New energy storage data acquisition and calculation method based on unmanned aerial vehicle network

Abstract

The invention relates to a new energy storage data acquisition and calculation method based on an unmanned aerial vehicle network. A discrete planning vision field is set and recorded as T = 1, 2,..., T, T time slots exist, T serves as an index, each time slot has two stages, the duration time of one stage is tau, and it is assumed that tau = 1 second; in the second stage, the duration time # imgabs0 # is used for enabling the unmanned aerial vehicle to go back and forth at different places; assuming a two-dimensional Euclidean coordinate, each device d belongs to D is located at a coordinate ld = (xd, yd); defining L = {1, 2,..., L as a hovering point set of the unmanned aerial vehicle u; each position is la, a belongs to L, and the coordinate is (xd, yd); defining a set E to contain directed edges, recorded as (i, j) or eij, where i and j respectively represent a device in D and a position in I, where a communication link exists between the device and the position when the UAV is located; enabling any E of the Z (lv) to return to a group of edges of which the end points are located at the lv position; writing the # imgabs 1 # into a path set or a position pair in the # imgabs 2 #; s (z) and e (z) are recorded as the starting point and the ending point of the path # imgabs3 # respectively; s ((1, 2)) = s (e12) = 1, and dab = la-lb2 is defined as the Euclidean distance between the device / position a and the device / position b.

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned aerial vehicles, data processing, optimization, energy collection, and sensor equipment, and specifically relates to a new energy storage data collection and calculation method based on an unmanned aerial vehicle network. Background Art

[0002] Devices in a renewable energy storage system (EMS) network generate sensor data and use environmental energy for charging, such as the temperature of the energy storage inverter running, and the voltage of the solar panel. A key issue is the need to ensure that the data collected from the device is quickly processed by a nearby server. However, the server may be placed far away from the data source, and the link / path from the device to the server may be limited / congested. To improve communication and computing in energy storage management networks, the following technologies can be utilized: (i) computing power network, (ii) unmanned aerial vehicle (UAV), and (iii) in-band full-duplex (IBFD) radio. In short, computing power network extends the mobile edge computing (MEC) paradigm to utilize any connected node with processing resources, where the code and data of the task can be routed and processed at any node. As for drones, they can be used to improve the communication and computing of the network and minimize the energy consumption rate of the device. Finally, IBFD radio allows nodes to send and receive simultaneously on the same frequency.

[0003] A Chinese invention patent with application number CN202311689842.8 discloses a sensor-based energy storage data collection and analysis method and system, including: real-time monitoring of energy storage equipment; connecting to the energy storage EMS system, setting search symbols, retrieving and downloading monitoring parameter records; based on monitoring parameter records and real-time monitoring parameters; obtaining technical indicator information of energy storage equipment; using real-time monitoring parameters as input data; using a predictive model for data analysis; solving the technical problem of comprehensive monitoring and management of energy storage systems, and improving their operating efficiency and safety;

[0004] A Chinese invention patent with application number CN202410445422.3 is a non-periodic operation and maintenance method and system for energy storage systems based on EMS, including: configuring the edge architecture according to the energy storage system deployment endpoint information and EMS controller, obtaining a set of distributed edge computing nodes, collecting operation information, obtaining a set of edge node operation data streams for edge computing, obtaining a set of node anomaly analysis information, and then obtaining a set of target node operation and maintenance strategy parameters to perform non-periodic operation and maintenance tasks on the target energy storage system, ensuring real-time dynamic operation and maintenance management of the energy storage system, and improving the operation and maintenance efficiency and accuracy of the energy storage system.

[0005] After investigation, previous work on using drones to assist computing only considered one of the following system settings: (i) The device uploads all or part of its tasks to one or more drones. The authors studied the use of drones to improve the connection between the device and the computing base station, (iii) the drone acts as a repeater and server. The drone can perform tasks or offload these tasks to a server located on the drone. (iv) Multiple drones work together in a cluster to perform tasks offloaded by the device or other drones. A drone cluster is considered to help compute tasks with dependencies. The drone uses reinforcement learning to select neighboring drones and offload the data volume to the selected neighboring drones. However, none of these works consider new energy storage systems that can be recharged from the environment and full-duplex drone communication methods. Summary of the invention

