Calculation unloading and resource optimization method for crop yield estimation task

Abstract

The embodiment of the invention discloses a calculation unloading and resource optimization method for a crop yield pre-estimation task. The method comprises the following steps: firstly, deducing emission work and calculation frequency of a resource allocation strategy through a preset condition and a Lagrange multiplication; solving an unloading selection problem to obtain an unloading selection strategy parameter; solving the resource allocation problem to obtain the transmitting power and the calculation frequency of a resource allocation strategy; obtaining total energy consumption according to the unloading selection strategy parameter and the transmitting power and the calculation frequency of the resource allocation strategy; judging whether the total energy consumption converges or whether the round reaches the maximum number of times, and if not, repeating the steps; and if yes, outputting an unloading selection strategy parameter, the transmitting power and the calculation frequency of the resource allocation strategy, and a suboptimal solution of the total energy consumption. By adopting the technical scheme of the invention, the performance of satellite task calculation can be improved, the energy consumption is effectively reduced, and the economic benefit is improved.

Description

Technical Field

[0001] The present invention relates to the field of edge computing technology, and in particular to a computational offloading and resource optimization method for a crop yield estimation task. Background Art

[0002] With the development of Internet of Things (IoT) technology, more and more devices have been interconnected, forming a pattern of interconnection of all things. These devices generate massive amounts of data that need to be processed in a timely and effective manner. Many emerging application scenarios, such as autonomous driving, industrial automation, and telemedicine, have extremely high requirements for the real-time performance of data processing. These applications require fast responses and cannot rely on long-distance data transmission and processing in cloud computing centers. As a result, a series of space-to-ground networks have been derived for offloading calculations. In the process of offloading calculations, energy consumption, latency, etc. are important indicators for measuring economic benefits and user experience, and are the main optimization goals.

[0003] Every summer or autumn harvest, different regions will evaluate the yield of important crops in their regions. Traditional manual statistical methods may lead to fraud. Therefore, the present invention proposes to use satellite networks, where the region initiates a task request to the satellite network, and completes the calculation of the yield estimation task in the satellite network. As for how to complete the calculation of this task with high energy efficiency, the present invention uses the mutual cooperation between satellites in the computing network composed of air, space and land, with the goal of minimizing system energy consumption, to achieve high-efficiency calculation of the task.

[0004] The article "Energy-Efficient Design of Satellite-Terrestrial Computing in 6G Wireless Networks" describes a satellite-to-terrestrial network computing model that takes into account the interference between ground (including ground users, base stations, low-altitude drones and other equipment) and satellite communications, and optimizes the task computing energy consumption under the constraints of latency and computing resource limitations. The optimization problem is split into three sub-problems: offloading selection, beamforming, and resource allocation using an alternating optimization method. The objective function is converted into a convex function for solution. The optimal solution for offloading selection is obtained through a relaxation mapping method, and the solution for beamforming is obtained through a convex approximation technique.

[0005] There are obvious deficiencies in the currently proposed methods for optimizing energy consumption in air-ground networks.

[0006] 1) Balance between computing reuse and energy consumption optimization: In drone-assisted edge computing systems, it is necessary to balance computing reuse and energy consumption optimization, but existing technologies may find it difficult to minimize energy consumption while ensuring service quality.

[0007] 2) Limitations of deep reinforcement learning algorithms: Although deep reinforcement learning (DRL) algorithms are used to solve energy optimization problems, such algorithms may suffer from the curse of dimensionality and slow convergence, especially in dynamic unloading problems in high-dimensional state space and action space. Summary of the invention

[0008] In order to solve the above technical problems, the present invention proposes a computational offloading and resource optimization method for crop yield estimation tasks, which can improve the performance of satellite mission calculations, effectively reduce energy consumption, and improve economic benefits.

[0009] An embodiment of the present invention provides a method for offloading calculations and optimizing resources for a crop yield estimation task, comprising the following steps:

[0010] Step 101: Derivation of the transmit power p of the resource allocation strategy through preset conditions and Lagrange multiplication 0 and calculate the frequency f 0 ;

[0011] Step 102: Solve the uninstall selection problem to obtain the uninstall selection strategy parameter α i ,β i ,ξ i ;

[0012] Step 103: Solve the resource allocation problem to obtain the transmit power p of the resource allocation strategy. i and calculate the frequency f i ;

[0013] Step 104: Select strategy parameter α according to the uninstallation i ,β i ,ξ i and the transmission power p of the resource allocation strategy i and calculate the frequency f i , and obtain the total energy consumption E i ;

[0014] Step 105: determine whether the total energy consumption converges or whether the rounds reach the maximum number of times. If yes, go to step 106, otherwise repeat steps 102 to 104;

[0015] Step 106: Output the uninstall selection strategy parameter α * ,β * ,ξ * , the transmission power p of the resource allocation strategy * and calculate the frequency f * , and the suboptimal solution E for total energy consumption * .

[0016] Step 101 further includes the following steps:

[0017] Preset objective function:

[0018] Constraints:

[0019] Where: E n is the total energy consumption of BS n task computation, E u is the total energy consumption of UAV u task calculation, T n The total delay calculated for BS n tasks, The maximum delay calculated for BS n tasks, T u The total latency calculated for the UAV u mission, The maximum delay calculated for UAV u mission, α n,m is the offloading selection strategy of BS n, β u,m Select an offloading strategy for UAV u, ξ n is the local offloading policy of BS n, is the transmission power of BS n, is the maximum transmission power of BS n, is the transmission power of UAV u, is the maximum transmission power of UAV u, f n,m is the calculated frequency allocated by LEO m to BS n, f u,m The calculated frequency assigned by LEO m to UAV u, The maximum value of the calculation frequency that can be provided for LEO m,

[0020] All constraints are added to the objective function using the Lagrange multiplier method. The Lagrange multiplier for each constraint is expressed as l 1,n , l 2,u , l 3,m Represented by, l represents the set of Lagrange multipliers;

[0021] The Lagrange multiplication of the objective function is represented by Λ(α, β, ξ, p, f), abbreviated as Λ;

[0022] Use α to represent α n,m The set composed of β represents β u,m The set composed of n,m The set composed of n The set composed of and The set composed of f n,m and f u,m A collection of components;

[0023] The partial derivatives of the transmission power p, computing resources f and Lagrange multiplier l in the Lagrange multiplication of the objective function are obtained respectively, that is, Let these equations equal 0, that is, Λ p (α,β,p,ξ,l)=0,Λ f (α,β,f,l)=0,Λ l (α,β,p,f,ξ)=0;

