required to input 1 bit of data;

[0014] (3.3) The drone establishes a reliable RF signal connection with the IoT device by optimizing its flight trajectory. The IoT device divides the RF signal it receives from the drone into two parts, α k [n] and 1 - α k [n], where α k [n] is used for backscattering to the base station, and 1 - α k [n] is collected to support the circuit consumption. The total energy harvested by the k-th IoT device within the time period T is EH k ; Each IoT device adopts a partial offloading mode, offloading a part of the computing task to the base station for calculation through backscattering, and the other part is calculated locally at the IoT device. The number of task bits executed locally by the k-th IoT device within time T is obtained and the energy consumption The number of task bits offloaded by the k-th IoT device to the base station within the time slot n and the transmission energy consumption as well as the flight energy consumption of the drone

[0015] (3.4) Calculate the total energy consumption E of the IoT device k within the task completion time T according to the following formula k and the total energy consumption E of the drone within the task completion time T U :

[0016]

[0017] where P U is the transmission power of the drone;

[0018] (3.5) Construct an expression to maximize the system energy efficiency :

[0019]

[0020] Among them, the optimization variables are the reflection coefficient α of backscattering = {α k [n]}, the CPU frequency f of local calculation at the IoT device = {f k}, the duration of each sub-time slot The phase shift θ of the RIS = {θ m [n]} and the trajectory Q of the drone = {q[n]}; ω U represents the energy consumption weight of the drone;

[0021] It is set that the IoT device and the drone satisfy the following constraint conditions:

[0022] represents the task completion constraint; represents the energy causality constraint, where Denote the initial energy of the k-th Internet of Things device; 0 ≤ α k [n] ≤ 1 represents the reflection coefficient constraint of backscattering; 0 ≤ f k ≤ f k,max represents the local computing CPU frequency constraint of the Internet of Things device, where f k,max represents the maximum available CPU frequency of the k-th Internet of Things device; represents the duration constraint of each sub-slot; |θ m [n]| 2 = 1 represents the phase shift constraint of the RIS; q[0] = q I , q[N] = q F represents the starting and ending position constraints of the drone; ||q[n] - q[n - 1]|| ≤ τV max represents the maximum speed constraint of the drone;

[0023] (3.6) Obtain the optimal system energy efficiency through the Dinkelbach algorithm, alternating optimization algorithm, Majorization Minimization algorithm, semidefinite relaxation algorithm, and successive convex approximation algorithm and its corresponding optimal optimization variables;

[0024] (4) Set the system operating parameters according to the optimal optimization variables corresponding to the optimal system energy efficiency and make the system run under these parameters to achieve system energy efficiency optimization.

[0025] The following are the advantages of the present invention compared with the prior art:

[0026] First, the present invention uses a drone as a mobile energy source to establish a reliable energy transmission connection with the Internet of Things device, thereby ensuring the stable operation of the system under high load or emergencies, and the drone can automatically adjust the flight path and charging time according to the needs of the system in the later stage, so as to maximize the energy utilization rate and the efficiency of the system.

[0027] Second, since the backscattering-assisted MEC of the present invention uses the radio frequency signal of the drone to collect energy and perform task offloading without the need to send additional signals, it can effectively reduce the communication energy consumption of the Internet of Things device. At the same time, it can achieve a higher data transmission rate and a longer transmission distance, thus improving the communication efficiency.

[0028] Third, the present invention uses the RIS to increase the channel capacity and coverage area, thereby enhancing the quality and stability of the energy transmission link between the drone and the Internet of Things device, as well as the link between the Internet of Things device and the base station, further improving the system throughput and reducing the system energy consumption. Description of the Drawings

[0029] Figure 1Schematic diagram of the application scenario of the method of the present invention;

[0030] Figure 2 Schematic diagram of the time slot structure of the present invention;

[0031] Figure 3 Flow chart for implementing the method of the present invention;

[0032] Figure 4 Schematic diagram of the algorithm convergence of different UAV transmission powers and the number of RIS reflection elements in the method of the present invention;

[0033] Figure 5 Comparison diagram of UAV trajectory maps obtained by using the phase random, trajectory fixed, and joint optimization methods in the method of the present invention;

[0034] Figure 6 Comparison diagram of UAV speeds obtained by using the phase random, trajectory fixed, and joint optimization methods in the method of the present invention;

[0035] Figure 7 Simulation result diagram of the influence of UAV weight on UAV energy consumption and IoT device energy consumption in the method of the present invention;

[0036] Figure 8 Simulation result diagram of the change of system energy efficiency with the task volume obtained by using the phase random, trajectory fixed, and joint optimization methods in the method of the present invention. Detailed implementation manners

[0037] The following refers to the accompanying drawings to describe in detail the implementation process of the technical solution of the present invention:

[0038] Example 1: Refer to Figure 3 , the present invention provides a method for maximizing the energy efficiency of MEC assisted by backscattering RIS enabled UAVs, and the specific implementation steps are as follows:

[0039] Step 1: Refer to Figure 1 , build an edge computing network model:

[0040] The edge computing network built by the present invention consists of a single-antenna base station equipped with an MEC server, a single-antenna rotary UAV, a RIS with M reflection elements, and K single-antenna IoT devices; let and respectively represent the sets of IoT devices and RIS reflection elements, where k and m represent the kth IoT device and the mth RIS reflection element respectively.