[0006] The purpose of this invention is to provide a new energy storage data acquisition and calculation method based on a drone network. It takes into account the rate problem of large amounts of local data being transmitted to the cloud in the energy storage system. Unlike previous work, it uses drones as base stations to share the data transmitted by the local end, thereby increasing the rate of data transmission to the cloud. We develop a distributed protocol and a mixed integer linear program to transmit data to the cloud at the fastest rate to ensure that reasonable decisions can be made in the shortest time.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] The new energy storage data collection and calculation method based on the UAV network includes a quad-rotor UAV u; a discrete programming horizon is set, denoted as T = 1, 2, ..., T, there are T time slots, indexed by T, each time slot has two stages, the duration of the first stage is τ, which is used for communication between the UAV and the device, assuming τ = 1 second; the duration of the second stage is Used for drones to travel between different locations; assuming two-dimensional Euclidean coordinates, each device d∈D is located at coordinate l d =(x d ,y d ); define L = {1,2,...,L} as the set of hovering points of UAV u; each position l a , where a∈L, the coordinates are (x d ,y d ); define a set E to contain directed edges, denoted as (i, j) or e ij , where i and j represent the device in D and the location in I, respectively. When the UAV is at this location, there is a communication link between the device and the location. Let Z(l v ) Any E returns a set of endpoints located at l v The edge of the position; Written The path set or position pair in The starting point and end point of s((1,2))=s(e 12 )=1, define d ab =||l a -l b ||2 is the Euclidean distance between devices / locations a and b.

[0009] In each time slot, UAV u selects a path in ε[L], i.e., a directed edge connecting two hovering positions; a binary variable is defined for each path z∈ε[L] represents whether UAV u traverses path z in time slot t. UAV u can only choose one path in ε[L]. Mathematically speaking,

[0010]

[0011] The next constraint ensures that UAV u is at one location in each time slot, assuming where a∈L represents its position l in time slot t a ,

[0012]

[0013] For each link z∈ε[L], we have That is, if UAV u chooses path z, then UAV u must be at positions s(z) and e(z) at time t and time t+1, respectively.

[0014] The device has a half-duplex radio and the UAV u has an IBFD radio. At each location lv∈L, the UAV selects one or more links in Z(lv) to activate. Let Γ[lv] be the set of transmission sets involving links in Z(lv). Some example transmission sets in Γ[l1] include {{e A1},{e A1 ,e 1A},{e A1 ,e 1B},{e B1}};definition To indicate whether to choose Γ[l v ], at each time slot t and each location l v , can only select one transmission set, formally,

[0015]

[0016] According to the SINR threshold δ, each link in the transmission set c has a capacity Bc, defined as to indicate whether links (d,u) and (u,d) are active at time t; that is, whether device d is sending data to UAV u or receiving data from UAV u. The device is half-duplex, so we have

[0017]

[0018] In each time slot t, only one device can transmit to the UAV:

[0019]

[0020] The channel model is outlined, assuming that mutual channels and line of sight (LoS) are dominant

[18] ; for the channels between the UAV and the device and between devices i and j, we assume that given the distance d ij The log-normal path loss (PL) model is formally

[0021]

[0022] Where θ is the path loss exponent and Xσ is a Gaussian distribution with zero mean and standard deviation;

[0023] set up is the channel power gain between i and j at time slot t, where i and j are the UAV hovering positions or devices, and is defined as and are the transmit powers of UAV u and device d, respectively, and are bounded as follows:

[0024]

[0025] If the SINR at device d exceeds δ, then there exists a link (u, d) from UAV u to device d, formally,

[0026]

[0027] The symbol N0 represents the noise power;

[0028] The SINR at UAV u when link (d,u) is active is

[0029]

[0030] The value of β is determined by the self-cancellation capability of the IBFD radio.

[0031] The rate at which each device d generates data in each time slot t is (bit / s), with an upper limit of r max ; The device generates samples during the communication phase and UAV movement phase of each time slot, using To represent the amount of data sampled by device d in time slot t, i,e,

[0032] Given Calculate The UAV can offload / receive data to / from a device; it can receive or transmit respectively and (bit / s) amount of data; and The value of is constrained by the set of transmissions selected in each time slot and decreases as the UAV u and the device compute data from their respective queues.