[0024] The equation relationship between the transmit power p, the calculation frequency f, the unloading selection parameters α, β, ξ and the Lagrange multiplier l is obtained, which is denoted as p = Λ p (α, β, ξ, l) and f = Λ f (α,β,l);

[0025] Communication rate between base station and satellite in the communication model (R n,m is the rate of the channel between BS n and LEO m, B 1 is the bandwidth of the satellite-to-ground channel, g n,m represents the channel gain from ground base station n to LEO m, is the signal transmission power from BS n to LEO m, represents the additive white Gaussian noise variance), in the objective function R n,m In the denominator, The derivation is more complicated and needs to be transformed before it can be solved;

[0026] make The above formula is transformed into (In the formula is the channel rate between BS n and the local satellite, is the transmission power of BS n to the local satellite), and a new transmission power replacement parameter is set make Convert the base station upload energy consumption into (In the formula, C BS is the message size of the task request initiated by BS n);

[0027] Communication rate between UAV and satellite in the communication model (R u,m is the channel rate between UAV u and LEO m, g u,m =Δq u,m .O u,m (r u ).τ u,m (C) is the effective satellite channel gain, B 2 represents the UAV and LEO channel bandwidth, represents the additive white Gaussian noise variance, du,m is the distance between UAV u and satellite m, q u,m is the channel frequency from UAV to LEO satellite) and the inter-satellite communication rate (In the formula is the transmission power of the local satellite of BS n, λ is the wavelength, is the distance between the local satellite of BS n and LEO m, B 3 represents the bandwidth of intersatellite communication, represents the additive white Gaussian noise variance, and are the aperture gain, pointing loss factor, aperture diameter, pointing error angle and transmitter optical efficiency at the local satellite of BS n, respectively. are the aperture gain, pointing loss factor, aperture diameter, pointing error angle and transmitter optical efficiency of the receiver at LEO m respectively. They are also in the denominator of the objective function. and The derivative is equally complicated;

[0028] Use the same method with new parameters and Make a transformation;

[0029] use Indicates the transmit power replacement parameter and A collection of components;

[0030] From the computational model (In the formula is the computational energy consumption of BS n, κ m is the chip energy architecture coefficient of LEO m, Calculate the number of clock cycles required for the task of BS n, f n,m The calculated frequency allocated to BS n by LEO m) and (In the formula is the computational energy consumption of UAV u, Calculate the number of clock cycles required for the task of UAV u, f u,m = is the CPU calculation frequency allocated to UAV u by LEO m) From the two equations, we can see that the less computing resources the satellite allocates to the device (the smaller f), the lower the computing energy consumption; from the calculation delay equation, we can see that the less computing resources the satellite allocates to the device (the smaller f), the greater the calculation delay;

[0031] Depend on (In the formula The relationship between computing energy consumption and computing delay can be obtained by using the two equations:

[0032] pass (Where T n is the delay of the entire task calculation process of BS n, is the maximum delay of the entire task calculation process of BS n) and we can get α n,m ,ξ n , and f n,m The equation relationship between them is denoted as p = Ψ p (α,β,ξ,f);

[0033] pass (Where T u is the latency of the entire mission calculation process of UAV u, is the maximum delay of the entire mission calculation process of UAV u) and we can get β u,m , and f u,m The equation relationship is recorded as f = Ψ f (α,β,ξ,p);

[0034] Combining the above, we get p = Λ p (α,β,ξ,l),f=Λ f (α, β, l) and p = Ψ p (α,β,ξ,f),f=Ψ f (α, β, ξ, p) These equations give the value of the transmitted power p and the calculated frequency f, denoted as p 0 and f 0 .

[0035] Step 102 further includes the following steps:

[0036] The problems and constraints are as follows:

[0037]

[0038] α n,m ,β u,m ,ξ n ∈{0,1}

[0039]

[0040] In order to solve the offloading selection strategy when the transmission power p and the task calculation frequency f are known, the variable offloading selection parameters α, β, ξ are first relaxed to continuous variables, that is, the constraints α, β, ξ∈{0,1} become α, β, ξ∈[0,1];

[0041] Substitute the values ​​of the transmission power p and the task calculation frequency f into the original problem P to obtain the offloading selection problem P1;

[0042] Based on problem P1, construct the Lagrangian function Λ P1 (α, β, ξ, Γ, l), the original integer programming problem P1 is transformed into a convex optimization problem;

[0043] The total energy consumption of the entire task calculation process of the nth BS is (E n is the total energy consumption of the entire task calculation process of BS n, The size of the task request message forwarded by the local satellite of BS n and the size of the image data) contains the BS offloading selection parameter α n,m and the locally calculated parameter ξ n Two unknown quantities, and the multiplication of the two unknown quantities will make the trend of the function uncertain. The two unknown quantities need to be combined into a new variable Γ n,m , that is, Γ n,m =α n,m *ξ n , the constraints will also increase Let Γ represent Γ n,m A collection of components;

[0044] After the two unknown quantities are combined, due to the derived The formula only contains Γ n,m and α n,m Two unknown quantities, we can get Γ n,m and α n,m The relationship between Γ=Ψ Γ (α);

[0045] Since the derived T u =T u max The formula only contains the unloading selection parameter β of UAV u,m An unknown quantity can directly determine the value of the unloading selection parameter β of the UAV;

[0046] The partial derivatives of the unloading selection parameters α, β, and ξ in the Lagrangian multiplication of the objective function are obtained respectively, that is, Setting these equations equal to 0 gives Λ α (α,l)=0,Λ β (β,l)=0,Λ ξ (ξ,l)=0,Λ Γ (Γ,l)=0,Λ l (α, β, ξ, Γ) = 0, these equations are combined to obtain the values ​​of the BS offloading selection parameter α and the local calculation parameter ξ;

[0047] The values ​​of the unloading selection parameters α, β, ξ calculated in the above manner are between 0 and 1. Since the values ​​of the unloading selection parameters in the actual problem must be 0 or 1, it is necessary to map the continuous solution back to a 0-1 variable;

[0048] Set the offloading selection parameter α of each BS n,m The maximum value in is set to 1, and the others are set to 0;

[0049] The offloading selection parameter β of each UAV u,m The maximum value in is set to 1, and the others are set to 0;

[0050] For the local calculation parameter ξ of the BS n , set a judgment value, and calculate the parameter ξ locally n When it is greater than this value, n Set to 1, otherwise set to 0, set the judgment value to 0.5, satisfying ξ n >0.5, let ξ n =1, otherwise let ξ n =0.

