A method for designing energy efficiency of unmanned aerial vehicle communication system considering wind interference

Abstract

The present application belongs to the field of unmanned aerial vehicle and wireless communication technology, and relates to a design scheme for energy efficiency of an unmanned aerial vehicle communication system considering the interference of wind. In order to reduce the influence of wind on the energy consumption of the unmanned aerial vehicle and the energy efficiency design problem of the communication system thereof, a general-purpose energy consumption propulsion model (GPECM) in a three-dimensional space is derived under the consideration of the effect of wind on the unmanned aerial vehicle, and an online design based on offline design (OBOA) scheme is proposed for the unmanned aerial vehicle communication system based on the model to maximize the energy efficiency of the system. The scheme includes the optimization of the offline unmanned aerial vehicle trajectory and user selection strategy based on the statistical distribution information of wind in the offline stage, and the online optimization of the flight speed and trajectory of the unmanned aerial vehicle based on the trajectory obtained in the offline stage in the online stage.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) and wireless communication technology, and relates to an energy efficiency design method for UAV communication systems that takes into account wind interference. Background Technology

[0002] The ability of drones to provide low-cost and efficient diversified services in future wireless communication systems has attracted widespread attention in the industry. Drones possess characteristics such as high mobility and rapid deployment capabilities, enabling them to easily establish line-of-sight (LoS) links with ground communication nodes and achieve high-quality data collection and delivery tasks, while adapting to various complex environments. Drones can therefore be used in many industrial applications, such as data collection for the Internet of Things (IoT), mobile edge computing, secure communications, and disaster recovery communication services.

[0003] Despite the numerous advantages of UAVs mentioned above, energy consumption has become a critical issue for UAV communication systems as their missions become more diverse. Specifically, due to the size and weight limitations of UAVs, their limited onboard energy significantly restricts their ability to perform long-duration flight missions, as various flight states of UAVs incur additional propulsion energy consumption. Existing research on UAV communication systems only considers simple energy consumption constraints, such as Z. Liu, R. Sengupta, and A. Kurzhanskiy, “A power consumption model for multi-rotor small unmanned aircraft systems,” in 2017 International Conference on Unmanned Aircraft Systems (ICUAS), 2017, pp. 310–315 and D. Yang, Q. Wu, Y. Zeng, and R. Zhang, “Energy tradeoff in ground-to-UAV communication via trajectory design,” IEEE Transactions on Vehicular Technology, vol. 67, no.7, pp. 6721–6726, 2018. Neither of these can effectively describe the propulsion energy consumption of UAVs in real-world environments. Therefore, establishing a more accurate and universal propulsion energy consumption model for UAVs to quantify their energy consumption is crucial for the energy-efficient design of UAV communication systems. Existing research has introduced various UAV propulsion energy consumption models from two-dimensional or three-dimensional perspectives and proposed corresponding energy-saving designs for UAV communication systems. For example, in the paper X. Xiong, C. Sun, W. Ni, and X. Wang, “Three-dimensional trajectory design for unmanned aerial vehicle-based secure and energy-efficient data collection,” IEEE Trans. Veh. Technol., 2022, the authors derived a two-dimensional UAV propulsion energy consumption model by considering the UAV's flight speed and optimized its trajectory design in a UAV communication system to minimize the total energy consumption of the UAV while satisfying communication constraints.In the paper X. Xiong, C. Sun, W. Ni, and X. Wang, “Three-dimensional trajectory design for unmanned aerial vehicle-based secure and energy-efficient data collection,” IEEE Trans. Veh. Technol., 2022, the authors derived a three-dimensional propulsion energy consumption model for UAVs using UAV flight dynamics. This model involves parameters such as the UAV's speed, acceleration, and fuselage plane pitch angle. Furthermore, it optimizes the UAV's trajectory and communication resource allocation under the consideration of data eavesdropping scenarios, thus achieving an energy-efficient design.

[0004] When considering real-world scenarios, the impact of wind on UAV flight cannot be ignored. Whether flying or hovering, UAVs must withstand the fluid resistance caused by wind, resulting in additional propulsion energy consumption. Therefore, wind interference poses a real challenge to the energy efficiency design of UAV communication systems, especially for rotorcraft UAVs which are more sensitive to wind. Considering the role of wind, as in the paper G. Nachmani, “Minimum-energy flight paths for UAVs using mesoscale wind forecasts and approximate dynamic programming,” Master's thesis, Monterey California Naval Postgraduate School, Tech. Rep., 2007, the authors propose an energy-efficient flight path design for UAVs that utilizes wind. In the paper Y. Zhang, J. Lyu, and L. Fu, “Energy-efficient trajectory design for UAV-aided maritime datacollection in wind,” IEEE Trans. Wireless Commun., vol. 21, no. 12, pp. 10871–10886, 2022, the authors describe the impact of wind on UAVs by considering the horizontal angular deviation between the UAV's horizontal flight direction and the wind direction, and minimize the propulsion energy consumption of the UAV under given data acquisition constraints by optimizing the UAV's horizontal trajectory.