[0033] Using the data queue model of "collect-then-use data queue", the data sampled / received in time slot t can only be processed / transmitted in the next time slot t+1. and are the data queues of device d at the beginning and end of time slot t, respectively; or represents the amount of data calculated by device d or UAV u at time t; the device and UAV u only communicate in the first phase of each time slot, the time length is τ seconds, then for each device d∈D, its queue evolves into

[0034]

[0035] Similarly, for UAV u,

[0036]

[0037] We constrain and The values ​​of , whose values ​​depend on the transmission set selected at each time slot t, if device d is an endpoint of link transmission set c, then let ψ(d,c) return the value 1; mathematically, these flow variables are constrained as follows:

[0038]

[0039]

[0040] Each device and UAV u has M homogeneous virtual machines (VMs) that they instantiate to process data, each with a capacity of μ (bits) and an operating cost of E α (Joule), in each time slot, each device and UAV u determines the number of virtual machines to instantiate, which in turn determines and The value of The number of virtual machines instantiated for device d, The number of virtual machines instantiated for UAV u, which are and As the boundary, that is, the maximum computing power, we have

[0041]

[0042] set up is the energy arriving at device d at time slot t, determined by the hidden-Markov model, which has four states corresponding to weather conditions, which we denote as S = {'Excellent', 'Good', 'Fair', 'Poor'}. Let Pss' be the transition probability from state s to s', where s, s'∈S, and each state s has a Gaussian distribution with mean μ s , variance σ s .

[0043] Given a processing speed f (cycles per second), the energy consumption for processing L bits is E p (L) = κP b Lf 2 , where κ represents the switch capacitance that depends on the processor architecture, P b is the number of CPU cycles required to process 1 bit;

[0044] UAV u and device d operate in time slot t and The energy costs incurred by the virtual machines are

[0045]

[0046] The communication cost of a node includes circuit and transmission power consumption. Let the circuit power consumption be c0. Each transmission set generates different transmission power for UAV u and the device. Let the transmission set c∈Γ[l v ], the transmission powers of UAV u and device d are and

[0047] The required SINR threshold is δ, and we have B c =B·log2(1+δ), where B is the channel bandwidth. When using transmission set c, the corresponding energy cost per bit is

[0048]

[0049] In the formula

[0050]

[0051]

[0052] The energy consumption of the UAV is due to hovering and moving between positions in l. Using the quadrotor energy model, for a given UAV velocity V, the following quantity is defined:

[0053]

[0054] Note that p0,U tip ,p i ,v0,d0,s,ρ,A are parameters; the power consumption caused by propulsion is P(V)=Ψ1(V)+Ψ2(V)+Ψ3(V), and the power consumption of UAV when hovering is p h =p0+p i [twenty two];

[0055] In each time slot t, UAV u flies between locations in l as sec,d ab Represents the distance between positions a and b, denoted by V ab is the flight distance d of UAV u in time τˉ ab If UAV u maintains its current position for two consecutive time slots, i.e., V aa =0, then its energy consumption rate is p h , the energy u consumed by the UAV flying in time slot t is

[0056]

[0057] Finally, we have the energy evolution of the device and UAV u, defining e_s as the cost of sampling one bit of data. Formally, for each device d∈D and time slot t∈T, we have

[0058]

[0059] For UAV u, we have

[0060]

[0061] In summary, we do not consider the packet loss problem. After evaluation, our results can be regarded as the best. Traditional energy storage communications cannot solve the problem of large-scale data transmission to the cloud. Using drones as base stations to share the data transmitted by the local energy storage system to the cloud can increase the transmission rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0063] Figure 1 This is a schematic diagram of the relationship between the number of devices and the computing rate;

[0064] Figure 2 Schematic diagram of the relationship between the number of time slots and the calculation rate;

[0065] Figure 3 Schematic diagram of the relationship between the maximum number of virtual machines and computing rate;

[0066] Figure 4 Schematic diagram of the relationship between initial energy and calculation rate;

[0067] Figure 5 Schematic diagram of the relationship between SINR threshold and calculation rate;

[0068] Figure 6 Schematic diagram of the relationship between the maximum value of the virtual machine on UVA and the computing rate. DETAILED DESCRIPTION

[0069] The technical solutions in the embodiments of the present invention are clearly and completely described below in conjunction with the accompanying drawings.