[0041] Step 2: Divide the time slot structure of users and UAVs:

[0042] Discretize the limited task completion time T into N equal time slots, and let Denote a set of \(N\) time slots, where \(n\) represents the \(n\)th time slot; the duration of each time slot is \(\tau = T / N\), assuming that the position of the UAV remains unchanged during each time slot; the IoT devices adopt a time division multiple access protocol and divide each time slot into \(K'\) sub - time slots, where \(K'=K\), and the duration of each sub - time slot is determined by the time allocation decision variable and satisfies

[0043] Step 3: Obtain the optimal optimization variables of the backscatter RIS - assisted UAV - enabled MEC network model:

[0044] (3.1) Adopt a three - dimensional Euclidean coordinate system. Assume that the positions of the base station and all IoT devices are fixed on the ground at zero altitude. The horizontal positions of the base station and IoT device \(k\) are obtained as \(\mathbf{w}\) A \(=(x\) A ,y\) A ) and \(\mathbf{w}\) k \(=(x\) k ,y\) k ); Assume that the RIS is installed on the outer facade of a building, and the horizontal position and height of the RIS are obtained as \(\mathbf{w}\) R \(=(x\) R ,y\) R ) and \(h\) R ; Let \(\theta\) m [n] represent the phase shift of the \(m\)th reflecting element of the RIS in the \(n\)th time slot. The diagonal reflection coefficient matrix of the RIS in the \(n\)th time slot is Assume that the UAV flies at a fixed height \(H\) within the mission completion time \(T\), and set the starting and ending points of the flight as \(\mathbf{q}\) I and \(\mathbf{q}\) F ; Based on the discrete path planning method, the horizontal position of the UAV in the \(n\)th time slot is obtained as \(\mathbf{q}[n]\), where \(\mathbf{q}[0]=\mathbf{q}\) I , \(\mathbf{q}[N]=\mathbf{q}\) F ; Assume that the line - of - sight links dominate between the UAV and the RIS, between the RIS and the IoT devices, and between the RIS and the base station. The channel gains between the UAV and the RIS, between the RIS and IoT device \(k\), between IoT device \(k\) and the RIS, and between the RIS and the base station in time slot \(n\) are obtained as \(h\) U,R [n], \(h\) R,k , \(g\) k,R and \(g\) R,A ; Assume that the wireless channels between the UAV and the IoT devices and between the IoT devices and the base station are subject to Rayleigh fading channel models. The channel gains between the UAV and the IoT devices and between the IoT devices and the base station in time slot \(n\) are obtained as \(h\) U,k [n] and \(g\) R,A, the equivalent channel gains from the UAV to the IoT device k and from the IoT device k to the base station are further obtained as h k [n] and g k [n];

[0045] Among them, the channel gains between the UAV and the RIS, the RIS and the IoT device k, the IoT device k and the RIS, and the RIS and the base station within the time slot n are h U,R [n], h R,k , g k,R and g R,A , specifically as follows:

[0046]

[0047] Among them, ρ is the channel power gain at 1m; represents the distance between the UAV and the RIS in the nth time slot; represents the distance between the RIS and the kth IoT device; represents the distance between the RIS and the base station; d represents the distance between the reflection elements; λ represents the carrier signal wavelength; represents the cosine of the angle of arrival of the signal from the UAV to the RIS in the nth time slot; represents the cosine of the deviation angle of the signal from the RIS to the kth IoT device; represents the cosine of the deviation angle of the signal from the RIS to the base station.

[0048] The obtained equivalent channel gains from the UAV to the IoT device k and from the IoT device k to the base station are h k [n] and g k [n], which is achieved as follows:

[0049] (3.1.1) Calculate the channel gains h U,k [n] and g R,A :

[0050]

[0051] Among them, represents the distance between the UAV and the kth IoT device in the nth time slot; represents the distance between the kth IoT device and the base station; γ1 and γ2 respectively represent the path loss exponents from the UAV to the IoT device and from the IoT device to the base station; and represent the random scattering components modeled by zero-mean and unit-variance circularly symmetric complex Gaussian random variables;

[0052] (3.1.2) The equivalent channel gains from the UAV to the IoT device k and from the IoT device k to the base station are obtained as h k [n] and g k [n] as follows:

[0053] h k [n] = h U,k [n] + h U,R [n] H Θ[n]h R,k ,

[0054]

[0055] (3.2) Assume that each IoT device has a definite computing task to be executed. The computing task is represented by a binary tuple where I k represents the size of the computing input task bits, and C k represents the computing resources required for inputting 1-bit data;

[0056] (3.3) The UAV establishes a reliable RF signal connection with the IoT device by optimizing the flight trajectory. The IoT device divides the RF signal received from the UAV into two parts, α k [n] and 1 - α k [n], where α k [n] is used for backscattering to the base station, and 1 - α k [n] is collected to support the circuit consumption. The total energy harvested by the k-th IoT device within the time period T is EH k ; Each IoT device adopts a partial offloading mode, offloading a part of the computing task to the base station for computing through backscattering, and computing the other part locally at the IoT device. The task bits locally computed by the k-th IoT device within time T and the energy consumption The task bits offloaded by the k-th IoT device to the base station within the time slot n and the transmission energy consumption as well as the flight energy consumption of the UAV are respectively calculated as follows:

[0057]

[0058] where P U is the transmission power of the UAV, η EH is the energy conversion efficiency of the energy harvesting circuit, κ k represents the chip capacitance coefficient of the IoT device k, B represents the bandwidth from the IoT device k to the base station, σ 2 represents the noise power at the base station, P BFor the constant circuit power consumption of backscattering, P(||v[n]||) represents the flight power consumption of the rotary-wing UAV, specifically as follows:

[0059]

[0060] Among them, P0 and P H are the profile power and induced power in the hover state respectively; U tip , v0, d0, g, s, and A are related to aerodynamics and represent the tip speed of the rotor blade, the average rotor induced speed during hovering, the fuselage drag ratio, the air density, the rotor solidity, and the rotor area respectively.