[0051] Step 103 further includes the following steps:

[0052] The problems and constraints are as follows:

[0053]

[0054] Construct the Lagrangian function Λ P2 (p,f,l), replace the parameters with transmit power and Transform the transmission rate R in the Lagrange multiplication of the objective function respectively n,m , and R u,m ;

[0055] because The formula only contains and f n,m Three unknown quantities can be obtained, and the equation relationship between these three unknown quantities can be obtained, that is,

[0056] because The formula only contains and f u,m Two unknown quantities can be obtained by equation relationship, that is,

[0057] Substitute the transmit power obtained above for the parameter The two types of relations for calculating the frequency f are combined into

[0058] The Lagrange multiplication of the objective function replaces the parameters of the calculation frequency f and the transformed transmission power respectively and Find the partial derivative, that is Setting these expressions equal to 0 gives Λ f (f,l)=0,

[0059] Through Lianli Λ f (f,l)=0, These equations lead directly to the alternative parameter for transmit power: and calculate the frequency f, and then replace the parameter of the transmission power with Perform an inverse transformation to the transmit power p, and then obtain the transmit power p.

[0060] Step 104 further includes the following steps:

[0061] Substitute the offloading selection strategy parameters α, β, ξ, the transmission power p and the calculation frequency f into the objective function Find the value of the objective function, which is represented by E.

[0062] Step 105 and step 106 further include the following steps:

[0063] Using the unloading selection parameter α obtained in the i-th round of iteration i ,β i ,ξ i , transmission power p i and calculate the frequency f i Find the objective function The value of E i , the system energy consumption value obtained in the i-1th round is E i-1 ;

[0064] The absolute value difference of the two total energy consumption E obtained from two consecutive iterations is used to determine whether the results converge. That is, the absolute value difference between the result of the i-th iteration and the result of the i-1th iteration is expressed as |E i -E i-1 |, if the value is less than a value Δ, the total energy consumption E i Convergence, the value of Δ can be assigned as needed, and it is set to 1e-7, expressed as |E i -E i-1 |<Δ;

[0065] At the same time, set the maximum number of iterations i max , when i>i max and |E i -E i-1|<ΔWhen one of the two conditions is met, the iteration process ends and the suboptimal solution total energy consumption E is obtained. * , uninstall selection parameter α * ,β * ,ξ * , transmission power p * and computing resources * ;

[0066] If the total energy consumption E obtained in the first round of iterations does not converge and does not reach the maximum number of iterations i max , then repeat steps 102 to 104, and make a judgment again on the result obtained by the second round of iteration. If it is not satisfied, continue the third round of iteration until one of the two conditions is satisfied.

[0067] In the above technical solution, a new iterative optimization method is proposed to decompose the problem into two sub-problems, namely, the calculation offloading selection problem and the resource allocation problem, and obtain the suboptimal solution through iterative solution. Focus on inter-satellite forwarding communication. Considering that the base station is located on the ground, the number of satellites that can directly communicate in the same time slice is small, while the high-altitude UAV is at a high altitude, and the number of satellites that can communicate in the same time slice is more. Therefore, when comprehensively considering the optimal energy consumption, the tasks initiated by the base station are sent to the unloading target satellite for calculation through inter-satellite forwarding. Performance improvement and economic benefits. Through the proposed model and method, the performance of satellite mission calculation can be improved, energy consumption can be effectively reduced, and economic benefits can be improved.

[0068] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] The above and other purposes, features and advantages of the present invention will become more apparent by describing the embodiments of the present invention in more detail in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0070] Figure 1 It is a flowchart of computational offloading and resource optimization of the crop yield estimation task in an embodiment of the present invention. DETAILED DESCRIPTION

[0071] Below, the exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described here.

[0072] It should be noted that the relative arrangement of components and steps, the numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.

[0073] Those skilled in the art can understand that the terms "first" and "second" in the embodiments of the present invention are only used to distinguish different steps, devices or modules, etc., and neither represent any specific technical meaning nor indicate the necessary logical order between them.

[0074] It should also be understood that, in the embodiments of the present invention, “plurality” may refer to two or more than two, and “at least one” may refer to one, two or more than two.

[0075] It should also be understood that any component, data or structure mentioned in the embodiments of the present invention can generally be understood as one or more, unless explicitly limited or otherwise indicated in the context.

[0076] In edge computing, energy consumption is an important indicator for evaluating economic benefits and an important optimization goal. The lower the energy consumption, the higher the economic benefits. Latency is an important indicator that affects user experience and is one of the key constraints. The lower the latency, the better the user experience.

[0077] 1) Offloading selection: involves offloading computing tasks from mobile devices to edge servers or cloud servers to optimize resource usage, reduce latency and energy consumption, etc., taking into account the processing power, storage capacity and energy status of mobile devices. The technical solution of the present invention involves task offloading selection from ground base stations to low-orbit satellites and task offloading selection from high-altitude UAVs to low-orbit satellites.

[0078] 2) Resource allocation: In edge computing, the computing resources of MEC are usually limited. In the model proposed by the technical solution of the present invention, the computing resources of each satellite are limited, and the CPU frequency needs to be reasonably allocated to UAV and base station to achieve the purpose of optimizing energy consumption.

[0079] The technical solution of the present invention aims to develop an energy consumption optimization method for an air-space-ground network to solve the problems existing in the prior art. The present invention proposes a novel energy consumption optimization method, firstly gives an overall architecture of an air-space-ground network, and describes the unloading strategy model of the UAV and the base station initiating computing task requests in the model, then uses Shannon's theorem to describe the communication models between the three devices respectively, and describes the strategy of the UAV and the base station initiating computing task requests in the model.

[0080] 1. System Model

[0081] The system model is an air-ground wireless network, which consists of N BSs, M LEOs, and U high-altitude UAVs. LEO satellites provide computing services for BSs and high-altitude UAVs. BS tasks can be calculated locally (i.e., the nearest satellite) or forwarded to other satellites for computing.

[0082] 2. Uninstallation Strategy Model

[0083] BS requests the nearest satellite to obtain image data and selects one of the M LEOs for task offloading, i.e., α n,m ∈{0,1},α n,m =1 means the task of the nth BS is offloaded to the mth LEO, otherwise it is 0. To ensure that the computing task of each BS can be calculated by a satellite, the constraint condition LEO can choose to calculate locally or forward data and requests to other satellites for calculation, so assuming a parameter ξ n ∈{0,1} is used to indicate whether the task request of the nth BS is processed locally, ξ n =0 means local processing. 0 Refers to the satellite closest to the nth BS, also known as the local satellite.