[0070] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0071] Example

[0072] The new energy storage data collection and calculation method based on the UAV network includes a quad-rotor UAV u; a discrete programming horizon is set, denoted as T = 1, 2, ..., T, there are T time slots, indexed by T, each time slot has two stages, the duration of the first stage is τ, which is used for communication between the UAV and the device, assuming τ = 1 second; the duration of the second stage is Used for drones to travel between different locations; assuming two-dimensional Euclidean coordinates, each device d∈D is located at coordinate l d =(x d ,y d ); define L = {1,2,...,L} as the set of hovering points of UAV u; each position l a , where a∈L, the coordinates are (x d ,y d ); define a set E to contain directed edges, denoted as (i, j) or e ij , where i and j represent the device in D and the location in I, respectively. When the UAV is at this location, there is a communication link between the device and the location. Let Z(l v) Any E returns a set of endpoints located at l v The edge of the position; Written The path set or position pair in The starting point and end point of s((1,2))=s(e 12 )=1, define d ab =||l a -l b ||2 is the Euclidean distance between devices / locations a and b.

[0073] In each time slot, UAV u selects a path in ε[L], i.e., a directed edge connecting two hovering positions; a binary variable is defined for each path z∈ε[L] represents whether UAV u traverses path z in time slot t. UAV u can only choose one path in ε[L]. Mathematically speaking,

[0074]

[0075] The next constraint ensures that UAV u is at one location in each time slot, assuming where a∈L represents its position l in time slot t a ,

[0076]

[0077] For each link z∈ε[L], we have That is, if UAV u chooses path z, then UAV u must be at positions s(z) and e(z) at time t and time t+1, respectively.

[0078] The device has a half-duplex radio and the UAV u has an IBFD radio. At each location lv∈L, the UAV selects one or more links in Z(lv) to activate. Let Γ[lv] be the set of transmission sets involving links in Z(lv). Some example transmission sets in Γ[l1] include {{e A1},{e A1 ,e 1A},{e A1 ,e 1B},{e B1}};definition To indicate whether to choose Γ[l v ], at each time slot t and each location l v , can only select one transmission set, formally,

[0079]

[0080] According to the SINR threshold δ, each link in the transmission set c has a capacity Bc, defined as to indicate whether links (d,u) and (u,d) are active at time t; that is, whether device d is sending data to UAV u or receiving data from UAV u. The device is half-duplex, so we have

[0081]

[0082] In each time slot t, only one device can transmit to the UAV:

[0083]

[0084] The channel model is outlined, assuming that mutual channels and line of sight (LoS) are dominant

[18] ; for the channels between the UAV and the device and between devices i and j, we assume that given the distance d ij The log-normal path loss (PL) model is formally

[0085]

[0086] Where θ is the path loss exponent and Xσ is a Gaussian distribution with zero mean and standard deviation;

[0087] set up is the channel power gain between i and j at time slot t, where i and j are the UAV hovering positions or devices, and is defined as and are the transmit powers of UAV u and device d, respectively, and are bounded as follows:

[0088]

[0089]

[0090] If the SINR at device d exceeds δ, then there exists a link (u, d) from UAV u to device d, formally,

[0091]

[0092] The symbol N0 represents the noise power;

[0093] The SINR at UAV u when link (d,u) is active is

[0094]

[0095] The value of β is determined by the self-cancellation capability of the IBFD radio.

[0096] The rate at which each device d generates data in each time slot t is (bit / s), with an upper limit of r max ; The device generates samples during the communication phase and UAV movement phase of each time slot, using To represent the amount of data sampled by device d in time slot t, i,e,

[0097] Given Calculate The UAV can offload / receive data to / from a device; it can receive or transmit respectively and (bit / s) amount of data; and The value of is constrained by the set of transmissions selected in each time slot and decreases as the UAV u and the device compute data from their respective queues.

[0098] Using the data queue model of "collect-then-use data queue", the data sampled / received in time slot t can only be processed / transmitted in the next time slot t+1. and are the data queues of device d at the beginning and end of time slot t, respectively; or represents the amount of data calculated by device d or UAV u at time t; the device and UAV u only communicate in the first phase of each time slot, the time length is τ seconds, then for each device d∈D, its queue evolves into

[0099]

[0100] Similarly, for UAV u,

[0101]

[0102] We constrain and The values ​​of , whose values ​​depend on the transmission set selected at each time slot t, if device d is an endpoint of link transmission set c, then let ψ(d,c) return the value 1; mathematically, these flow variables are constrained as follows:

[0103]

[0104] Each device and UAV u has M homogeneous virtual machines (VMs) that they instantiate to process data, each with a capacity of μ (bits) and an operating cost of E α (Joule), in each time slot, each device and UAV u determines the number of virtual machines to instantiate, which in turn determines and The value of The number of virtual machines instantiated for device d, The number of virtual machines instantiated for UAV u, which are and As the boundary, that is, the maximum computing power, we have

[0105]

[0106] set up is the energy arriving at device d at time slot t, determined by the hidden-Markov model, which has four states corresponding to weather conditions, which we denote as S = {'Excellent', 'Good', 'Fair', 'Poor'}. Let Pss' be the transition probability from state s to s', where s, s'∈S, and each state s has a Gaussian distribution with mean μ s , variance σ s .