[0061] (3.4) Calculate the total energy consumption E of the Internet of Things device k within the task completion time T according to the following formula k and the total energy consumption E of the UAV within the task completion time T U :

[0062]

[0063] Among them, P U is the transmission power of the UAV;

[0064] (3.5) Construct an expression for maximizing the system energy efficiency :

[0065]

[0066] Among them, the optimization variables are the reflection coefficient α = {α k [n]} of backscattering, the CPU frequency f = {f k} of local computing of the Internet of Things device, the duration of each sub-slot the phase shift θ = {θ m [n]} of the RIS, and the trajectory Q = {q[n]} of the UAV; ω U represents the energy consumption weight of the UAV;

[0067] Set the Internet of Things device and the UAV to satisfy the following constraints:

[0068] represents the task completion constraint; represents the energy causality constraint, where represents the initial energy of the k-th Internet of Things device; 0 ≤ α k [n] ≤ 1 represents the reflection coefficient constraint of backscattering; 0 ≤ f k ≤ f k,max represents the CPU frequency constraint of local computing of the Internet of Things device, where f k,max represents the maximum available CPU frequency of the k-th Internet of Things device; Denote the duration constraint of each sub - time slot; |θ m [n]| 2 = 1 represents the phase - shift constraint of the RIS; q[0]=q I , q[N]=q F represents the starting and ending position constraints of the UAV; ||q[n] - q[n - 1]||≤τV max represents the maximum speed constraint of the UAV;

[0069] (3.6) Obtain the optimal system energy efficiency through the Dinkelbach algorithm, alternating optimization algorithm, Majorization Minimization algorithm, semidefinite relaxation algorithm, and successive convex approximation algorithm and its corresponding optimal optimization variables; specifically, maximize the system energy efficiency in this scenario First, solve the fractional programming problem through the Dinkelbach algorithm, and then decouple the highly complex non - convex optimization problem into a resource allocation problem, RIS phase - shift optimization problem, and UAV trajectory design problem through the alternating optimization algorithm, as follows:

[0070] (3.6.1) Introduce an auxiliary variable ξ>0, and use the Dinkelbach method to solve the fractional programming problem, that is, convert the expression constructed in (3.5) into

[0071] (3.6.2) Fix the RIS phase - shift and UAV trajectory, and convert the resource allocation problem into a convex problem using variable substitution, that is, convert the expression constructed in step (3.6.1) into a convex expression using variable substitution, and obtain the reflection coefficient α of backscattering, the CPU frequency f of local computing of IoT devices, and the duration t of each sub - time slot using standard convex optimization tools;

[0072] (3.6.3) Fix the UAV trajectory and α, f, t in (3.6.2), and obtain the RIS phase - shift θ through the Majorization Minimization algorithm and semidefinite relaxation algorithm;

[0073] (3.6.4) Fix α, f, t in (3.6.2) and θ in (3.6.3), and obtain the UAV trajectory Q using the successive convex approximation algorithm;

[0074] (3.6.5) Substitute the solved optimization variables α, f, t, θ, and Q into the expression constructed in (3.5) to obtain the system energy efficiency Repeat steps (3.6.1)-(3.6.5) until the algorithm converges, and solve to obtain the optimal system energy efficiency and the corresponding optimal optimization variables α * , f * , t * , θ* and Q * 。

[0075] Step 4: According to the optimal system energy efficiency Set the system operating parameters according to the corresponding optimal optimization variables, and make the system operate under these parameters to achieve system energy efficiency optimization.

[0076] Embodiment 2: The method for maximizing energy efficiency in the MEC scenario enabled by backscatter RIS-assisted UAVs provided in this embodiment has the same overall implementation steps as Embodiment 1, and is further described for the MEC scenario enabled by backscatter RIS-assisted UAVs:

[0077] Step a: Build an edge computing network consisting of a single-antenna base station equipped with an MEC server, a single-antenna rotary UAV, an RIS with M reflecting elements, and K single-antenna Internet of Things devices; let and represent the sets of the Internet of Things devices and the RIS reflecting elements respectively, where k and m represent the k-th Internet of Things device and the m-th RIS reflecting element respectively;

[0078] Step b: Discretize the finite task completion time T into N equal time slots, and let represent the set of N time slots, where n represents the n-th time slot. The duration of each time slot is τ = T / N, where τ is small enough so that the position of the UAV can be assumed to be unchanged during each time slot. To avoid interference between Internet of Things devices during the offloading process, the Internet of Things devices adopt a time division multiple access protocol, and each time slot is further divided into K' sub-time slots, where K' = K, and the duration of each sub-time slot is determined by the time allocation decision variable and satisfies

[0079] Step c: For the sake of convenience of explanation, a three-dimensional Euclidean coordinate system is adopted, and its coordinates are measured in meters. Assume that the positions of the base station and all Internet of Things devices are fixed on the ground at zero altitude, and the horizontal positions of the base station and Internet of Things device k are obtained as w A =(x A , y A ) and w k =(x k , y k ) respectively; assume that the RIS is installed on the building facade, and the horizontal position and height of the RIS are obtained as w R =(x R , y R ) and h R respectively; let θ m [n] represent the phase shift of the m-th RIS reflecting element in the n-th time slot, and the diagonal reflection coefficient matrix of the RIS in the n-th time slot is obtained as Assume that the UAV flies at a fixed altitude H within the mission completion time T, and sets the flight starting point and ending point as q I =(x I ,y I ) and q F =(x F ,y F ); Based on the discrete path planning method, the horizontal position q[n] of the UAV at the nth time slot is obtained, where q[0]=q I , q[N]=q F ; Assume that the line-of-sight links dominate between the UAV and the RIS, between the RIS and the IoT device, and between the RIS and the base station. The channel gains h U,R [n], h R,k , g k,R and g R,A at the nth time slot are obtained as follows:

[0080]

[0081] where ρ is the channel power gain at 1m; represents the distance between the UAV and the RIS at the nth time slot; represents the distance between the RIS and the kth IoT device; represents the distance between the RIS and the base station; d represents the distance between the reflecting elements; λ represents the carrier signal wavelength; represents the cosine of the signal arrival angle from the UAV to the RIS at the nth time slot; represents the cosine of the signal deviation angle from the RIS to the kth IoT device; represents the cosine of the signal deviation angle from the RIS to the base station.