[0084] The UAV obtains image data by taking photos at high altitude, selects one of the M LEOs to send a request and data through the uplink channel, and then the LEO performs calculations, that is, β u,m ∈{0,1},β u,m =1 indicates that the mission and data of the u-th UAV are offloaded to the m-th LEO, otherwise it is 0.

[0085] 3. Communication model

[0086] The mission computation model involves three different communication channels, namely, ground satellite channel, UAV satellite channel and inter-satellite channel.

[0087] 1) Ground satellite communication model

[0088] For communications between ground base stations and satellites, there is usually a problem of co-channel interference during transmission due to frequency reuse. We use NOMA technology and enable the receiving end to use SIC technology to solve the interference problem between ground base stations, thereby eliminating the impact caused by interference.

[0089] The rate of the channel between BS n and LEO m is expressed as follows

[0090] Among them: B 1 is the bandwidth of the satellite-to-ground channel, g n,m represents the channel gain from ground base station n to LEO m, is the signal transmission power from BS n to LEO m, represents the additive white Gaussian noise variance.

[0091] The time delay required for BS n to upload the task request to the nearest satellite is

[0092] Where: C BS The message size of the task request initiated by BS n.

[0093] The propagation delay of BS n is

[0094] Where: c is the speed of light.

[0095] The energy consumption of task request upload of BS n is

[0096] 2) UAV satellite communication model

[0097] During the optical communication between high-altitude UAV and LEO satellite, the altitude of the UAV is often affected by atmospheric turbulence, and the UAV flying at a fast speed will be affected by Doppler frequency shift and pointing errors caused during the turning process.

[0098] The altitude of high-altitude UAV and LEO is relatively high from the ground and cannot be ignored, so the flight radius of the UAV is modeled as r u =r e +h u ; Model the LEO radius as r m =r e +h m , r e is the radius of the earth. The frequency shift due to the Doppler effect is

[0099] Where: q u,m is the channel frequency from UAV to LEO satellite, r m is the distance from the satellite to the center of the Earth, ν u is the angular velocity of UAV u, t is ν u The flight time of high altitude UAVs is usually faster, and the pointing error is more obvious, so the pointing error is calculated as

[0100] Where: u is the beam radius, φ = φ j d lnk , ψ=ψ j d lnk , d lnk is the length of the link, d lnk =(r m-h u )sec(z), where z is the zenith angle.

[0101] Since the normal distribution can well describe the light intensity flicker of atmospheric turbulence, we model it as

[0102] in: S ATM is the atmospheric turbulence intensity, λ is the wavelength, and D is the antenna diameter.

[0103] The effective satellite channel gain is g u,m =Δq u,m .O u,m (r u ).τ u,m (C)

[0104] The data transmission rate from UAV to LEO is

[0105] Among them: B 2 represents the UAV and LEO channel bandwidth, represents the additive white Gaussian noise variance, is the transmission power of UAV u.

[0106] The delay required for UAV u to upload mission requests and data to the satellite is

[0107] in It is the sum of the image data sent by UAV u and the message data size of the task request.

[0108] The propagation delay of UAV u is

[0109] The energy consumption of UAV u’s mission request and data upload is

[0110] 3) Inter-satellite communication model

[0111] For inter-satellite channels between LEO satellites, free space optical (FSO) communication is used to reduce interference. The local satellite LEO m of BS n 0 The data transmission rate between LEO m is

[0112] in: is the local satellite m of BS n 0 The transmission power, λ is the wavelength, is the local satellite LEOm of BS n0 Distance from LEO m, B 3 represents the bandwidth of intersatellite communication, represents the additive white Gaussian noise variance. and LEO m 0 Aperture gain, pointing loss factor, aperture diameter, pointing error angle, and transmitter optical efficiency at . are the aperture gain, pointing loss factor, aperture diameter, pointing error angle, and transmitter optical efficiency of the receiver at LEO m, respectively.

[0113] Local satellite m of BS n 0 The transmission delay is

[0114] in: LEO m 0 The sum of the forwarded task request message size and the image data size.

[0115] Local satellite m of BS n 0 The propagation delay is

[0116] Local satellite m of BS n 0 The transmission energy consumption is

[0117] 4. Task Calculation Model

[0118] This involves the computing models of two types of equipment, namely BS and UAV. The model mainly focuses on the energy consumption and latency in the computing process.

[0119] 1) Computational model from base station to satellite

[0120] The delay of the task calculation process mainly includes three parts: data transmission delay, propagation delay and task calculation delay. The first two parts have been explained in the communication model.

[0121] The task computation delay of BS n is

[0122] in Calculate the number of clock cycles required for the task of BS n, f n,m Calculate the frequency allocated to BS n for LEO m.

[0123] The total delay of the entire task calculation process of BS n is

[0124] The total task computing delay is limited and needs to be completed within a limited range:

[0125] The energy consumption of the task calculation process mainly includes two parts: data transmission energy consumption and task calculation energy consumption. The former is described in the communication model.

[0126] The task computing energy consumption of BS n is

[0127] where κ m is the energy coefficient of the CPU chip architecture of LEO m.

[0128] The total energy consumption of the entire task calculation process of BS n is

[0129] 2) Computational model from UAV to LEO

[0130] The delay of the task calculation process mainly includes the delay of uploading task requests and data, propagation delay and calculation delay. The first two parts have been explained in the communication model.

[0131] The task calculation delay of UAV u is

[0132] in: Calculate the number of clock cycles required for the task of UAV u, f u,m Calculate the CPU frequency assigned to UAV u for LEO m.

[0133] The total delay of the entire mission calculation process of UAV u is

[0134] The total task computing delay is limited and needs to be completed within a limited range, that is,

[0135] The energy consumption of the task calculation process mainly includes two parts: data transmission energy consumption and task calculation energy consumption. The former is described in the communication model.

[0136] The energy consumption of the UAV u task calculation is

[0137] The total energy consumption of the entire mission calculation process of UAV u is

[0138] 5. Problem Formulation and Conditional Constraints

[0139] In this system, the task calculation time needs to be completed within the required range. The computing resources on the LEO satellite are limited. The satellite does not occupy the CPU frequency during transmission. Assuming that the channel bandwidth is large enough, under this condition, we mainly optimize the total energy consumption of the system. The problem and constraints are as follows:

[0140]

[0141] α n,m ,β u,m ,ξ n ∈{0,1}

[0142]

[0143] Figure 1 It is a flowchart of computational offloading and resource optimization of the crop yield estimation task in an embodiment of the present invention.