[0005] Therefore, the role of wind, as a key factor affecting the flight status and propulsion energy consumption of UAVs, cannot be ignored in the energy efficiency design of UAV communication systems. However, existing designs for UAV communication systems either fail to consider the role of wind or do not accurately describe its impact on the UAV's flight status and propulsion energy consumption. Furthermore, actual wind conditions are usually random, thus limiting the practical significance of traditional energy efficiency designs.

[0006] This invention will provide corresponding solutions to these problems. Summary of the Invention

[0007] This invention proposes an energy efficiency design method for unmanned aerial vehicle (UAV) communication systems that considers wind interference. The method includes a three-dimensional generalized energy propulsion model (GPECM) derived considering the effects of wind on the UAV, and an energy efficiency design algorithm for UAV communication systems based on this model. This aims to address the aforementioned impact of wind on UAV energy consumption and the energy efficiency design issues of their communication systems.

[0008] This invention considers an energy-efficient design scheme for a drone communication system that takes into account wind interference, including a system operated by a single drone under wind interference. A drone communication system for ground users (GUs) with fixed locations, where the two-dimensional coordinates of each user are defined as follows: The system includes a discretized mobility model for the UAV and a probabilistic channel model between the UAV and ground users.

[0009] The discretized mobility model of the UAV considers the UAV's flight cycle as follows: Furthermore, considering time discretization, the UAV flight cycle is discretized into N time slots, each time slot being represented as... Therefore, this invention considers the position of the UAV in any time slot n within its flight cycle as... ,in and These represent the horizontal position and flight altitude of the UAV in time slot n, respectively. This represents the transpose operation on a matrix or vector. Therefore, the starting and ending positions of the UAV in a given flight cycle are represented as follows: and The distance between the UAV and GUk in any time slot n is expressed as: ,in Let represent the Euclidean norm. This invention considers the discretized flight velocity and acceleration of the UAV in any time slot n as follows: and This invention considers that the UAV is limited by the maximum flight distance in the horizontal and vertical directions within a given time slot, which are respectively expressed as: and ,in and These represent the maximum flight speeds of the drone in the horizontal and vertical directions, respectively. This invention considers the maximum and minimum flight altitudes of the drone as follows: and .

[0010] The Loss Channel Probability (LoS) of the UAV and GUk in any time slot n can be expressed as: ,in and These are all constant parameters. This represents the elevation angle between the UAV and GUk within the corresponding time slot, specifically expressed as... Therefore, the NLoS channel probability of the UAV and GUk in any time slot n can be expressed as: This invention considers path loss in free space and shadowing effects in communication transmission. The channel gains of the UAV and GUk in any time slot n under LoS and NLoS states are expressed as follows: and ,in This represents the signal power attenuation coefficient per unit distance (1 m). This represents the additional signal attenuation factor caused by the more complex electromagnetic propagation environment in the NLoS state. and Let represent the path loss exponents of the signal in free space under the LoS and NLoS states, respectively. Therefore, the achievable code rate of GUk in any time slot n under the LoS and NLoS states are respectively expressed as: and ,in , This refers to the signal transmission power of the drone. The power of additive white Gaussian noise, This represents the signal-to-noise ratio loss due to actual modulation and coding losses. Therefore, under the probabilistic Loss channel model, the expected achievable code rate of GUk in any time slot n is expressed as: .in It represents the mathematical expectation.

[0011] The technical solution adopted in this invention includes the following steps:

[0012] S1. This invention considers a UAV communication system where the UAV flies in three-dimensional space and is affected by wind in the horizontal direction, and the wind is represented by a vector in the horizontal direction. Specifically as follows:

[0013]

[0014] In the formula, Indicates the corresponding reference height The average wind speed below, This represents the wind direction angle, and both the average wind speed and the direction angle can follow some random distribution. This indicates the altitude at which the wind being measured is located. Let be an exponential parameter representing the variation of average wind speed with altitude, and defined here. This indicates the average wind speed.

[0015] S2, such as Figure 1 As shown, assuming that the air resistance and wind resistance experienced by the drone during flight are independent, and ignoring the viscosity effect of the drone flying in a fluid, according to the definition of fluid mechanics, the air resistance of the drone during flight is... And the drag caused by wind on drones The resultant force can be expressed as

[0016]

[0017] in and This indicates air density and the equivalent horizontal area of ​​the drone's fuselage. According to... Figure 1 Based on the force analysis and Newton's second law, the lift of the drone can be expressed as:

[0018]

[0019] The propulsion power of a rotary-wing UAV can be expressed as: ,in and Let represent the rotor solidity and fuselage disk area of ​​the UAV, respectively, and let the UAV's torque coefficient satisfy . ,in, The drag coefficient of the wing is represented by the following specific parameters and related equations: ① Rotor speed normalized to flight speed ② Fuselage lift coefficient ③ Average induced flight speed ,in ④ Fuselage gravity coefficient , among which and Indicates fuselage mass and gravitational acceleration; ⑤ Fuselage drag ratio ⑥ ,in Let represent the lift coefficient of the UAV. Based on the above equation and the discretized movement model of the UAV described above, the propulsion power of the UAV in any time slot n, i.e., GPECM, can be expressed as:

[0020]

[0021] in This indicates the power required for the drone's propellers to rotate and support its flight. This represents the induced power of the drone in hovering mode. The last two terms in the formula represent the climb power of the drone during vertical flight and the power to overcome wind and air resistance, respectively. This indicates the angle of climb of the drone in the vertical direction.