[0107] Given a processing speed f (cycles per second), the energy consumption for processing L bits is E p (L) = κP b Lf 2 , where κ represents the switch capacitance that depends on the processor architecture, P b is the number of CPU cycles required to process 1 bit;

[0108] UAV u and device d operate in time slot t and The energy costs incurred by the virtual machines are

[0109]

[0110] The communication cost of a node includes circuit and transmission power consumption. Let the circuit power consumption be c0. Each transmission set generates different transmission power for UAV u and the device. Let the transmission set c∈Γ[l v ], the transmission powers of UAV u and device d are and

[0111] The required SINR threshold is δ, and we have B c =B·log2(1+δ), where B is the channel bandwidth. When using transmission set c, the corresponding energy cost per bit is

[0112]

[0113] In the formula

[0114]

[0115] The energy consumption of the UAV is due to hovering and moving between positions in l. Using the quadrotor energy model, for a given UAV velocity V, the following quantity is defined:

[0116]

[0117] Note that p0,U tip ,p i ,v0,d0,s,ρ,A are parameters; the power consumption caused by propulsion is P(V)=Ψ1(V)+Ψ2(V)+Ψ3(V), and the power consumption of UAV when hovering is p h =p0+p i [twenty two];

[0118] In each time slot t, UAV u flies between locations in l as sec,d ab Represents the distance between positions a and b, denoted by V ab is the flight distance d of UAV u in time τˉ ab If UAV u maintains its current position for two consecutive time slots, i.e., V aa =0, then its energy consumption rate is p h , the energy u consumed by the UAV flying in time slot t is

[0119]

[0120] Finally, we have the energy evolution of the device and UAV u, defining e_s as the cost of sampling one bit of data. Formally, for each device d∈D and time slot t∈T, we have

[0121]

[0122] For UAV u, we have

[0123]

[0124] Assume that all generated data must be calculated by the UAV and the device before time T, that is:

[0125]

[0126] Our problem is to maximize the minimum computation rate of all devices, formally,

[0127]

[0128] The decision variables include: 1. The sampling rate of each device 2. On each device and The number of virtual machines instantiated in each time slot,3. The number of virtual machines instantiated in each time slot by UAV u The location of,4,the transmission set used by UAV u in each time slot,the transmission set used by UAV u in each time slot, and the,transmission set used by UAV u in each time slot, and the,travel time slot The path of our MILP is as follows:

[0129]

[0130] (1)-(3),(11)-(18),(19)-(24),(28)-(30)

[0131] Finally, problem (P1) requires as input a set of transmission sets Γ[L], for each transmission set, the link has a transmit power, i.e. and In the next section, we propose a distributed protocol for computing Rmin for UAVs and devices.

[0132] The distributed protocol is to position the UAV u above the device with the smallest sampling rate and energy calculated in each time t, and forward its data to another device with remaining energy. The low-energy device only performs data sampling and uploading, and does not perform data calculation.

[0133] UAVu runs the algorithm and collects the status of each ground device d∈D at the beginning of each time slot t. The status information includes (a) current energy level (b) Sampling rate (c) Channel gain g ud , then the UAV calculates the following weights:

[0134]

[0135] A device with a lower weight means that it can only sample and process a small amount of data. To this end, UAV u selects the device with the smallest weight, i.e. Then, a REQ_UL message is sent to device d', i.e., Send(REQ_UL,d,), requesting data upload in the next time slot. In the current time slot, the UAV performs the following three tasks:

[0136] 1) Upload: If the UAVu queue is empty, And from the device After receiving the START_UL message, the slave device Collect data and update its queue use indicates; note that this means that the device A REQ_UL message was received in the previous time slot; the amount of data τ by device otherwise, if its data queue is not empty, it sends a STOP_UL message to the device. Cancel the upload.

[0137] 2) Download: When UAVu transmits at the maximum transmit power, it constructs a set W of devices with SINR greater than β, that is:

[0138]

[0139] Next, it sends a * send The maximum weight of τ in W is This amount is

[0140]

[0141] The transmission power is

[0142]

[0143] We call Will τ bits sent to device d * ; UAV sends REQ_DL message to d * Notify download power, i.e.