[0144] Step 101: Derivation of the transmit power p of the resource allocation strategy through preset conditions and Lagrange multiplication 0 and calculate the frequency f 0 .

[0145] Step 101 further includes the following steps:

[0146] Preset objective function:

[0147] Constraints:

[0148] Where: E n is the total energy consumption of BS n task computation, E u is the total energy consumption of UAV u task calculation, T n The total delay calculated for BS n tasks, The maximum delay calculated for BS n tasks, T u The total latency calculated for the UAV u mission, The maximum delay calculated for UAV u mission, α n,m is the offloading selection strategy of BS n, β u,m Select an offloading strategy for UAV u, ξ n is the local offloading policy of BS n, is the transmission power of BS n, is the maximum transmission power of BS n, is the transmission power of UAV u, is the maximum transmission power of UAV u, f n,m is the calculated frequency allocated by LEO m to BS n, f u,m is the calculated frequency assigned by LEO m to UAV u, The maximum value of the calculation frequency that can be provided for LEO m,

[0149] All constraints are added to the objective function using the Lagrange multiplier method. The Lagrange multiplier for each constraint is expressed as l 1,n, l 2,u , l 3,m Represented by, l represents the set of Lagrange multipliers;

[0150] The Lagrange multiplication of the objective function is represented by Λ(α, β, ξ, p, f), abbreviated as Λ;

[0151] Use α to represent α n,m The set composed of β represents β u,m The set composed of n,m The set composed of n The set composed of and The set composed of f n,m and f u,m A collection of components;

[0152] The partial derivatives of the transmission power p, computing resources f and Lagrange multiplier l in the Lagrange multiplication of the objective function are obtained respectively, that is, Let these equations equal 0, that is, Λ p (α,β,p,ξ,l)=0,Λ f (α,β,f,l)=0,Λ l (α,β,p,f,ξ)=0;

[0153] The equation relationship between the transmit power p, the calculation frequency f, the unloading selection parameters α, β, ξ and the Lagrange multiplier l is obtained, which is denoted as p = Λ p (α, β, ξ, l) and f = Λ f (α,β,l);

[0154] Communication rate between base station and satellite in the communication model (R n,m is the rate of the channel between BS n and LEO m, B 1 is the bandwidth of the satellite-to-ground channel, g n,m represents the channel gain from ground base station n to LEO m, is the signal transmission power from BS n to LEO m, represents the additive white Gaussian noise variance), in the objective function R n,m In the denominator, The derivation is more complicated and needs to be transformed before it can be solved;

[0155] make The above formula is transformed into (In the formula is the channel rate between BS n and the local satellite, is the transmission power of BS n to the local satellite), and a new transmission power replacement parameter is set make Convert the base station upload energy consumption into (In the formula, C BS is the message size of the task request initiated by BS n);

[0156] Communication rate between UAV and satellite in the communication model (R u,m is the channel rate between UAV u and LEO m, g u,m =Δq u,m .O u,m (r u ).τ u,m (C) is the effective satellite channel gain, B 2 represents the UAV and LEO channel bandwidth, represents the additive white Gaussian noise variance, d u,m is the distance between UAV u and satellite m, q u,m is the channel frequency from UAV to LEO satellite) and the inter-satellite communication rate (In the formula is the transmission power of the local satellite of BS n, λ is the wavelength, is the distance between the local satellite of BS n and LEO m, B 3 represents the bandwidth of intersatellite communication, represents the additive white Gaussian noise variance, and are the aperture gain, pointing loss factor, aperture diameter, pointing error angle and transmitter optical efficiency at the local satellite of BS n, respectively. are the aperture gain, pointing loss factor, aperture diameter, pointing error angle and transmitter optical efficiency of the receiver at LEO m respectively. They are also in the denominator of the objective function. and The derivative is equally complicated;

[0157] Use the same method with new parameters and Make a transformation;

[0158] use Indicates the transmit power replacement parameter and A collection of components;

[0159] From the computational model (E n c is the computational energy consumption of BS n, κm is the chip energy architecture coefficient of LEO m, Calculate the number of clock cycles required for the task of BS n, f n,m The calculated frequency allocated to BS n by LEO m) and (In the formula is the computational energy consumption of UAV u, Calculate the number of clock cycles required for the task of UAV u, f u,m = is the CPU calculation frequency allocated to UAV u by LEO m) From the two equations, we can see that the less computing resources the satellite allocates to the device (the smaller f), the lower the computing energy consumption; from the calculation delay equation, we can see that the less computing resources the satellite allocates to the device (the smaller f), the greater the calculation delay;

[0160] Depend on (Where T n c The relationship between computing energy consumption and computing delay can be obtained by using the two equations:

[0161] pass (Where T n is the delay of the entire task calculation process of BS n, is the maximum delay of the entire task calculation process of BSn) and we can get α n,m ,ξ n , and f n,m The equation relationship between them is denoted as p = Ψ p (α,β,ξ,f);

[0162] pass (Where T u is the latency of the entire mission calculation process of UAV u, is the maximum delay of the entire mission calculation process of UAV u) and we can get β u,m , and f u,m The equation relationship is recorded as f = Ψ f (α,β,ξ,p);

[0163] Combining the above, we get p = Λ p (α,β,ξ,l),f=Λ f (α, β, l) and p = Ψ p (α,β,ξ,f),f=Ψ f (α, β, ξ, p) These equations give the value of the transmitted power p and the calculated frequency f, denoted as p 0 and f 0 .

[0164] Step 102: Solve the uninstall selection problem to obtain the uninstall selection strategy parameter α i ,β i ,ξ i .