[0022] S3. In an unmanned aerial vehicle (UAV) communication system, assume that in any time slot n, the UAV can only choose to provide communication services to one user, and define a user selection variable. In any time slot n, if GU k is provided with communication services by a drone, then ,otherwise It is also assumed that the location information and probability Loss (LoS) channel information of all GUs are known in advance to the UAV, and that the UAV can estimate the channel state information in real time during flight. An auxiliary variable is introduced. And maximize the energy efficiency of the drone by solving the following optimization problem:

[0023] (P1)

[0024] (C1)

[0025] (C2)

[0026] (C3)

[0027] (C4)

[0028] (C5)

[0029] (C6)

[0030] Where (C1) represents the starting position constraint, (C2) represents the fairness constraint of the UAV to each user's communication, (C3) and (C4) are the corresponding flight constraints, and (C5) and (C6) are the user selection constraints.

[0031] S4, such as Figure 2 As shown, since the optimization problem (P1) in S3 lacks specific wind information, this invention proposes an online design-online-offline (OBOA) scheme based on offline design to maximize the energy efficiency of the UAV communication system. In the offline phase, the following is defined: And introduce auxiliary variables , , , And these auxiliary variables satisfy the following in any time slot n: , , , Therefore, using Taylor expansion and successive convex approximation methods, the following three inequality constraints can be obtained:

[0032]

[0033]

[0034]

[0035]

[0036] in ,and , , , as well as Let be the local iteration variable of the corresponding variable in the t-th iteration of the successive convex approximation iterations. Since Therefore, we can obtain the upper bound of the GPECM expression within any time slot n:

[0037]

[0038] S5. The lower bound of the user's expected communication reachability can be obtained through the following inequality:

[0039]

[0040] And introduce auxiliary variables and satisfy Therefore, as Figure 2 As shown, the corresponding offline system energy efficiency design is obtained by solving the following optimization problem in the offline stage:

[0041] (P2)

[0042] (C1)

[0043] (C2)

[0044] (C3)

[0045] (C4)

[0046] (C5)

[0047] (C6)

[0048] (C7)

[0049] (C8)

[0050]

[0051] (C9)

[0052] (C10)

[0053] (C11)

[0054] in, , (P2) The denominator of the objective function and the constraints in (C7) This represents S random samples generated using known information about wind distribution. The average of the samples is calculated within each time slot n, which can be specifically expressed as follows: Next, S random samples are obtained using the known distribution information. After that, initialize To make these problems fall within the feasible region of the optimization problem (P2), (P2) is solved by dividing it into three subproblems and solving them alternately, and an index is defined for the iteration rounds. Proceed to the next step.

[0055] S6. Assume that in the r-th iteration, given Solve the following optimization problem.

[0056] (P3)

[0057] (C1)

[0058] (C2)

[0059] (C3)

[0060] The obtained optimal solution is expressed as Continue to the next step.

[0061] S7, Given , and record as Solve the following optimization problem.

[0062] (P4)

[0063] (C1)

[0064] (C2)

[0065] (C3)

[0066] (C4)

[0067] (C5)

[0068] (C6)

[0069] (C7)

[0070]

[0071] (C8)

[0072] (C9)

[0073] (C10)

[0074] in Similar to the Taylor approximation in S4, and They respectively satisfy:

[0075]

[0076]

[0077] in , , , Furthermore, the corresponding variable containing the superscript t is also defined as the local iteration variable in the t-th iteration of the successive convex approximation iteration. The optimal solution obtained from problem (P4) is expressed as... Continue to the next step.

[0078] S8, Given , and record as Solve the following optimization problem.

[0079] (P5)

[0080] (C1)

[0081] (C2)

[0082] (C3)

[0083] (C4)

[0084] (C5)

[0085] (C6)

[0086] (C7)

[0087] (C8)

[0088]

[0089] (C9)

[0090] (C10)

[0091] (C10)

[0092] in, It is for variables Taylor expansion and in S7 A similar formal expression. The optimal solution to problem (P5) is expressed as... The results obtained using S6-S8 , , Substituting into S5, the objective function value of (P2) is calculated and denoted as... ,like Convergence to a given algorithm threshold After the iteration ends, the energy efficiency design results obtained in the offline stage are recorded as follows: [Records of offline user-selected schemes and three-dimensional trajectories]. , Otherwise, update. Return to S6 and execute S6-S8.