[0082] Assume that the wireless channels between the UAV and the IoT device and between the IoT device and the base station are subject to the Rayleigh fading channel model. The channel gains h U,k [n] and g R,A at the nth time slot are obtained as follows:

[0083]

[0084] where, represents the distance between the UAV and the kth IoT device at the nth time slot; represents the distance between the kth IoT device and the base station; γ1 and γ2 respectively represent the path loss exponents from the UAV to the IoT device and from the IoT device to the base station; and represents a random scattering component modeled by a circularly symmetric complex Gaussian random variable with zero mean and unit variance.

[0085] Therefore, the equivalent channel gains from the UAV to the IoT device k and from the IoT device k to the base station can be obtained as h k [n] = h U,k [n] + h U,R [n] H Θ[n]h R,k and

[0086] Step d: Assume that each IoT device has a definite computing task to be executed, and the computing task is represented by a binary tuple where I k represents the size of the computing input task bits, and C k represents the computing resources required for inputting 1-bit data;

[0087] Step e: The UAV establishes a reliable RF signal connection with the IoT device by optimizing the flight trajectory. The RF signal received by the IoT device from the UAV is divided into two parts. The α k [n] part is used for backscattering to the base station, and the (1 - α k [n]) part is collected to support its circuit consumption, and the total energy EH harvested by the k-th IoT device within the time period T is obtained as k as follows:

[0088]

[0089] where P U is the transmission power of the UAV, and η EH is the energy conversion efficiency of the energy harvesting circuit.

[0090] Each IoT device adopts a partial offloading mode, offloading a part of the computing task to the base station for computing through backscattering, and the other part is computed locally at the IoT device. Let and respectively represent the task bits and energy consumption executed by the local computing of the k-th IoT device within the time slot n, and the specific expressions are as follows:

[0091]

[0092] where κ k represents the chip capacitance coefficient of the IoT device k.

[0093] Let and respectively represent the task bits and energy consumption offloaded by the k-th IoT device to the base station within the time slot n, and the specific expressions are as follows:

[0094]

[0095] Among them, B represents the bandwidth from the Internet of Things device k to the base station, and σ 2 represents the noise power at the base station, and P B is the constant circuit power consumption of backscattering. It is assumed that the circuit power consumption of all K Internet of Things devices is the same.

[0096] Let represent the flight energy consumption of the drone, and the specific expression is as follows:

[0097]

[0098] In the formula, P(||v[n]||) represents the flight power consumption of the rotary-wing drone, specifically as follows:

[0099]

[0100] Among them, P0 and P H are the blade profile power and induced power in the hover state respectively, and U tip , v0, d0, g, s, and A are related to aerodynamics, and represent the tip speed of the rotor blade, the average rotor induced speed during hovering, the fuselage drag ratio, the air density, the rotor solidity, and the rotor area respectively.

[0101] Step f: Calculate the total energy consumption Ek of the Internet of Things device k within the task completion time T according to the following formula k and the total energy consumption Eu of the drone within the task completion time T U :

[0102]

[0103] Step g: Establish a problem of maximizing the energy efficiency of MEC enabled by backscattering RIS-assisted drones, and the expression is as follows:

[0104]

[0105] Among them, the optimization variables are the reflection coefficient α = {α k [n]} of backscattering, the CPU frequency f = {f k} of local computing of Internet of Things devices, the duration of each sub-slot, the phase shift θ = {θ m [n]} of RIS, and the trajectory Q = {q[n]} of the drone; ω U represents the energy consumption weight of the drone;

[0106] It is set that the Internet of Things devices and drones in this scenario satisfy the following constraint conditions:

[0107] Represents the task completion constraint; Represents the energy causality constraint, where represents the initial energy of the k-th Internet of Things device; 0 ≤ α k [n] ≤ 1 represents the reflection coefficient constraint of backscattering; 0 ≤ f k ≤ f k,max represents the local computing CPU frequency constraint of the Internet of Things device, where f k,max represents the maximum available CPU frequency of the k-th Internet of Things device; represents the duration constraint of each sub-slot; |θ m [n]| 2 = 1 represents the phase shift constraint of the RIS; q[0] = q I , q[N] = q F represents the starting and ending position constraints of the UAV; ||q[n] - q[n - 1]|| ≤ τV max represents the maximum speed constraint of the UAV;

[0108] Step h: Use the Dinkelbach algorithm, alternating optimization algorithm, Majorization Minimization algorithm, semidefinite relaxation algorithm, and successive convex approximation algorithm to solve formula <1.1> to obtain the optimal parameters.

[0109] (8.1) Use the Dinkelbach algorithm to solve the fractional programming problem of the objective function. Introduce an auxiliary variable ξ > 0 and transform formula <1.1> into:

[0110]

[0111] Therefore, if and only if can the solutions α, f, t, θ, and Q of formula <1.1> be obtained, where ξ * is the optimal system energy efficiency of formula <1.1> For a given ξ, formula <1.2> is still non-convex. Therefore, a three-step alternating optimization algorithm is proposed to solve formula <1.2>, and the specific algorithm is as follows.

[0112] (8.2) In the first step, focus on solving the optimization variables α, f, t with the RIS phase shift θ and the UAV trajectory Q.

