UAV-assisted green stereo coverage method for ultra-large-scale machine-like scenes

Abstract

The present invention provides a green stereoscopic coverage method in a super-large-scale machine-like scenario assisted by a drone, and belongs to the field of resource allocation optimization for drone downlink communication. The drone and the ground base station respectively use non-orthogonal multiple access to serve users distributed in three dimensions at the edge and center of the cell to achieve wide-area coverage and large-scale connection. In order to overcome the energy consumption bottleneck of the drone, the energy efficiency of the drone is first optimized. Next, the transmission power of the ground base station is minimized under the condition of achieving the same average throughput to reduce system energy consumption. Based on the model, the present invention is a design method for jointly optimizing the flight trajectory and power allocation of the drone, which can achieve wide-area coverage and large-scale connection and improve system energy efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned aerial vehicle air-to-ground communication resource allocation, and relates to a green three-dimensional coverage method in a super-large-scale machine-type scenario assisted by a unmanned aerial vehicle, specifically referring to the use of unmanned aerial vehicle assistance and non-orthogonal multiple access to achieve wide-area coverage and large-scale connection in a cell where users are three-dimensionally distributed, and to achieve green coverage through resource allocation between unmanned aerial vehicles and ground base stations. Background Art

[0002] Ultra-large-scale machine-type communication is one of the classic scenarios of 5G, focusing on information exchange between people and things, and aiming at the interconnection of all things. It requires the network to carry more device connections, requires low latency, high reliability and wide area coverage, and needs to achieve low cost and low power consumption.

[0003] With the popularization of drone technology, drones are gradually being applied to the communication field. Drone base stations are easy to deploy and have low setup costs, which can effectively cope with the problem of flexible demand. Compared with traditional ground base stations, the advantage of using drones as flying base stations is that they can adjust their altitude, avoid obstacles, and increase the possibility of establishing line-of-sight (LoS) communication links with ground users, reduce shadow and blocking effects, and provide high-quality channels. Therefore, drones as aerial base stations can effectively utilize the advantages of flexible mobility and good line-of-sight connections, increase the coverage range, and achieve wide-area coverage of wireless communications. However, since the onboard energy of drones is limited, the energy of drone base stations is mainly used for signal processing and transmission as well as flight propulsion. Therefore, how to reasonably plan the flight trajectory, improve the energy efficiency of drone base stations, and extend the endurance of drones is a problem that needs to be solved.

[0004] In order to solve the problem of scarce spectrum resources and realize spectrum sharing among multiple users, Japan's Do Co Mo company proposed non-orthogonal multiple access technology (NOMA). It uses superposition coding technology at the transmitting end to transmit the information of different users in the same resource block, and uses serial interference cancellation (SIC) technology at the receiving end to decode the signals of different users. In the NOMA scheme, each time-frequency resource block carries the signals of multiple non-orthogonal users. Compared with the traditional orthogonal multiple access scheme, NOMA can realize spectrum sharing among multiple users, improve spectrum efficiency, and effectively increase the number of device access. Combining drones with NOMA can achieve large-scale user access while achieving wide-area coverage, effectively responding to the needs of the Internet of Everything.

[0005] In order to solve the problems of access scale and cost consumption, drone base stations can be used to assist ground base stations based on NOMA to serve the devices at the edge of the cell, thereby achieving high-density wide-area coverage. However, at present, in the large-scale connection of drones to the air-ground communication network based on NOMA, the energy supply of drones is still a bottleneck problem. In order to seek a balance between the transmission rate and energy consumption of drones and extend the flight time of drones, the present invention improves the energy efficiency of drones by optimizing the flight trajectory and transmission power of drone base stations, and minimizes the base station transmission power by optimizing the power allocation of ground base stations while achieving the same average throughput, so as to reduce the energy consumption of the communication system. Summary of the invention

[0006] The present invention provides a method for green stereoscopic coverage in a large-scale machine-like scenario assisted by a drone. In a cell with three-dimensional distribution of users, users in the center of the cell are served by ground base stations, and users at the edge of the cell are served by drone base stations to expand the coverage area. All base stations use NOMA to serve users to increase the number of connected users. The specific scheme is as follows: Figure 1 Based on this model, the present invention provides a solution for jointly optimizing the flight trajectory and power allocation of UAVs, which is committed to improving the energy efficiency of UAVs to break through the energy consumption bottleneck of UAVs. In addition, under the condition of achieving the same average throughput, the transmission power consumption of the base station is reduced through power allocation optimization to achieve energy saving.