[0093] S9. After the offline phase is completed, it is assumed that during the actual flight of the UAV, the wind speed and direction within the current time slot can be measured by the onboard sensors in all time slots. For example... Figure 2As shown, this invention considers the user selection scheme and three-dimensional trajectory obtained in the offline phase, as well as the real-time wind information obtained from measurements. The UAV continuously and dynamically adjusts its flight speed within each time slot n to save propulsion energy. In the online phase, for time slots 1 to N-1, the reference speed for any time slot n is initialized to the flight speed obtained in the offline phase, i.e. Furthermore, the initial and final positions are the same as those in the offline scheme.

[0094] S10. The UAV uses onboard sensors to measure real-time wind information within the current time slot n, denoted as... And based on this information, optimize the following issues:

[0095] (P5)

[0096] (C1)

[0097] (C2)

[0098] (C3)

[0099] (C4)

[0100] (C5)

[0101]

[0102] (C6)

[0103] (C7)

[0104] (C8)

[0105] in, This indicates the current actual position of the UAV, which is also the target position obtained in the previous time slot based on the above optimization problem. and This represents the dynamic adjustment tolerance of the drone's actual flight speed and trajectory compared to its offline speed, referring to constraint (C2).

[0106] Constraints (C3 and C4) indicate that the distance from the UAV's flight position to the endpoint within each time slot n must be less than the distance from the reference position to the endpoint obtained during the offline phase. This constraint ensures that the UAV's flight speed can be reasonably allocated within the global time slots while adjusting its flight direction. The optimized speed within the corresponding time slot n is denoted as... .

[0107] S11, The drone updates and obtains the target position for the next time slot n+1. .like Then return to S10 to continue updating the speed and target position for the next time slot.

[0108] S12. During the online phase, while the drone dynamically adjusts its speed, it still follows the user selection strategy obtained during the offline phase. Choose to provide communication services to different users within the corresponding time slot.

[0109] The innovation of this invention can be summarized as follows: First, this invention considers an energy-efficient design method for a UAV communication system that takes wind interference into account. This invention first derives a general propulsion energy consumption model for a rotorcraft UAV under wind interference in three-dimensional space, which facilitates more accurate design of energy-efficient schemes and calculation of energy efficiency. Second, based on this energy consumption model, this invention proposes an online design scheme based on offline design, providing a new energy-efficient design scheme for UAV communication systems that consider wind interference. This design scheme operates in two stages: offline and online. In the offline stage, offline flight trajectories and user selection strategies are optimized based on statistical wind information. In the online stage, flight speed is dynamically adjusted based on the trajectory dynamics from the offline stage and the measured real-time wind information, further reducing the impact of random wind and saving UAV propulsion energy to improve system energy efficiency. Attached Figure Description

[0110] Figure 1 Schematic diagram of force analysis of UAV in three-dimensional space

[0111] Figure 2 Example diagram of OBOA solution

[0112] Figure 3 Comparison of offline trajectories of drones based on different average wind directions

[0113] Figure 4 Comparison of offline trajectories of drones based on different average wind speeds

[0114] Figure 5 Comparison of system energy efficiency based on offline solutions with different average wind directions

[0115] Figure 6 Comparison of system energy efficiency based on offline solutions with different average wind speeds

[0116] Figure 7 Comparison chart of online and offline drone trajectories

[0117] Figure 8 Comparison of system energy efficiency based on variance of wind direction changes

[0118] Figure 9Comparison of system energy efficiency based on variance of wind speed variation Detailed Implementation

[0119] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and specific simulation examples.

[0120] The specific parameter settings are as follows:

[0121] This invention considers four Gateway Units (GUs) receiving communication services from a UAV. The UAV's flight cycle is 150 s, and the length of a single time slot is 1 s. The UAV's starting and ending positions are (0, 500, 100) m and (1000, 500, 100) m, respectively. The UAV's maximum horizontal and vertical flight speeds are 40 m / s and 20 m / s, respectively. The UAV's maximum and minimum flight altitudes are 300 m and 50 m, respectively. The horizontal positions of the four GUs are (100, 300) m, (500, 800) m, (500, 200) m, and (900, 600) m, respectively. The four constant parameters of the probabilistic Loss-of-Stake (LoS) channel model are also considered. The values ​​are -1, 0.05, 0.1, and 0.9 respectively; the relevant communication parameters are set to... -60 dB, -20 dB, 8.2 dB, 2.5, 5. Communication bandwidth 1MHz, -110 dBm, 0.1 W; the drone's mass and gravitational acceleration are 2 kg and 9.8 m / s², respectively. 2 The drone's fuselage parameters are set as follows: 1.225 kg / m 3 , 0.79 m 2 , 0.1, 0.012, 0.01 m 2 , 0.3, 0.13; the reference height for wind speed measurement is 50 m, and the wind speed variation index is 0.5; the speed variation tolerance during the online phase is 20 m / s; the convergence threshold for both alternating optimization and successive convex approximation is 10. -3 .