[0113] (8.2.1) Given the RIS phase shift θ and the UAV trajectory Q. Then, the resource allocation problem of the optimization variables α, f, and t can be rewritten as:

[0114]

[0115] where the constraints are 0 ≤ αk [n] ≤ 1; 0 ≤ f k ≤ f k,max ;

[0116]

[0117] (8.2.2) Since there is a non - linear coupling between the variables α and t in formula <1.3>, formula <1.3> is a complex non - convex expression. To solve it, an auxiliary variable β k [n] is introduced such that β k [n] = t k b [n]α k [n], we can obtain

[0118]

[0119] Let Obviously is the perspective function of Ψ k [n]. At this time, Ψ k [n] is a concave function, then is also a concave function. Thus, formula <1.3> is transformed into:

[0120]

[0121] where β = {β U,k,m [n]}; the constraint becomes The constraint becomes The constraint 0 ≤ α k [n] ≤ 1 becomes

[0122] It can be seen that formula <1.4> is a standard convex expression, and the backscattering reflection coefficient α, the CPU frequency f of the local computing of the Internet of Things device, and the duration t of each sub - time slot can be obtained by using the interior - point method.

[0123] (8.2) In the second step, focus on solving the RIS phase shift θ with the UAV trajectory Q and the optimization variables α, f, t.

[0124] (8.2.1) By fixing the previously optimized backscattering reflection coefficient, the CPU frequency of the local computing of the Internet of Things device, and the duration of each sub - time slot. Then, the RIS phase - shift optimization problem can be expressed as follows:

[0125]

[0126] where the constraint is |θ m [n]|2 = 1;

[0127] (8.2.2) Due to the complex coupling of variables in formula <1.5> and the constraint there is a non-convexity in formula <1.5>. First, rewrite |h k [n]| 2 and |g k [n]| 2 as follows:

[0128]

[0129] where, and In addition, represent the matrix forms of formula <1.6> and formula <1.7> as:

[0130]

[0131] where, and

[0132]

[0133] Let then we have

[0134] |h k [n]| 2 = Tr(H k [n]Φ[n]) + |h U,k [n]| 2

[0135] |g k [n]| 2 = Tr(G k Φ[n]) + |g k,A | 2

[0136] (8.2.3) Since the quadratic terms |h k [n]| 2 |g k [n]| 2 appear in formula <1.5> and the constraint Therefore, rewrite |h k [n]| 2 |g k [n]| 2 as follows:

[0137]

[0138] According to the Majorization Minimization algorithm, taking the first-order Taylor expansion of Equation <1.8>, the lower bound of Equation <1.8> is obtained as follows:

[0139]

[0140] where,

[0141]

[0142] The matrix form of Equation <1.9> is expressed as:

[0143]

[0144] where, and

[0145]

[0146] Let then there is

[0147]

[0148] (8.2.4) Since it satisfies a rank-one constraint is introduced However, due to the non-convexity of the rank-one constraint, the semidefinite relaxation algorithm is used to relax the rank-one constraint Then Equation <1.5> can be rewritten as:

[0149]

[0150] where the constraint becomes The constraint becomes |θ m [n]| 2 = 1 becomes In the formula, is a matrix with additional zero rows and zero columns;

[0151] Obviously, Equation <1.5> is a standard convex semidefinite programming expression and can be solved using standard convex optimization tools. By solving Equation <1.5> to iteratively update until convergence. During the iteration process, if has a rank of 1, the RIS phase shift θ can be obtained through the principal eigenvector. If does not have a rank of 1, then Gaussian randomization can be used to make have a rank of 1, and then the RIS phase shift θ can be obtained through the principal eigenvector.

[0152] (8.3) Step 2: Focus on solving the UAV trajectory $\mathbf{Q}$ with the optimized variables $\alpha$, $f$, $t$, and the RIS phase shift $\boldsymbol{\theta}$.

[0153] (8.3.1) By fixing the reflection coefficient of the previously optimized backscattering, the CPU frequency of the local computing of the IoT device, the duration of each sub-slot, and the RIS phase shift. Then, the UAV trajectory optimization problem can be formulated as follows:

[0154]

[0155] where the constraints are $\mathbf{q}[0]=\mathbf{q}_0$ I , $\mathbf{q}[N]=\mathbf{q}_N$ F ; $\|\mathbf{q}[n] - \mathbf{q}[n - 1]\| \leq \tau V$ max ;

[0156] (8.3.2) Obviously, Equation <1.5> is a non-convex expression. Therefore, the continuous convex approximation algorithm is used to optimize the non-convex terms in Equation <1.5> to obtain its approximate expression. First, for the non-convex term $P(\|\mathbf{v}[n]\|)$ in the objective function, slack variables $\mathbf{v}_1[n]$ and $\mathbf{v}_2[n]$ are introduced, satisfying:

[0157] $\mathbf{v}_1[n] \geq \|\mathbf{v}[n]\|$

[0158]

[0159] Thus, we can obtain:

[0160]

[0161] Using the continuous convex approximation algorithm, given any feasible solutions of $\mathbf{v}_1[n]$ and $\mathbf{v}_2[n]$ and Equation <1.11> is approximated by the following convex constraints:

[0162]

[0163] where,

[0164]

[0165] Thus, the non-convex term $P(\|\mathbf{v}[n]\|)$ is replaced by the following convex approximation

[0166]

[0167] (8.3.3) It can be seen from that $h$ U,R [n] is relative to

[0168] Since the UAV trajectory q[n] is complex and non-linear because d U,R [n] and are both related to q[n]. To solve this problem, we use q[n] from the j-th iteration in the next iteration to obtain an approximate

[0169] Let Then we have

[0170]

[0171] where

[0172]