[0144] 3) Compute: UAVu processes any data in its queue; the number of virtual machines to be instantiated is

[0145]

[0146] If the algorithm finds that the UAV does not have enough energy to complete the above three tasks,

[0147]

[0148] It terminates; otherwise, it enters the second stage.

[0149] In the second phase, the UAV Check if you have enough energy to move to the position (x d, ,y d, ),Right now

[0150]

[0151] If UAVu has enough energy, it will continue to operate; otherwise, it will terminate.

[0152] Algorithm 2 is executed by the ground device. In each time slot t, the ground device d determines the following quantities: (i) the amount of data to be sampled (ii) The amount of data that needs to be uploaded to the UAV in the next time slot, expressed as represents, (iii) the amount of data that needs to be processed, Indicates, attention, Each ground device has a data queue

[0153] The device first checks whether it has received a REQ_DL message from the UAV; if so, it retrieves the and After this, if the device decides to upload data to the UAV, it sets its transmit power To upload data, it sends a START_UL message to the UAV, and if it receives a STOP_UL message, it sets its upload power to Otherwise, it uses

[0154]

[0155] Next, the device determines the data to be uploaded and processed in the next time slot based on the following two situations:

[0156] 1) Case-1: At this time, device d does not receive the REQ_UL message, or there is a non-empty data queue in time slot t, then in time slot t+1, it needs to process all the sampled data in time slot t and set Device d first assumes that it does not have enough energy to operate at maximum computing power in time slot t+1, i.e. Then at time t+1, the amount of data calculated by device d is

[0157]

[0158] in τ represents the amount of data downloaded from the drone, such as in REQ D L message, on the other hand, for

[0159]

[0160] Note that if the device receives a STOP_UL message, we have Formula (8) can be written as

[0161]

[0162] Therefore, we have,

[0163]

[0164] Bundle Substituting into formula (12) we get The minimum number of virtual machines is

[0165]

[0166] Note that the available energy The number of exported virtual machines may be higher than and / or Therefore, device d needs to be recalculated The number of virtual machines in time period t+1 is based on actual numbers:

[0167]

[0168] If device d does not have enough energy to process the data sent by UAVu, then

[0169] 2) Case-2: Here, device d receives a REQ_UL message at time slot t and has an empty data queue It samples as much data as possible at time slot t, uploads it to the UAV at time slot t+1 and / or performs computation. Similar to Case-1, the device first assumes that it does not have enough energy, thus giving priority to uploading data to avoid wasting energy on computation. During the data upload process, the device may experience energy interruption. Alternatively, it may have enough energy to upload at maximum capacity, but not enough energy to process the data. Device d first assumes that it Insufficient energy for uploading. Assume that the UAV downloads at the maximum transmission power in time slot t+1. The amount of data that can be uploaded in time slot t+1 is

[0170]

[0171] Newton's law can be used to find the best Then the device sets for

[0172]

[0173] If the device has enough energy to upload at maximum capacity, i.e.

[0174]

[0175] Then device d has extra energy to calculate data, otherwise,

[0176] If inequality (49) holds and It is determined that device d assumes that its energy is insufficient to handle The following formula is used to calculate

[0177]

[0178] Then the device is inserted Calculate

[0179] At the end of each time slot, device d calculates its estimated available energy in time slot t+1, using The estimated available energy is the remaining energy minus the uploaded energy, assuming the maximum transmit power P max , calculate, where

[0180]

[0181] It also calculates the sample rate for calculations by ignoring data that it has not processed or sent to the UAV Specifically, we have

[0182]

[0183] Device d calculates and After that, and Send it to UAV. Finally, device d updates the actual remaining energy using equation (12):

[0184] Test results:

[0185] 1) Number of devices: Figure 1 It is shown that the max-min computation rate of all solutions decreases as the number of devices increases. This is because it is more likely that a device with lower initial energy will be found.

[0186] 2) Number of time slots: Figure 2 In , we see that the max-min computation rate decreases as the number of time slots increases. This is because a large number of time slots results in a longer processing time and thus a lower computation rate.

[0187] 3) Maximum number of virtual machines on the device: Figure 3 The results show that when Vdmax is 2 and 4, the max-min computation rate of the distributed protocol is lower than that of MILP half-duplex; the reason is that it greedily sends data to the device with the highest weight in Eq. (33); this means that a device with low computation rate but high energy level needs to compute data for other devices; after increasing the maximum number of virtual machines, the device with surplus energy instantiates a large number of virtual machines; therefore, it has the opportunity to compute its own sampled data.