[0122] In particular, during the simulation, this invention chooses to use the Weibull distribution to describe the randomness of wind speed [9], whose probability density function is:

[0123]

[0124] in This is the scale parameter of the distribution; the larger the value, the greater the average wind speed (distribution mean). The smaller the value of is, the greater the variance of wind speed variation. This invention chooses the Von-Mises distribution to describe the randomness of wind direction

[10] , whose probability density function is .

[0125]

[0126] in This represents the average wind direction angle (distribution mean). The variance of wind direction is a parameter representing the central tendency of the wind. The larger the value, the smaller the range of wind direction variation and the smaller the variance. This represents the zeroth-order Bessel function. In the simulation results, this invention marks the results obtained in the offline stage as "offline" and uses the baseline scheme, i.e., a windless environment. The baseline scheme is labeled as windless, and the energy efficiency index is abbreviated as EE.

[0127] Based on the above parameter settings, the specific steps of this simulation are as follows:

[0128] S1. This invention considers a UAV communication system where the UAV flies in three-dimensional space and is affected by wind in the horizontal direction, and the wind is represented by a vector in the horizontal direction. Specifically as follows:

[0129]

[0130] In the formula, Indicates the corresponding reference height The average wind speed below, This represents the wind direction angle, and both the average wind speed and the direction angle can follow some random distribution. This indicates the altitude at which the wind being measured is located. Let be an exponential parameter representing the variation of average wind speed with altitude, and defined here. This indicates the average wind speed.

[0131] S2, such as Figure 1 As shown, assuming that the air resistance and wind resistance experienced by the drone during flight are independent, and ignoring the viscosity effect of the drone flying in a fluid, according to the definition of fluid mechanics, the air resistance of the drone during flight is... And the drag caused by wind on drones The resultant force can be expressed as

[0132]

[0133] in and This indicates air density and the equivalent horizontal area of ​​the drone's fuselage. According to... Figure 1 Based on the force analysis and Newton's second law, the lift of the drone can be expressed as:

[0134]

[0135] The propulsion power of a rotary-wing UAV can be expressed as: ,in and Let represent the rotor solidity and fuselage disk area of ​​the UAV, respectively, and let the UAV's torque coefficient satisfy . ,in, The drag coefficient of the wing is represented by the following specific parameters and related equations: ① Rotor speed normalized to flight speed ② Fuselage lift coefficient ③ Average induced flight speed ,in ④ Fuselage gravity coefficient , among which and Indicates fuselage mass and gravitational acceleration; ⑤ Fuselage drag ratio ⑥ ,in Let represent the lift coefficient of the UAV. Based on the above equation and the discretized movement model of the UAV described above, the propulsion power of the UAV in any time slot n, i.e., GPECM, can be expressed as:

[0136]

[0137] in This indicates the power required for the drone's propellers to rotate and support its flight. This represents the induced power of the drone in hovering mode. The last two terms in the formula represent the climb power of the drone during vertical flight and the power to overcome wind and air resistance, respectively. This indicates the angle of climb of the drone in the vertical direction.

[0138] S3. In an unmanned aerial vehicle (UAV) communication system, assume that in any time slot n, the UAV can only choose to provide communication services to one user, and define a user selection variable. In any time slot n, if GU k is provided with communication services by a drone, then ,otherwise It is also assumed that the location information and probability Loss (LoS) channel information of all GUs are known in advance to the UAV, and that the UAV can estimate the channel state information in real time during flight. An auxiliary variable is introduced. And maximize the energy efficiency of the drone by solving the following optimization problem:

[0139] (P1)

[0140] (C1)

[0141] (C2)

[0142] (C3)

[0143] (C4)

[0144] (C5)

[0145] (C6)

[0146] Where (C1) represents the starting position constraint, (C2) represents the fairness constraint of the UAV to each user's communication, (C3) and (C4) are the corresponding flight constraints, and (C5) and (C6) are the user selection constraints.

[0147] S4, such as Figure 2 As shown, the optimization problem (P1) in S3 is difficult to solve due to the lack of specific wind information. Therefore, this invention proposes an online design based on offline design (OBOA) scheme to maximize the energy efficiency of the UAV communication system. In the offline phase, the following is defined: And introduce auxiliary variables , , , And these auxiliary variables satisfy the following in any time slot n: , , , Therefore, using Taylor expansion and successive convex approximation methods, the following three inequality constraints can be obtained:

[0148]

[0149]

[0150]

[0151]

[0152] in ,and , , , as well as Let be the local iteration variable of the corresponding variable in the t-th iteration of the successive convex approximation iterations. Since Therefore, we can obtain the upper bound of the GPECM expression within any time slot n:

[0153]

[0154] S5. The lower bound of the user's expected communication reachability can be obtained through the following inequality:

[0155]

[0156] And introduce auxiliary variables and satisfy Therefore, as Figure 2 As shown, the corresponding offline system energy efficiency design is obtained by solving the following optimization problem in the offline stage:

[0157] (P2)

[0158] (C1)

[0159] (C2)

[0160] (C3)

[0161] (C4)

[0162] (C5)

[0163] (C6)

[0164] (C7)

[0165] (C8)

[0166]

[0167] (C9)

[0168] (C10)

[0169] (C11)

[0170] in, , (P2) The denominator of the objective function and the constraints in (C7) This represents S random samples generated using known information about wind distribution. The average of the samples is calculated within each time slot n, which can be specifically expressed as follows: Next, S random samples are obtained using the known distribution information. After that, initialize To make these problems fall within the feasible region of the optimization problem (P2), (P2) is solved by dividing it into three subproblems and solving them alternately, and an index is defined for the iteration rounds. Proceed to the next step.

[0171] S6. Assume that in the r-th iteration, given Solve the following optimization problem.

[0172] (P3)

[0173] (C1)

[0174] (C2)

[0175] (C3)

[0176] The obtained optimal solution is expressed as Continue to the next step.

[0177] S7, Given , and record as Solve the following optimization problem.

[0178] (P4)

[0179] (C1)

[0180] (C2)

[0181] (C3)

[0182] (C4)

[0183] (C5)

[0184] (C6)

[0185] (C7)

[0186]

[0187] (C8)

[0188] (C9)

[0189] (C10)

[0190] in Similar to the Taylor approximation in S4, and They respectively satisfy:

[0191]

[0192]

[0193] in , , , Furthermore, the corresponding variable containing the superscript t is also defined as the local iteration variable in the t-th iteration of the successive convex approximation iteration. The optimal solution obtained from problem (P4) is expressed as... Continue to the next step.

[0194] S8, Given , and record as Solve the following optimization problem.

[0195] (P5)

[0196] (C1)

[0197] (C2)

[0198] (C3)

[0199] (C4)

[0200] (C5)

[0201] (C6)

[0202] (C7)

[0203] (C8)

[0204]

[0205] (C9)

[0206] (C10)

[0207] (C10)

[0208] in, It is for variables Taylor expansion and in S7 Similar in form. The optimal solution obtained from problem (P5) can be expressed as... The results obtained using S6-S8 , , Substituting into S5, the objective function value of (P2) is calculated and denoted as... ,like Convergence to a given algorithm threshold After the iteration ends, the energy efficiency design results obtained in the offline stage are recorded as follows: [Records of offline user-selected schemes and three-dimensional trajectories]. , Otherwise, update. Return to S6 and execute S6-S8.

[0209] S9. After the offline phase is completed, it is assumed that during the actual flight of the UAV, the wind speed and direction within the current time slot can be measured by the onboard sensors in all time slots. For example... Figure 2 As shown, this invention considers the user selection scheme and three-dimensional trajectory obtained in the offline phase, as well as the real-time wind information obtained from measurements. The UAV continuously and dynamically adjusts its flight speed within each time slot n to save propulsion energy. In the online phase, for time slots 1 to N-1, the reference speed for any time slot n is initialized to the flight speed obtained in the offline phase, i.e. Furthermore, the initial and final positions are the same as those in the offline scheme.

[0210] S10. The UAV uses onboard sensors to measure real-time wind information within the current time slot n, denoted as... And based on this information, optimize the following issues:

[0211] (P5)

[0212] (C1)

[0213] (C2)

[0214] (C3)

[0215] (C4)

[0216] (C5)

[0217]

[0218] (C6)

[0219] (C7)

[0220] (C8)

[0221] in, This indicates the current actual position of the UAV, which is also the target position obtained in the previous time slot based on the above optimization problem. and This represents the dynamic adjustment tolerance of the drone's actual flight speed and trajectory compared to its offline speed, referring to constraint (C2).

[0222] Constraints (C3 and C4) indicate that the distance from the UAV's flight position to the endpoint within each time slot n must be less than the distance from the reference position to the endpoint obtained during the offline phase. This constraint ensures that the UAV's flight speed can be reasonably allocated within the global time slots while adjusting its flight direction. The optimized speed within the corresponding time slot n is denoted as... .

[0223] S11, The drone updates and obtains the target position for the next time slot n+1. .like Then return to S10 to continue updating the speed and target position for the next time slot.

[0224] S12. During the online phase, while the drone dynamically adjusts its speed, it still follows the user selection strategy obtained during the offline phase. Choose to provide communication services to different users within the corresponding time slot.

[0225] exist Figure 3 In the text, we present the results of the offline phase of the drone's performance. Comparison of trajectories under different conditions and in windless conditions, with other parameters set as follows: 20, 10, 10. Since the average wind speed increases with altitude, we can see from the graph that the drone flies higher than others affected by wind because it is not affected by wind. Based on our proposed energy-efficient design scheme, when the drone is affected by winds with different average wind directions, it reduces its flight altitude to varying degrees to reduce the impact of wind on propulsion energy consumption and improve system energy efficiency.