[0007] The technical solution adopted by the present invention to solve the technical problem is as follows:

[0008] A method for green stereoscopic coverage in a large-scale machine scene assisted by a drone includes the following steps:

[0009] The first step is to build a system model:

[0010] 1.1) In order to achieve dense coverage connection and avoid deploying ultra-high density ground base stations in this scenario, drone base stations are used to serve users at the edge of the cell. The hexagonal area covered by the equivalent four ground base stations and two drone base stations can be regarded as a whole, which is equivalent to covering four cells. Assume that the drone base station, ground base station and user all have single antennas, and the ground base station and user are stationary. Due to the flexibility of drones and good channels, drone base stations are used to serve users at the edge of the cell, and base stations serve users in the center of the cell. Both drone base stations and ground base stations use NOMA. In order to ensure the effect of SIC, each NOMA pair consists of two users, sharing the same resource block with a bandwidth of B. Different NOMA pairs are allocated resource blocks of different frequencies. Assume that the sets of drone base stations and ground base stations are u∈{1,2} and w∈{1,2,3,4} respectively, and the set of NOMA pairs they serve is expressed as That is, both the drone base station and the ground base station occupy I resource blocks to serve the same number of 2I users, and the radius of each macro unit is R. In order to ensure that the drone base station and the ground base station serve the same number of users, the service range of the ground base station is

[0011] Since continuous time variables are difficult to handle, they are converted into discrete models. The flight time T of the drone is discretized into N units with δ t =T / N is the time slot with a step length, and the set of moments can be expressed as n∈{0,1,..,N}. For simplicity, assuming that the drone base station flies at a fixed height H, the two-dimensional coordinates of the u-th drone base station can be defined as In addition, the location of the wth ground base station is L w =[x w ,y w ,z w ],user The coordinates of Where l∈{U,G} indicates whether the user is served by a drone base station or a ground base station, m indicates the base station number, and j∈{1,2} indicates the user in the NOMA pair. express The two-dimensional position of .

[0012] 1.2) In the system model represents a matrix of dimension N×M; ||a|| represents the bi-norm of vector a; It means that the mean is μ and the variance is σ 2 The complex Gaussian distribution of .

[0013] The uth drone base station to the user The channel is a LoS channel, and the channel at the nth moment can be expressed as:

[0014]

[0015] where γ U is the channel gain at 1m. The channel obeys Rayleigh fading and can be expressed as

[0016]

[0017] in γ G is the channel gain at 1 m, and α represents the channel loss exponent.

[0018] The interference between NOMA pairs is eliminated by using different resource blocks. In one time slot, The signal received from the u-th drone base station is:

[0019]

[0020] in, It is given by the u-th drone base station in the n-th time slot The power distribution, Indicates the unit power transmission signal, represents additive white Gaussian noise.

[0021] The signal received from the w-th ground base station can be expressed as:

[0022]

[0023] Where P Gt is the maximum transmission power of the ground base station, yes The power allocation factor of the message.

[0024] Since the energy supply for UAV communication is a major bottleneck issue, the energy efficiency of the UAV is first optimized, and then the transmission power of the ground base station is minimized while achieving the same throughput, thereby reducing the overall energy consumption.

[0025] In the second step, we build a drone base station energy efficiency optimization model based on the first step, determine the objective function, and list the optimization problems:

[0026] Without loss of generality, we assume that the first user in a NOMA pair, i.e. is allocated more transmission power, and its message is decoded first. According to the received signal of (3) and the decoding characteristics of NOMA, In decoding When other signals are regarded as interference, the achievable rate at the nth moment can be expressed as:

[0027]

[0028] Conditions should be met and Where P Ut To ensure that the UAV transmit power is limited. It can be decoded at both users in a NOMA pair, and the achievable rate needs to satisfy:

[0029]

[0030] decoding hour, First, the decoded signal is eliminated by SIC. The decoding rate can be expressed as:

[0031]

[0032] In addition, since fixed-wing UAVs have a longer flight time, they are used as air base stations to serve users. Since the UAV flies periodically in the air, its propulsion power limit can be expressed as:

[0033]

[0034] in and represents the two-dimensional velocity and acceleration of the u-th UAV at the nth moment, c1 and c2 are parameters related to the UAV itself, and g represents the gravitational acceleration.

[0035] According to (6), (7) and (8), the energy efficiency of the UAV during its flight cycle can be expressed as:

[0036]

[0037] Where P Uc The circuit consumes a fixed amount of power.