[0188] 4) Initial energy: from Figure 4It can be seen that the max-min computation rate increases with increasing initial energy; this is expected since each device has more energy to process its sampled data.

[0189] 5) SINR threshold: reference Figure 5 , for the case without UAV, all solutions converge to the same result of MILP calculation; this is because when the SINR threshold is large, the device cannot transmit its data to the UAV, so there is no full-duplex or half-duplex link; we also notice that when δ ≥ 30, the maximum minimum calculation rate of the distributed protocol is smaller than that of MILP half-duplex; the reason is that in the case of large SINR, the number of links in full-duplex is less than that in half-duplex due to self-interference.

[0190] 6) Maximum number of virtual machines on UAV: Figure 6 It can be seen that when Vumax = 1, the maximum-minimum computation rate of MILP half-duplex is 68% of that of the distributed protocol, which is the same as when Vumax ≥ 4; this is because the lowest-energy device using half-duplex can only send its data to the UAV for computation; therefore, the maximum-minimum computation rate increases with the increase in the computational power of the UAV.

[0191] In the description of the present invention, it is necessary to understand that the directions or positional relationships indicated by the terms "front and back", "left and right", etc. are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or component referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the invention.

[0192] Of course, in the present technical solution, those skilled in the art should understand that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the element may be multiple, and the term "one" should not be understood as a limitation on the quantity.

[0193] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art under the technical guidance of the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A new energy storage data collection and calculation method based on drone network, characterized in that: Including a quadrotor drone u; set a discrete programming horizon, denoted as T = 1, 2, ..., T, with T time slots, indexed by T, each time slot has two phases, where the duration of the first phase is τ, which is used for communication between the drone and the device, assuming τ = 1 second; Phase II, duration Used for drones to travel between different locations; assuming two-dimensional Euclidean coordinates, each device d∈D is located at coordinate l d =(x d ,y d ); define L = {1,2,...,L} as the set of hovering points of UAV u; each position l a , where a∈L, the coordinates are (x d ,y d ); define a set E to contain directed edges, denoted as (i, j) or e ij , where i and j represent the device in D and the location in l, respectively. When the UAV is at this location, there is a communication link between the device and the location. Let Z(l v ) Any E returns a set of endpoints located at l v The edge of the position; Written The path set or position pair in The starting point and end point of s((1,2))=s(e 12 )=1, define d ab =||l a -l b ||2 is the Euclidean distance between devices / locations a and b.

2. The new energy storage data collection and calculation method based on drone network according to claim 1 is characterized in that: In each time slot, UAV u selects a path in ε[L], i.e., a directed edge connecting two hovering positions; a binary variable is defined for each path z∈ε[L] represents whether UAV u traverses path z in time slot t. UAV u can only choose one path in ε[L]. Mathematically speaking, The next constraint ensures that UAV u is at one location in each time slot, assuming where a∈L It represents its position l in time slot t a , 3. The new energy storage data collection and calculation method based on drone network according to claim 2 is characterized in that: For each link z∈ε[L], we have That is, if UAV u chooses path z, then UAV u must be at positions s(z) and e(z) at time t and time t+1, respectively.

4. The new energy storage data acquisition and calculation method based on drone network according to claim 1 is characterized in that: The device has a half-duplex radio and the UAV has an IBFD radio, At each location lv∈L, the UAV selects one or more links in Z(lv) to activate. Let Γ[lv] be the set of transmission sets involving links in Z(lv). Some example transmission sets in Γ[l1] include {{e A1 },{e A1 ,e 1A },{e A1 ,e 1B },{e B1 }};definition To indicate whether to choose Γ[l v ], at each time slot t and each location l v , can only select one transmission set, formally, According to the SINR threshold δ, each link in the transmission set c has a capacity Bc, defined as to indicate whether links (d,u) and (u,d) are active at time t; that is, whether device d is sending data to UAV u or receiving data from UAV u. The device is half-duplex, so we have In each time slot t, only one device can transmit to the UAV: The channel model is outlined, assuming that mutual channels and line of sight (LoS) are dominant [18]; for the channels between the UAV and the device and between devices i and j, we assume that given the distance d ij The log-normal path loss (PL) model is formally in is the path loss exponent, Xσ is a Gaussian distribution with zero mean and standard deviation; set up is the channel power gain between i and j at time slot t, where i and j are the UAV hovering positions or devices, and is defined as and are the transmit powers of UAV u and device d, respectively, and are bounded as follows: If the SINR at device d exceeds δ, then there exists a link (u, d) from UAV u to device d, formally, The symbol N0 represents the noise power; The SINR at UAV u when link (d,u) is active is The value of β is determined by the self-cancellation capability of the IBFD radio.