[0173] To address the non-convexity of the problem, we introduce slack variables w k [n] and r[n], let d U,k [n] ≤ w k [n] and d U,R [n] ≤ r[n], and using the successive convex approximation algorithm, given any feasible solution of w k [n] and r[n] and r (j) [n], we can obtain:

[0174]

[0175] In addition, Γ k [n] is approximated by its lower bound as follows:

[0176] [[ID=5i]]

[0177] where

[0178]

[0179]

[0180] Similarly, |h k [n]| 2 can be equivalently expressed as:

[0181]

[0182] where

[0183]

[0184] Using the successive convex approximation algorithm, |h k [n]| 2 is approximated by its lower bound as follows:

[0185]

[0186] Among them,

[0187]

[0188] (8.3.4) Based on the successive convex approximation algorithm, a suboptimal solution to the original problem can be obtained by solving the following convex approximation expression:

[0189]

[0190] where the constraint becomes constraint becomes q[0] = q I , q[N] = q F ; ||q[n] - q[n - 1]|| ≤ τV max ; v1[n] ≥ ||v[n]||;

[0191] Equation <1.12> is a standard convex expression. However, the positions of the UAVs in different time slots are coupled with each other, so it is difficult to obtain a closed-form solution for q[n]. In this case, resort to standard convex optimization tools to solve Equation <1.7> to obtain the UAV trajectory Q.

[0192] (8.4) Substitute the solved optimization variables α, f, t, θ, and Q into Equation <1.1> to obtain the system energy efficiency Repeat steps (8.1)-(8.4) until the algorithm converges, and solve to obtain the optimal system energy efficiency and the corresponding optimal optimization variables α * , f * , t * , θ * and Q * .

[0193] Step i: The system selects operating parameters according to the optimal optimization variables to make the system performance optimal.

[0194] The present invention solves the problems of severe channel fading and limited energy of Internet of Things devices in traditional mobile edge computing networks. First, the MEC and backscattering technologies are used to effectively improve the system throughput while significantly reducing the communication and computing energy consumption of Internet of Things devices. Second, the advantages of easy deployment, flexible movement, and line-of-sight link connection of unmanned aerial vehicles are utilized to establish a reliable energy transmission link with ground Internet of Things devices. Finally, the RIS is used to reconfigure the wireless communication environment to provide a higher passive beamforming gain.

[0195] The effects of the present invention are further described below in combination with simulation experiments:

[0196] A. Simulation conditions

[0197] Computer simulation software is used for simulation. The simulation considers a geographical area of 40×50 square meters on the ground, where K = 6 Internet of Things devices, a base station equipped with an MEC server, and an RIS with M = 10 reflecting elements are deployed. The position of the base station is set to (0, 30), the position of the RIS is set to (15, 40), the starting and ending points of the unmanned aerial vehicle flight are set to (0, 0) and (40, 50) respectively, the height of the RIS is set to h R = 10m, the height of the unmanned aerial vehicle is set to H = 20m, and the maximum flight speed of the unmanned aerial vehicle is set to V max = 10m / s. Table 1 lists the simulation parameters used in the power consumption model of the rotary-wing unmanned aerial vehicle. Unless otherwise specified, other parameter settings are as follows: task completion time T = 15s, number of time slots N = 15, the task volume I of each Internet of Things device k = 1 / kbit, the initial energy of each Internet of Things device The constant circuit power consumption P of Internet of Things device backscattering B = 10 -4 W, the transmission power P of the unmanned aerial vehicle U = 10 -4 W, the maximum CPU clock frequency f of Internet of Things devices k,max = 1MHz, the energy consumption weight ω of the unmanned aerial vehicle U = 0.02, the distance d between reflecting elements = λ / 2, the channel power gain ρ at a reference distance of 1m = -20dB, the path loss exponents from the unmanned aerial vehicle to Internet of Things devices and from Internet of Things devices to the base station are respectively set to γ1 = 2.2 and γ2 = 3, the bandwidth from Internet of Things devices to the base station is set to B = 2MHz, the noise at the base station is set to σ 2 = -70dBm, the computing resources required for Internet of Things devices to input 1 bit of data are set to C k = 1000, the chip capacitance coefficient of Internet of Things devices is set to κ k = 10 -27 .

[0198] Table 1 Simulation Parameters of Rotor UAV

[0199]

[0200] B. Simulation Content

[0201] Simulation 1: The influence of different UAV transmission powers and the number of RIS reflection array elements on the algorithm convergence. The simulation results are as Figure 4 shown;

[0202] Simulation 2: Obtain the UAV trajectory by using phase randomization, fixed trajectory, and joint optimization methods respectively. The simulation results are as Figure 5 shown;

[0203] Simulation 3: Obtain the UAV speed by using phase randomization, fixed trajectory, and joint optimization methods respectively. The simulation results are as Figure 6 shown;

[0204] Simulation 4: The influence of UAV weight on the energy consumption of UAV and IoT devices. The simulation results are as Figure 7 shown;

[0205] Simulation 5: The influence of the task volume on the system energy efficiency when using different methods of phase randomization, fixed trajectory, and joint optimization. The simulation results are as Figure 8 shown;

[0206] C. Simulation Results

[0207] It can be seen from Figure 4 that the method proposed in the present invention can converge within fewer iteration times, which proves the effectiveness of the proposed method. In addition, it can be observed from Figure 4 that the system energy efficiency increases with the increase of P U because the number of offloaded task bits is an increasing function of P U . It can also be seen from Figure 4 that the system energy efficiency increases with the increase of the number of RIS array elements. This is because the increase in the number of RIS array elements brings more degrees of freedom, which can provide higher passive beamforming gain.