[0165] Step 102 further includes the following steps:

[0166] The problems and constraints are as follows:

[0167]

[0168] α n,m ,β u,m ,ξ n ∈{0,1}

[0169]

[0170] In order to solve the offloading selection strategy when the transmission power p and the task calculation frequency f are known, the variable offloading selection parameters α, β, ξ are first relaxed to continuous variables, that is, the constraints α, β, ξ∈{0,1} become α, β, ξ∈[0,1];

[0171] Substitute the values ​​of the transmission power p and the task calculation frequency f into the original problem P to obtain the offloading selection problem P1;

[0172] Based on problem P1, construct the Lagrangian function Λ P1 (α, β, ξ, Γ, l), the original integer programming problem P1 is transformed into a convex optimization problem;

[0173] The total energy consumption of the entire task calculation process of the nth BS is (E n is the total energy consumption of the entire task calculation process of BS n, The size of the task request message forwarded by the local satellite of BS n and the size of the image data) contains the BS offloading selection parameter α n,m and the locally calculated parameter ξ n Two unknown quantities, and the multiplication of the two unknown quantities will make the trend of the function uncertain. The two unknown quantities need to be combined into a new variable Γ n,m , that is, Γ n,m =α n,m *ξ n , the constraints will also increase Let Γ represent Γ n,m A collection of components;

[0174] After the two unknown quantities are combined, due to the derived The formula only contains Γ n,m and α n,mTwo unknown quantities, we can get Γ n,m and α n,m The relationship between Γ=Ψ Γ (α);

[0175] Due to the derived The formula only contains the unloading selection parameter β of UAV u,m An unknown quantity can directly determine the value of the unloading selection parameter β of the UAV;

[0176] The partial derivatives of the unloading selection parameters α, β, and ξ in the Lagrangian multiplication of the objective function are obtained respectively, that is, Setting these equations equal to 0 gives Λ α (α,l)=0,Λ β (β,l)=0,Λ ξ (ξ,l)=0,Λ Γ (Γ,l)=0,Λ l (α, β, ξ, Γ) = 0, these equations are combined to obtain the values ​​of the BS offloading selection parameter α and the local calculation parameter ξ;

[0177] The values ​​of the unloading selection parameters α, β, ξ calculated in the above manner are between 0 and 1. Since the values ​​of the unloading selection parameters in the actual problem must be 0 or 1, it is necessary to map the continuous solution back to a 0-1 variable;

[0178] Set the offloading selection parameter α of each BS n,m The maximum value in is set to 1, and the others are set to 0;

[0179] The offloading selection parameter β of each UAV u,m The maximum value in is set to 1, and the others are set to 0;

[0180] For the local calculation parameter ξ of the BS n , set a judgment value, and calculate the parameter ξ locally n When it is greater than this value, n Set to 1, otherwise set to 0, set the judgment value to 0.5, satisfying ξ n >0.5, let ξ n =1, otherwise let ξ n =0.

[0181] Step 103: Solve the resource allocation problem to obtain the transmit power p of the resource allocation strategy. i and calculate the frequency f i .

[0182] Step 103 further includes the following steps:

[0183] The problems and constraints are as follows:

[0184]

[0185] Construct the Lagrangian function Λ P2 (p,f,l), replace the parameters with transmit power and Transform the transmission rate R in the Lagrange multiplication of the objective function respectively n,m , and R u,m ;

[0186] because The formula only contains and f n,m Three unknown quantities can be obtained, and the equation relationship between these three unknown quantities can be obtained, that is,

[0187] because The formula only contains and f u,m Two unknown quantities can be obtained by equation relationship, that is,

[0188] The two types of relations between the transmission power replacement parameter p and the calculation frequency f obtained above are combined into

[0189] The Lagrange multiplication of the objective function replaces the parameters of the calculation frequency f and the transformed transmission power respectively and Find the partial derivative, that is Setting these expressions equal to 0 gives Λ f (f,l)=0,

[0190] Through Lianli Λ f (f,l)=0, These equations lead directly to the alternative parameter for transmit power: and calculate the frequency f, and then replace the parameter of the transmission power with Perform an inverse transformation to the transmit power p, and then obtain the transmit power p.

[0191] Step 104: Select strategy parameter α according to the uninstallation i ,β i ,ξ i and the transmission power p of the resource allocation strategy i and calculate the frequency f i , and obtain the total energy consumption E i .

[0192] Step 104 further includes the following steps:

[0193] Substitute the offloading selection strategy parameters α, β, ξ, the transmission power p and the calculation frequency f into the objective function Find the value of the objective function, which is represented by E.

[0194] Step 105 , determine whether the total energy consumption converges or whether the rounds reach the maximum number, if yes, go to step 106 , otherwise repeat steps 102 to 104 .

[0195] Step 106: Output the uninstall selection strategy parameter α * ,β * ,ξ * , the transmission power p of the resource allocation strategy * and calculate the frequency f * , and the suboptimal solution E for total energy consumption * .

[0196] Using the unloading selection parameter α obtained in the i-th round of iteration i ,β i ,ξ i , transmission power p i and calculate the frequency f i Find the objective function The value of E i , the system energy consumption value obtained in the i-1th round is E i-1 ;

[0197] The absolute value difference of the two total energy consumption E obtained from two consecutive iterations is used to determine whether the results converge. That is, the absolute value difference between the result of the i-th iteration and the result of the i-1th iteration is expressed as |E i -E i-1 |, if the value is less than a value Δ, the total energy consumption E i Convergence, the value of Δ can be assigned as needed, and it is set to 1e-7, expressed as |E i -E i-1 |<Δ;

[0198] At the same time, set the maximum number of iterations i max , when i>i max and |E i -E i-1 |<ΔWhen one of the two conditions is met, the iteration process ends and the suboptimal solution total energy consumption E is obtained. * , uninstall selection parameter α * ,β * ,ξ * , transmission power p * and computing resources * ;

[0199] If the total energy consumption E obtained in the first round of iterations does not converge and does not reach the maximum number of iterations i max , then repeat steps 102 to 104, and make a judgment again on the result obtained by the second round of iteration. If it is not satisfied, continue the third round of iteration until one of the two conditions is satisfied.

[0200] In the above embodiment, a new iterative optimization method is proposed, which decomposes the problem into two sub-problems, namely, the calculation offloading selection problem and the resource allocation problem, and obtains a suboptimal solution through iterative solution. Focusing on inter-satellite forwarding communication, considering that the base station is located on the ground, the number of satellites that can directly communicate in the same time slice is small, while the high-altitude UAV is at a high altitude, and the number of satellites that can communicate in the same time slice is more, so when comprehensively considering the optimal energy consumption, the tasks initiated by the base station are sent to the unloading target satellite for calculation through the inter-satellite forwarding method. Performance improvement and economic benefits, through the proposed model and method, the performance of satellite mission calculation can be improved, energy consumption can be effectively reduced, and economic benefits can be improved.

[0201] The basic principle of the technical solution is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the technical solution are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the technical solution. In addition, the specific details of the above technical solution are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the technical solution to the need to adopt the above specific details to be implemented.