[0038] In order to maximize the energy efficiency under the constraints of the actual UAV motion model and user service quality, the UAV flight trajectory Q u , speed V u , acceleration A u and power distribution P u After joint optimization, the optimization problem (10) can be described as:

[0039]

[0040]

[0041]

[0042]

[0043]

[0044]

[0045]

[0046]

[0047] v min ≤||v u [n]||≤v max , (10i)

[0048] ||a u [n]||≤amax , (10j)

[0049] Where R th represents the user service quality threshold, d min is the minimum distance to avoid collision between drones, v min is the minimum speed of the fixed-wing UAV to maintain the analysis, v max and a max are the maximum speed and acceleration of the UAV. (10b) is the user service quality limit, (10e) avoids collision between two UAVs, (10f) and (10g) satisfy the physical model of UAV flight, and (10h) shows that the UAV is in periodic motion.

[0050] The third step is to design an algorithm to solve the drone optimization problem:

[0051] The optimization problem represented by (10) is a non-convex optimization problem with multiple variables. Therefore, the block coordinate descent (BCD) technique, which divides the complex problem into subproblems based on coordinates, can effectively solve the problem and converge to the local optimum.

[0052] 3.1) First, given the power allocation, the trajectory of the UAV is optimized. The trajectory optimization problem can be formulated as:

[0053]

[0054] st(10b),(10e)-(10j),

[0055] in is the optimal value obtained from the last power optimization. (11) is a non-convex fraction. In order to effectively process it, it is first converted into a form with a concave denominator and a convex numerator, and then solved by the Dinkelbach method. First, By converting to concave through the first-order Taylor expansion, we can get:

[0056]

[0057] in is the trajectory obtained from the last optimization, It can be expressed as:

[0058]

[0059] Then, for a given transmit power, It can be expressed as:

[0060]

[0061] Similarly, for a given transmit power It can be expressed as:

[0062]

[0063] in It can be expressed as:

[0064]

[0065] Then, by introducing the variable V u [n]≤‖v u [n]‖, the denominator can be transformed into:

[0066]

[0067] The additional variables introduced can be The first-order Taylor expansion of is approximately transformed into:

[0068]

[0069] In addition, the limit (10e) is still non-convex, which can be transformed into

[0070]

[0071] in

[0072] According to the above simplification, the original trajectory optimization problem (11) is transformed into a concave-convex fractional programming problem with convex set constraints. This problem can be solved by the Dinkelbach method and can be modified as follows:

[0073]

[0074]

[0075]

[0076] ‖v u [n]‖≤v max , (20d)

[0077] V u [n]≥v min , (20e)

[0078] (10f),(10g),(10h),(10j),(19),

[0079] where λ r is a constant whose value is updated after each iteration.

[0080] 3.2) Next, the power allocation is optimized according to the UAV trajectory obtained by (20). Since the propulsion power is related to the UAV trajectory and is fixed, the power allocation optimization problem can be expressed as:

[0081]

[0082] st(10b),(10c),(10d).

[0083] in The numerator of problem (21) is non-convex, and it can be transformed using the idea of ​​continuous convex approximation. Given the flight trajectory of the drone, It can be transformed into:

[0084]

[0085] in and They can be expressed as:

[0086]

[0087] in It can be obtained through the last power allocation optimization.

[0088] Therefore, given the flight trajectory satisfy:

[0089]

[0090] Similarly, when the trajectory is fixed It can be expressed as:

[0091]

[0092] Therefore, the power allocation optimization problem (21) can be expressed as:

[0093]

[0094]

[0095]

[0096] (10c),(10d).

[0097] In this way, the original trajectory and power allocation joint optimization problem is transformed into two convex optimization subproblems that can be solved by the algorithm summarized in Algorithm 1. The constant λ in Algorithm 1 is r+1 Updated to: based on previous optimization values:

[0098]

[0099] 3.3) Design of alternating optimization algorithm

[0100] Through the above method, the two sub-problems (20) and (26) can be transformed into convex problems respectively and can be solved using CVX tools. In order to solve problem (10), an algorithm is proposed to alternately optimize the two sub-problems to solve the original problem. The specific steps are as follows:

[0101] 3.3.1) Set the initial value r = 0, the drone trajectory Q u [n] r , UAV speed V u [n] r , power allocation P[n] r ,λ r And the error threshold e

[0102] 3.3.2) Repeat

[0103] 3.3.3) Use the Q obtained from the last optimization u [n] r , V u [n] r , P u [n] r Solve the optimization problem (20) and obtain the optimal flight trajectory Q u [n] * , V u [n] * , A u [n] *

[0104] 3.3.4) Update Q u [n] r+1 =Q u [n] * and V u [n] r+1 =V u [n] *

[0105] 3.3.5) Using Q u [n] r+1 and P u [n] r Solve the optimization problem (26) and get the optimal power allocation P u [n] *