[0208] It can be seen from Figure 5 that under the joint optimization method and phase randomization method proposed in the present invention, the UAV can use its mobility to get closer to the location of the IoT device. This is because the UAV flying closer to the IoT device can reduce the path loss, enabling the IoT device to reflect more computing tasks to the base station and collect more energy for local computing. Combining Figure 6 it can also be observed that the UAV first flies at the maximum speed, then decelerates, and even tends to hover at a fixed point, which can optimally enable the IoT device to perform task offloading and collect energy.

[0209] From Figure 7 It can be seen that as the task volume increases, the system energy efficiency of the phase randomization, trajectory fixed, and joint optimization methods proposed in the present invention first increases and then decreases. This is because the local computing energy consumption is an exponential function, which grows faster than the logarithmic function of the task bits unloaded by the Internet of Things devices to the base station. From Figure 7 it can also be observed that when the UAV trajectory is fixed, the system energy efficiency is the worst. This can be explained as when the UAV trajectory is fixed, the flight energy consumption is relatively high, resulting in a low system energy efficiency. At the same time, it shows the importance of optimizing the UAV trajectory.

[0210] From Figure 8 it can be seen that as the UAV energy consumption weight increases, the optimization objective function pays more attention to the UAV energy consumption. Therefore, the flight energy consumption of the UAV decreases as the UAV energy consumption weight increases. On the contrary, the communication and computing energy consumption of the Internet of Things devices increase as the UAV energy consumption weight increases.

[0211] The above simulation analysis proves the correctness and effectiveness of the method proposed in the present invention.

[0212] The parts not detailed in the present invention belong to the common general knowledge of those skilled in the art.

[0213] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Obviously, for those skilled in the art, after understanding the content and principle of the present invention, various modifications and changes in form and details may be made without departing from the principle and structure of the present invention. However, these modifications and changes based on the idea of the present invention are still within the protection scope of the claims of the present invention.

Claims

1. A method for maximizing the energy efficiency of backscatter RIS-assisted UAV-enabled MEC, characterized in that, Including the following steps: (1) Build an edge computing network model: An edge computing network model is constructed by a single-antenna base station equipped with an MEC server, a single-antenna rotary-wing UAV, a RIS with M reflecting elements, and K single-antenna Internet of Things devices; let and respectively represent the sets of Internet of Things devices and RIS reflecting elements, where k and m represent the k-th Internet of Things device and the m-th RIS reflecting element respectively; (2) Divide the time slot structure of users and drones: Discretize the limited task completion time T into N equal time slots, and let denote the set of N time slots, where n represents the nth time slot; the duration of each time slot is τ = T / N, and it is assumed that the position of the UAV remains unchanged during each time slot; the IoT device adopts a time division multiple access protocol and divides each time slot into K' sub-time slots, where K' = K, and the duration of each sub-time slot is determined by the time allocation decision variable and satisfies (3) Obtain the optimal optimization variables of the backscatter RIS-assisted UAV-enabled MEC network model: (3.1) Adopt a three-dimensional Euclidean coordinate system. Assume that the positions of the base station and all Internet of Things (IoT) devices are fixed on the ground at zero altitude. The horizontal positions of the base station and IoT device k are obtained as w A =(x A , y A ) and w k =(x k , y k ); Assume that the RIS is installed on the exterior facade of a building. The horizontal position and height of the RIS are obtained as w R =(x R , y R ) and h R ; Let θ m [n] represent the phase shift of the m-th reflecting element of the RIS in the n-th time slot. The diagonal reflection coefficient matrix of the RIS in the n-th time slot is Assume that the UAV flies at a fixed height H within the mission completion time T, and the starting and ending points of the flight are set as q I and q F respectively; Based on the discrete path planning method, the horizontal position q[n] of the UAV in the n-th time slot is obtained, where q[0]=q I , q[N]=q F ; Assume that the line-of-sight (LOS) links dominate between the UAV and the RIS, between the RIS and the IoT devices, and between the RIS and the base station. The channel gains between the UAV and the RIS, between the RIS and IoT device k, between IoT device k and the RIS, and between the RIS and the base station in the n-th time slot are obtained as h U,R [n], h R,k , g k,R and g R,A respectively; Assume that the wireless channels between the UAV and the IoT devices and between the IoT devices and the base station are subject to Rayleigh fading channel models. The channel gains between the UAV and the IoT devices and between the IoT devices and the base station in the n-th time slot are obtained as h U,k [n] and g k,A respectively. Furthermore, the equivalent channel gains from the UAV to IoT device k and from IoT device k to the base station are obtained as h k [n] and g k [n]; (3.2) Assume that each Internet of Things device has a definite computing task to be executed, and the computing task is represented by a binary tuple where I k represents the size of the computing input task bits, and C k represents the computing resources required to input 1-bit data; (3.3) The drone establishes a reliable RF signal connection with the IoT device by optimizing its flight trajectory. The IoT device divides the RF signal it receives from the drone into two parts, α k [n] and 1 - α k [n], where α k [n] is used for backscattering to the base station, and 1 - α k [n] is collected to support the circuit consumption. The total energy harvested by the k-th IoT device within the time period T is EH k ; Each IoT device adopts a partial offloading mode, offloading a part of the computing task to the base station for calculation through backscattering, and the other part is calculated locally by the IoT device. The number of task bits executed by the local calculation of the k-th IoT device within time T is obtained and the energy consumption The number of task bits offloaded by the k-th IoT device to the base station within the time slot n and the transmission energy consumption as well as the flight energy consumption of the drone (3.4) Calculate the total energy consumption \(E\) of the IoT device \(k\) within the task completion time \(T\) according to the following formula k and the total energy consumption \(E\) of the UAV within the task completion time \(T\) U : Where P U is the transmission power of the drone; (3.5)Construct a system to maximize energy efficiency The expression of: Among them, the optimization variables are the reflection coefficient α of backscattering = {α k [n]}, the CPU frequency f of local computing of the IoT device = {f k}, the duration of each sub-slot the phase shift θ of the RIS = {θ m [n]} and the trajectory Q of the UAV = {q[n]}; ω U represents the energy consumption weight of the UAV; Set the following constraints for the Internet of Things devices and drones: Represents the task completion constraint; Represents the energy causality constraint, where represents the initial energy of the k-th IoT device; 0 ≤ α k [n] ≤ 1 represents the reflection coefficient constraint of backscattering; 0 ≤ f k ≤ f k,max represents the local computing CPU frequency constraint of the IoT device, where f k,max represents the maximum available CPU frequency of the k-th IoT device; Represents the duration constraint of each sub-slot; |θ m [n]| 2 = 1 represents the phase shift constraint of the RIS; q[0] = q I , q[N] = q F represents the start and end position constraints of the UAV; ||q[n] - q[n - 1]|| ≤ τV max represents the maximum speed constraint of the UAV; (3.6) Obtain the optimal system energy efficiency through the Dinkelbach algorithm, the alternating optimization algorithm, the Majorization Minimization algorithm, the semidefinite relaxation algorithm, and the successive convex approximation algorithm and its corresponding optimal optimization variables; (4)According to the optimal system energy efficiency Set the system operating parameters corresponding to the optimal optimization variables, and make the system run under these parameters to achieve system energy efficiency optimization.