[0202] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0203] The block diagrams of the devices, apparatuses, equipment, and systems involved in the present technical solution are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, meaning "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The word "such as" used here refers to the phrase "such as but not limited to", and can be used interchangeably with it.

[0204] The method and apparatus of the present technical solution may be implemented in many ways. For example, the method and apparatus of the present technical solution may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present technical solution are not limited to the order specifically described above, unless otherwise specifically stated. In addition, in some embodiments, the present technical solution may also be implemented as a program recorded in a recording medium, which includes machine-readable instructions for implementing the method according to the present technical solution. Therefore, the present technical solution also covers a recording medium storing a program for executing the method according to the present technical solution.

[0205] It should also be noted that, in the apparatus, equipment and method of the present technical solution, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent solutions of the present technical solution. The above description of the aspects of the technical solution is provided to enable any technician in the field to make or use the present technical solution. Various modifications to these aspects are very obvious to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the present technical solution. Therefore, the present technical solution is not intended to be limited to the aspects shown here, but in accordance with the widest range consistent with the principles and novel features of this technical solution.

[0206] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present technical solution to the form of the technical solution herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A method for computational offloading and resource optimization of a crop yield estimation task, characterized in that: The following steps are involved: Step 101: derive the transmit power p0 and calculation frequency f0 of the resource allocation strategy through preset conditions and Lagrange multiplication; Step 102: Solve the uninstall selection problem to obtain the uninstall selection strategy parameter α i ,β i ,ξ i ; Step 103: Solve the resource allocation problem to obtain the transmit power p of the resource allocation strategy. i and calculate the frequency f i ; Step 104: Select strategy parameter α according to the uninstallation i ,β i ,ξ i and the transmission power p of the resource allocation strategy i and calculate the frequency f i , and obtain the total energy consumption E i ; Step 105: determine whether the total energy consumption has converged or whether the rounds have reached the maximum number of times. If yes, go to step 106, otherwise repeat steps 102 to 104; Step 106: Output the uninstall selection strategy parameter α * ,β * ,ξ * , the transmission power p of the resource allocation strategy * and calculate the frequency f * , and the suboptimal solution E for total energy consumption * .

2. The method for computing offloading and optimizing resources for crop yield estimation tasks according to claim 1, characterized in that: Step 101 further includes the following steps: Preset objective function: Constraints: Where: E n is the total energy consumption of BS n task computation, E u is the total energy consumption of UAV u task calculation, T n The total delay calculated for BS n tasks, T n max The maximum delay calculated for BS n tasks, T u The total delay calculated for UAV u mission, T u max The maximum delay calculated for UAV u mission, α n,m is the offloading selection strategy of BS n, β u,m Select an offloading strategy for UAV u, ξ n is the local offloading policy of BS n, is the transmission power of BS n, is the maximum transmission power of BS n, is the transmission power of UAV u, is the maximum transmission power of UAV u, f n,m is the calculated frequency allocated by LEO m to BS n, f u,m is the calculated frequency assigned by LEO m to UAV u, The maximum value of the calculation frequency that can be provided for LEO m; All constraints are added to the objective function using the Lagrange multiplier method. The Lagrange multiplier for each constraint is expressed as l 1,n , l 2,u , l 3,m Represented by, l represents the set of Lagrange multipliers; The Lagrange multiplication of the objective function is represented by Λ(α, β, ξ, p, f), abbreviated as Λ; Use α to represent α n,m The set composed of β represents β u,m The set composed of n,m The set composed of n The set composed of and The set composed of f n,m and f u,m A collection of components; The partial derivatives of the transmission power p, computing resources f and Lagrange multiplier l in the Lagrange multiplication of the objective function are obtained respectively, that is, Let these equations equal 0, that is, Λ p (α,β,p,ξ,l)=0,Λ f (α,β,f,l)=0,Λ l (α,β,p,f,ξ)=0; The equation relationship between the transmit power p, the calculation frequency f, the unloading selection parameters α, β, ξ and the Lagrange multiplier l is obtained, which is denoted as p = Λ p (α, β, ξ, l) and f = Λ f (α,β,l); Communication rate between base station and satellite in the communication model (R n,m is the rate of the channel between BS n and LEOm, B1 is the bandwidth of the satellite-to-ground channel, g n,m represents the channel gain from ground base station n to LEO m, is the signal transmission power from BSn to LEO m, represents the additive white Gaussian noise variance), in the objective function R n,m In the denominator, The derivation is more complicated and needs to be transformed before it can be solved; make The above formula is transformed into (In the formula is the channel rate between BS n and the local satellite, is the transmission power of BS n to the local satellite), and a new transmission power replacement parameter is set make Convert the base station upload energy consumption into (In the formula, C BS is the message size of the task request initiated by BS n); Communication rate between UAV and satellite in the communication model (R u,m is the channel rate between UAV u and LEO m, g u,m =Δq u,m .O u,m (r u ).τ u,m (C) is the effective satellite channel gain, B2 represents the UAV and LEO channel bandwidth, represents the additive white Gaussian noise variance, d u,m is the distance between UAV u and satellite m, q u,m is the channel frequency from UAV to LEO satellite) and the inter-satellite communication rate (In the formula is the transmission power of the local satellite of BS n, λ is the wavelength, is the distance between the local satellite of BS n and LEO m, B3 represents the bandwidth of intersatellite communication, represents the additive white Gaussian noise variance, and are the aperture gain, pointing loss factor, aperture diameter, pointing error angle and transmitter optical efficiency at the local satellite of BS n, respectively. are the aperture gain, pointing loss factor, aperture diameter, pointing error angle and transmitter optical efficiency of the receiver at LEO m respectively. They are also in the denominator of the objective function. and The derivative is equally complicated; Use the same method with new parameters and Make a transformation; use Indicates the transmit power replacement parameter and A collection of components; From the computational model (In the formula is the computational energy consumption of BS n, κ m is the chip energy architecture coefficient of LEO m, Calculate the number of clock cycles required for the task of BS n, f n,m The calculated frequency allocated to BS n by LEO m) and (In the formula is the computational energy consumption of UAV u, Calculate the number of clock cycles required for the task of UAV u, f u,m = is the CPU calculation frequency allocated to UAV u by LEO m) From the two equations, we can see that the less computing resources the satellite allocates to the device (the smaller f), the lower the computing energy consumption; from the calculation delay equation, we can see that the less computing resources the satellite allocates to the device (the smaller f), the greater the calculation delay; Depend on (In the formula The relationship between computing energy consumption and computing delay can be obtained by using the two equations: pass (Where T n is the delay of the entire task calculation process of BS n, is the maximum delay of the entire task calculation process of BS n) and we can get α n,m ,ξ n , and f n,m The equation relationship between them is denoted as p = Ψ p (α,β,ξ,f); pass (Where T u is the latency of the entire mission calculation process of UAV u, is the maximum delay of the entire mission calculation process of UAV u) and we can get β u,m , and f u,m The equation relationship is recorded as f = Ψ f (α,β,ξ,p); Combining the above, we get p = Λ p (α,β,ξ,l),f=Λ f (α, β, l) and p = Ψ p (α,β,ξ,f),f=Ψ f (α, β, ξ, p) These equations yield the values ​​of the transmitted power p and the calculated frequency f, denoted as p0 and f0.