[0226] exist Figure 4In the text, we present the results of the offline phase of the drone's performance. Comparison of trajectories under different conditions and in windless conditions, with other parameters set as follows: 20, 270, 10. Same Figure 3 Consistent with our conclusions, we can see that when the average wind speed in the environment increases, based on our proposed energy-efficient design scheme, the average flight altitude of the UAV will also decrease, in order to reduce the impact of wind on its propulsion energy consumption and ensure the energy efficiency of the system.

[0227] exist Figure 5 In the middle, we gave the following: A comparison chart of system energy efficiency (EE) obtained during the offline phase and under windless conditions, with other parameters set as follows: 20, 10, 10. Based on our proposed energy efficiency design scheme, the energy efficiency of the two schemes will differ under the influence of wind from different directions, but the energy efficiency based on our scheme is higher than that obtained under windless conditions.

[0228] exist Figure 6 In the middle, we gave the following: A comparison chart of system energy efficiency (EE) obtained during the offline phase and under windless conditions, with other parameters set as follows: 20, 270, 10. Based on our proposed energy efficiency design scheme, the energy efficiency of both schemes decreases with the increase of average wind speed. However, the energy efficiency of our scheme is higher than that obtained under windless conditions, and the improvement is greater with the increase of average wind speed.

[0229] exist Figure 7 In the figure, we present a trajectory comparison chart of three schemes: OBOA scheme, offline scheme, and windless condition, with other parameters set as follows: 50 m 5, 180, 10, 5. Through dynamic adjustments during the online phase, the trajectory based on OBOA is generally similar to that of the offline phase, but the positions differ in different time slots. In particular, it can be seen that the drone makes timely adjustments to its speed based on real-time wind information using the OBOA-based solution.

[0230] exist Figure 8 In the previous section, we presented the EE (Electrical Engineering) of three solutions: the OBOA (Online Operational Architecture) solution, the offline solution, and the solution under windless conditions. Comparison charts under different conditions, with other parameters set as follows: 100 m 180, 10, 5. We can see that under different wind direction changes with randomness, the system energy efficiency obtained based on our proposed OBOA scheme and the offline scheme is much higher than the energy efficiency obtained under windless conditions. Furthermore, the OBOA scheme is more energy efficient than the offline scheme due to further adjustments in the online phase.

[0231] exist Figure 9 In the previous section, we presented the EE (Electrical Engineering) of three solutions: the OBOA (Online Operational Architecture) solution, the offline solution, and the solution under windless conditions. Comparison charts under different conditions, with other parameters set as follows: 100 m 180, 10, 5. We can see that under different wind speed variations, the system energy efficiency obtained based on our proposed OBOA scheme and the offline scheme is much higher than the energy efficiency obtained under windless conditions. Similarly, the OBOA scheme is more energy efficient than the offline scheme.