2. The method according to claim 1, characterized in that: In step (3.1), the channel gains between the UAV and the RIS, between the RIS and IoT device k, between IoT device k and the RIS, and between the RIS and the base station within time slot n are h U,R [n], h R,k , g k,R and g R,A , specifically as follows: where ρ is the channel power gain at 1 m; denotes the distance between the UAV and the RIS in the n-th time slot; denotes the distance between the RIS and the k-th IoT device; denotes the distance between the RIS and the base station; d denotes the distance between the reflection elements; λ denotes the carrier signal wavelength; denotes the cosine of the angle of arrival of the signal from the UAV to the RIS in the n-th time slot; denotes the cosine of the angle of departure of the signal from the RIS to the k-th IoT device; denotes the cosine of the angle of departure of the signal from the RIS to the base station.

3. The method according to claim 2, characterized in that: The equivalent channel gains from the UAV to the IoT device k and from the IoT device k to the base station obtained in step (3.1) are h k [n] and g k [n], and the implementation is as follows: (3.1.1) Calculate the channel gains h U,k [n] and g k,A : wherein, represents the distance between the UAV and the k-th Internet of Things device in the n-th time slot; represents the distance between the k-th Internet of Things device and the base station; γ1 and γ2 respectively represent the path loss exponents from the UAV to the Internet of Things device and from the Internet of Things device to the base station; and represent the random scattering components modeled by zero-mean and unit-variance circularly symmetric complex Gaussian random variables; (3.1.2) The equivalent channel gains from the UAV to the IoT device k and from the IoT device k to the base station are obtained as h k [n] and g k [n]: h k [n] = h U,k [n] + h U,R [n] H Θ[n]h R,k , 4. The method according to claim 1, characterized in that: The total energy EH harvested by the k-th Internet of Things device during the time period T in step (3.3) k , the number of task bits executed by the local computing of the k-th Internet of Things device within time T and the energy consumption The number of task bits unloaded by the k-th Internet of Things device to the base station within time slot n and the transmission energy consumption And the flight energy consumption of the drone Are calculated respectively according to the following: where P U is the transmission power of the UAV, η EH is the energy conversion efficiency of the energy harvesting circuit, κ k represents the chip capacitance coefficient of the Internet of Things device k, B represents the bandwidth from the Internet of Things device k to the base station, σ 2 represents the noise power at the base station, P B is the constant circuit power consumption of backscattering, and P(||v[n]||) represents the flight power consumption of the rotor UAV, which is specifically as follows: Among them, P0 and P H are the blade element power and induced power in the hover state, respectively; U tip , v0, d0, g, s, and A are related to aerodynamics and represent the rotor blade tip speed, the average rotor induced velocity in hover, the fuselage drag ratio, the air density, the rotor solidity, and the rotor area, respectively.

5. The method according to claim 1, characterized in that: In step (3.6), obtaining the optimal system energy efficiency through the Dinkelbach algorithm, the alternating optimization algorithm, the Majorization Minimization algorithm, the semidefinite relaxation algorithm, and the successive convex approximation algorithm and the corresponding optimal optimization variables is realized as follows: (3.6.1) Introduce an auxiliary variable ξ > 0 and use the Dinkelbach method to solve the fractional programming problem, that is, transform the expression constructed in (3.5) into (3.6.2) Fix the RIS phase shift and UAV trajectory, and use variable substitution to transform the resource allocation problem into a convex problem, that is, use variable substitution to transform the expression constructed in step (3.6.1) into a convex expression, and use standard convex optimization tools to obtain the reflection coefficient α of backscattering, the CPU frequency f of local computing of Internet of Things devices, and the duration t of each sub-time slot; (3.6.3) Fix the UAV trajectory and α, f, t in (3.6.2), and obtain the RIS phase shift θ through the Majorization Minimization algorithm and the semidefinite relaxation algorithm; (3.6.4) Fix α, f, t in (3.6.2) and θ in (3.6.3), and use the successive convex approximation algorithm to obtain the UAV trajectory Q; Substitute the optimized variables α, f, t, θ, and Q obtained from the solution into the expression constructed in (3.5) to obtain the system energy efficiency. Repeat steps (3.6.1)-(3.6.5) until the algorithm converges, and solve for the optimal system energy efficiency. And the corresponding optimal optimized variable α * , f * , t * , θ * and Q * .

Images