5. The new energy storage data acquisition and calculation method based on drone network according to claim 1 is characterized in that: The rate at which each device d generates data in each time slot t is The upper limit is r max ; The device generates samples during the communication phase and UAV movement phase of each time slot, using To represent the amount of data sampled by device d in time slot t, i,e, .

6. The new energy storage data collection and calculation method based on drone network according to claim 5 is characterized in that: Given Calculate The UAV can offload / receive data to / from a device; it can receive or transmit respectively and (bit / s) amount of data; and The value of is constrained by the set of transmissions selected in each time slot and decreases as the UAV u and the device compute data from their respective queues; Using the "collect-then-use data queue" data queue model, the data sampled / received in time slot t can only be processed / transmitted in the next time slot t+1. Define and are the data queues of device d at the beginning and end of time slot t, respectively; or represents the amount of data calculated by device d or UAV u at time t; the device and UAV u only communicate in the first phase of each time slot, the time length is τ seconds, then for each device d∈D, its queue evolves into Similarly, for UAV u, We constrain and The values ​​of , whose values ​​depend on the transmission set selected at each time slot t, if device d is an endpoint of link transmission set c, then let ψ(d,c) return the value 1; mathematically, these flow variables are constrained as follows:

7. The new energy storage data collection and calculation method based on drone network according to claim 1 is characterized in that: Each device and UAV u has M homogeneous virtual machines (VMs) that they instantiate to process data, each with a capacity of μ (bits) and an operating cost of E α (Joule), in each time slot, each device and UAV u determines the number of virtual machines to instantiate, which in turn determines and The value of The number of virtual machines instantiated for device d, The number of virtual machines instantiated for UAV u, which are and As the boundary, that is, the maximum computing power, we have 8. The new energy storage data collection and calculation method based on drone network according to claim 1 is characterized in that: set up is the energy arriving at device d at time slot t, determined by the hidden-Markov model, which has four states corresponding to weather conditions, which we denote as S = {'Excellent', 'Good', 'Fair', 'Poor'}. Let Pss' be the transition probability from state s to s', where s, s'∈S, and each state s has a Gaussian distribution with mean μ s , variance σ s .

9. The new energy storage data collection and calculation method based on drone network according to claim 1 is characterized in that: Given a processing speed f (cycles per second), the energy consumption for processing L bits is E p (L) = κP b Lf 2 , where k represents the switch capacitance that depends on the processor architecture, P b is the number of CPU cycles required to process 1 bit; UAV u and device d operate in time slot t and The energy costs incurred by the virtual machines are The communication cost of a node includes circuit and transmission power consumption. Let the circuit power consumption be c0. Each transmission set generates different transmission power for UAV u and the device. Let the transmission set c∈Γ[l v ], the transmission powers of UAV u and device d are and .

10. The new energy storage data collection and calculation method based on drone network according to claim 9 is characterized in that: The required SINR threshold is δ, and we have B c =B·log2(1+δ), where B is the channel bandwidth. When using transmission set c, the corresponding energy cost per bit is In the formula The energy consumption of the UAV is due to hovering and moving between positions in l. Using the quadrotor energy model, for a given UAV velocity V, the following quantity is defined: Note that p0,U tip ,p i ,v0,d0,s,ρ,A are parameters; the power consumption caused by propulsion is P(V)=Ψ1(V)+Ψ2(V)+Ψ3(V), and the power consumption of UAV when hovering is p h =p0+p i [twenty two]; In each time slot t, UAV u flies between locations in l as sec,d ab Represents the distance between positions a and b, denoted by V ab is the flight distance d of UAV u in time τˉ ab If UAV u maintains its current position for two consecutive time slots, i.e., V aa =0, then its energy consumption rate is p h , the energy u consumed by the UAV flying in time slot t is Finally, we have the energy evolution of the device and UAV u, defining e_s as the cost of sampling one bit of data. Formally, for each device d∈D and time slot t∈T, we have For UAV u, we have

Images