Claims

1. A method for energy efficiency design of an unmanned aerial vehicle (UAV) communication system considering wind interference, comprising the following steps: S1. Consider a UAV communication system where the UAV flies in three-dimensional space and is affected by horizontal wind, and the wind is represented by a vector in the horizontal direction. Specifically as follows: In the formula, Indicates the corresponding reference height The average wind speed below, The wind direction angle is represented by the average wind speed, which follows a Weibull distribution, and the wind direction angle follows a Von-Mises distribution. Indicates the altitude at which the wind being measured is located. Let be an exponential parameter representing the variation of average wind speed with altitude, and define . This indicates the average wind speed. S2. Assuming that the air resistance and wind resistance experienced by the drone during flight are independent, and ignoring the viscosity effect of the drone flying in a fluid, according to the definition of fluid mechanics, the air resistance of the drone during flight... And the drag caused by wind on drones The resultant force can be expressed as in and Let the air density and the equivalent horizontal area of ​​the drone's fuselage be represented. Based on the force analysis of the drone and Newton's second law, the lift of the drone can be expressed as: The propulsion power of a rotary-wing UAV can be expressed as: ,in and Let represent the rotor solidity and fuselage disk area of ​​the UAV, respectively, and let the UAV's torque coefficient satisfy . ,in, The wing drag coefficient is represented by the following parameters and related equations: Rotor speed, Normalized flight speed. Fuselage lift coefficient Mean induced flight speed ,in fuselage gravity coefficient , among which and Indicates fuselage mass and gravitational acceleration; fuselage drag ratio ; ,in Let GPECM represent the lift coefficient of the UAV. Based on the above equation and the discretized movement model of the UAV, the propulsion power of the UAV in any time slot n, i.e., GPECM, can be expressed as: in This indicates the power required for the drone's propellers to rotate and support its flight. This represents the induced power of the drone in hovering mode. The last two terms in the formula represent the climb power of the drone during vertical flight and the power to overcome wind and air resistance, respectively. This indicates the angle of climb of the drone in the vertical direction; S3. In an unmanned aerial vehicle (UAV) communication system, assume that in any time slot n, the UAV can only choose to provide communication services to one user, and define a user selection variable. In any time slot n, if ground user GU k is provided with communication services by a drone, then ,otherwise Meanwhile, it is assumed that the location information and probability Loss channel information of all ground user GUs are known in advance to the UAV, and that the UAV can estimate the channel state information in real time during flight, thus introducing auxiliary variables. And maximize the energy efficiency of the drone by solving the following optimization problem: （P1） （C1） （C2） （C3） （C4） （C5） （C6） Where (C1) represents the starting position constraint, (C2) represents the fairness constraint of the UAV's communication with each user, (C3) and (C4) are the corresponding flight constraints, and (C5) and (C6) are the user selection constraints; S4. Propose an online design OBOA scheme based on offline design. In the offline stage, define... And introduce auxiliary variables , , , And these auxiliary variables satisfy the following in any time slot n: , , , In particular, using Taylor expansion and successive convex approximation methods, the following three inequality constraints can be obtained: in ,and , , , as well as Let be the local iteration variable of the corresponding variable in the t-th iteration of the successive convex approximation iteration, since Therefore, we can obtain the upper bound of the GPECM expression within any time slot n: ； S5. The lower bound of the user's expected communication reachability can be obtained through the following inequality: And we introduce auxiliary variables and satisfy The following optimization problem is solved offline to obtain the corresponding offline system energy efficiency design: （P2） （C1） （C2） （C3） （C4） （C5） （C6） （C7） （C8） （C9） （C10） （C11） in, , (P2) The denominator of the objective function and the constraints in (C7) This represents S random samples generated using known information about wind distribution. The average of the samples is calculated within each time slot n, which can be specifically expressed as follows: Next, S random samples are obtained using the known distribution information. After that, initialize To make these problems fall within the feasible region of the optimization problem (P2), (P2) is solved by dividing it into three subproblems and solving them alternately, and an index is defined for the iteration rounds. Proceed to the next step; S6. Assume that in the r-th iteration, given Solve the following optimization problem. （P3） （C1） （C2） （C3） The obtained optimal solution is expressed as Proceed to the next step; S7, Given , and record as Solve the following optimization problem. （P4） （C1） （C2） （C3） （C4） （C5） （C6） （C7） （C8） （C9） （C10） in , and They respectively satisfy: in , , , Furthermore, the corresponding variable containing the superscript t is also defined as the local iteration variable in the t-th iteration of the successive convex approximation iterations. The optimal solution obtained from problem (P4) is expressed as: Proceed to the next step; S8, Given , and record as Solve the following optimization problem. （P5） （C1） （C2） （C3） （C4） （C5） （C6） （C7） （C8） （C9） （C10） （C10） in, It is for variables Taylor expansion and in S7 A similar expression is used to represent the optimal solution to problem (P5) as follows: The results obtained using S6-S8 , , Substituting into S5, the objective function value of (P2) is calculated and denoted as... ,like Convergence to a given algorithm threshold After the iteration ends, the energy efficiency design results obtained in the offline stage are recorded as follows: [Records of offline user-selected schemes and three-dimensional trajectories]. , Otherwise, update. Return to S6 and execute S6-S8; S9. After the offline phase is completed, assuming that the UAV can measure the wind speed and direction in all time slots during actual flight using onboard sensors, and considering the user selection scheme and 3D trajectory obtained in the offline phase, as well as the measured real-time wind information, the UAV continuously and dynamically adjusts its flight speed in each time slot n to save propulsion energy. In the online phase, for time slots 1 to N-1, the base speed for any time slot n is initialized to the flight speed obtained in the offline phase, i.e. Furthermore, the initial and final positions are the same as in the offline scheme; S10. The UAV uses onboard sensors to measure real-time wind information within the current time slot n, denoted as... And based on this information, optimize the following issues: （P5） （C1） （C2） （C3） （C4） （C5） （C6） （C7） （C8） in, This indicates the current actual position of the UAV, which is also the target position obtained in the previous time slot based on the above optimization problem. and This represents the dynamic adjustment tolerance of the drone's actual flight speed and trajectory compared to its offline speed, referring to constraint (C2). Constraints (C3 and C4) indicate that the distance between the UAV's flight position and the destination within each time slot n must be less than the distance between the reference position and the destination obtained during the offline phase. This constraint ensures that the UAV's flight speed can be reasonably allocated within the global time slots while adjusting its flight direction. The optimized speed within the corresponding time slot n is denoted as... ; S11, The drone updates and obtains the target position for the next time slot n+1. ,like If so, return to S10 to continue updating the speed and target position for the next time slot; S12. During the online phase, while the drone dynamically adjusts its speed, it still follows the user selection strategy obtained during the offline phase. Choose to provide communication services to different users within the corresponding time slot.

Images