3. The method for computing offloading and optimizing resources for crop yield estimation tasks according to claim 1, characterized in that: Step 102 further includes the following steps: The problems and constraints are as follows: In order to solve the offloading selection strategy when the transmission power p and the task calculation frequency f are known, the variable offloading selection parameters α, β, ξ are first relaxed to continuous variables, that is, the constraints α, β, ξ∈{0,1} become α, β, ξ∈[0,1]; Substitute the values ​​of the transmission power p and the task calculation frequency f into the original problem P to obtain the offloading selection problem P1; Based on problem P1, construct the Lagrangian function Λ P1 (α, β, ξ, Γ, l), the original integer programming problem P1 is transformed into a convex optimization problem; The total energy consumption of the entire task calculation process of the nth BS is (E n is the total energy consumption of the entire task calculation process of BS n, The size of the task request message forwarded by the local satellite of BS n and the size of the image data) contains the BS offloading selection parameter α n,m and the locally calculated parameter ξ n Two unknown quantities, and the multiplication of the two unknown quantities will make the trend of the function uncertain. The two unknown quantities need to be combined into a new variable Γ n,m , that is, Γ n,m =α n,m *ξ n , the constraints will also increase Let Γ represent Γ n,m A collection of components; After the two unknowns are combined, due to the derived T n =T n max The formula only contains Γ n,m and α n,m Two unknown quantities, we can get Γ n,m and α n,m The relationship between Γ=Ψ Γ (α); Since the derived T u =T u max The formula only contains the unloading selection parameter β of UAV u,m An unknown quantity can directly determine the value of the unloading selection parameter β of the UAV; The partial derivatives of the unloading selection parameters α, β, and ξ in the Lagrangian multiplication of the objective function are obtained respectively, that is, Setting these equations equal to 0 gives Λ α (α,l)=0,Λ β (β,l)=0,Λ ξ (ξ,l)=0,Λ Γ (Γ,l)=0,Λ l (α, β, ξ, Γ) = 0, these equations are combined to obtain the values ​​of the BS offloading selection parameter α and the local calculation parameter ξ; The values ​​of the unloading selection parameters α, β, ξ calculated in the above manner are between 0 and 1. Since the values ​​of the unloading selection parameters in the actual problem must be 0 or 1, it is necessary to map the continuous solution back to a 0-1 variable; The offloading selection parameter α of each BS n,m The maximum value in is set to 1, and the others are set to 0; The offloading selection parameter β of each UAV u,m The maximum value in is set to 1, and the others are set to 0; For the local calculation parameter ξ of the BS n , set a judgment value, and calculate the parameter ξ locally n When it is greater than this value, n Set to 1, otherwise set to 0, set the judgment value to 0.5, satisfying ξ n >0.5, let ξ n =1, otherwise let ξ n =0.

4. The method for computing offloading and optimizing resources for crop yield estimation tasks according to claim 1, characterized in that: Step 103 further includes the following steps: The problems and constraints are as follows: Construct the Lagrangian function Λ P2 (p,f,l), replace the parameters with transmit power and Transform the transmission rate R in the Lagrange multiplication of the objective function respectively n,m , and R u,m ; Due to T n =T n max The formula only contains and f n,m Three unknown quantities can be obtained, and the equation relationship between these three unknown quantities can be obtained, that is, Due to T u =T u max The formula only contains and f u,m Two unknown quantities can be obtained by the equation relationship between them, that is, Substitute the transmit power obtained above for the parameter The two types of relations for calculating the frequency f are combined into The Lagrange multiplication of the objective function replaces the parameters of the calculation frequency f and the transformed transmission power respectively and Find the partial derivative, that is Setting these expressions equal to 0 gives Λ f (f,l)=0, Through Lianli Λ f (f,l)=0, These equations lead directly to the alternative parameter for transmit power: and calculate the frequency f, and then replace the parameter of the transmission power with Perform an inverse transformation to the transmit power p, and then obtain the transmit power p.

5. The method for computing offloading and optimizing resources for crop yield estimation tasks according to claim 1, characterized in that: Step 104 further includes the following steps: Substitute the offloading selection strategy parameters α, β, ξ, the transmission power p and the calculation frequency f into the objective function Find the value of the objective function, which is represented by E.

6. The method for computing offloading and optimizing resources for crop yield estimation tasks according to claim 1, characterized in that: Step 105 and step 106 further include the following steps: Using the unloading selection parameter α obtained in the i-th round of iteration i ,β i ,ξ i , transmission power p i and calculate the frequency f i Find the objective function The value of E i , the system energy consumption value obtained in the i-1th round is E i-1 ; The absolute value difference of the two total energy consumption E obtained from two consecutive iterations is used to determine whether the results converge. That is, the absolute value difference between the result of the i-th iteration and the result of the i-1th iteration is expressed as |E i -E i-1 |, if the value is less than a value Δ, the total energy consumption E i Convergence, the value of Δ can be assigned as needed, and it is set to 1e-7, expressed as |E i -E i-1 |<Δ; At the same time, set the maximum number of iterations i max , when i>i max and |E i -E i-1 |<ΔWhen one of the two conditions is met, the iteration process ends and the suboptimal solution total energy consumption E is obtained. * , uninstall selection parameter α * ,β * ,ξ * , transmission power p * and computing resources * ; If the total energy consumption E obtained in the first round of iterations does not converge and does not reach the maximum number of iterations i max , then repeat steps 102 to 104, and make a judgment again on the result obtained by the second round of iteration. If it is not satisfied, continue the third round of iteration until one of the two conditions is satisfied.

Images