[0106] 3.3.6) Update P u [n] r+1 =P u [n] *and λ r+1

[0107] 3.3.7) r = r + 1

[0108] 3.3.8) until |λ r+1 -λ r |<e

[0109] 3.3.9) Get the optimized energy efficiency λ r , trajectory Q u [n] * , V u [n] * , A u [n] * and power distribution P u [n] *

[0110] During the flight of the drone, due to the user distance switching during the flight of the drone and the uncertainty of the drone trajectory in each iteration, for the sake of simplicity, the user and decoding order of each NOMA pair are given. The decoding order is determined by the user and the fixed point The distance between them determines the

[0111] Step 4: Optimize ground base station power allocation and minimize transmission power:

[0112] 4.1) First, establish the ground base station optimization problem. Since we assume The information is decoded first, so It should be the user with better channel conditions in the NOMA pair, that is So The decoding rate at the two users can be expressed as:

[0113]

[0114] And because Can get In order for both users to successfully decode,

[0115] After eliminating interference through SIC, exist The decoding rate at is:

[0116]

[0117] Under the constraint that the average throughput of users is the same as that of the users served by the drone, the power minimization problem can be expressed as:

[0118]

[0119]

[0120]

[0121] in is the threshold to achieve the average throughput of the drone service users, (30b) indicates that the drone base station and the ground base station achieve the same service quality. In order to transform the non-convex restriction into a convex set, the same method as (22) can be used to obtain:

[0122]

[0123] in right Simplifying, we can get:

[0124]

[0125] According to (31) and (32), the original optimization problem (30) can be transformed into:

[0126]

[0127]

[0128]

[0129]

[0130] In addition, the fixed power consumption of the ground base station circuit is P Gc .

[0131] 4.2) Design iterative algorithms to solve optimization problems

[0132] Through the above method, problem (30) is transformed into a convex problem (33), which can be solved using the CVX tool. In order to solve problem (33), an iterative algorithm is proposed to solve the original problem. The specific steps are as follows:

[0133] 4.2.1) Set the initial value r = 0, the base station power allocation A' w r , optimization target initial value μ r And the error threshold e

[0134] 4.2.2) Repeat

[0135] 4.2.3) Use A' obtained from the last optimization w r Solve the optimization problem (33) and get the optimal power allocation A' w *

[0136] 4.2.4) Update A' w r+1 =A' w * , μ r+1 =μ r

[0137] 4.2.5) r = r + 1

[0138] 4.2.6) until |μ r+1 -μ r |<e

[0139] 4.2.7) Get the optimized transmission power μ r and power distribution A' w *

[0140] The beneficial effects of the present invention are:

[0141] The present invention combines drones with NOMA, uses drones to assist existing communication facilities, and designs the system for energy saving while achieving wide-area coverage and large-scale connection. On the one hand, the drone's cruising time can be extended by optimizing its trajectory, and on the other hand, the energy efficiency of the system can be improved through the assistance of drones. BRIEF DESCRIPTION OF THE DRAWINGS

[0142] Figure 1 Diagram of the NOMA-based drone-assisted large-scale connectivity communication network system.

[0143] Figure 2 Optimizing trajectories and user distribution for large-radius cell drones.

[0144] Figure 3 Optimizing trajectories and user distribution for small radius cell drones.

[0145] Figure 4 The flight trajectories of the UAV in different periods.

[0146] Figure 5 The impact of drone flight cycle on energy efficiency.

[0147] Figure 6 The impact of UAV transmission power on energy efficiency. DETAILED DESCRIPTION

[0148] In order to better understand the above technical solution, the following is a detailed analysis in conjunction with the accompanying drawings and specific implementation methods. The specific steps of this embodiment include:

[0149] The first step is to build a system model:

[0150] 1.1) If Figure 1This is a drone-assisted large-scale access network system model. The drone base station serves the cell edge users. Both the drone base station and the ground base station adopt NOMA. Each NOMA pair consists of two users, sharing the same resource block. In this embodiment, the bandwidth is B = 180kHz, and each drone base station and ground base station occupies I = 200 resource blocks. The radius of each macro unit is R = 2000m, r' = 1485m.

[0151] Discretize the flight time T of the drone into N t is the time slot with a step length. In this embodiment, T = 100s and N = 50. In this embodiment, the drone base station flies at a fixed height H = 1000m. The position of the central ground base station is L1 = [0, 0, 50], and the base station height z w = 50m, user's location Generated randomly by MATLAB in the service area according to uniform distribution, and the height

[0152] 1.2) In the system model, represents a matrix of dimension N×M; ||a|| represents the bi-norm of vector a; It means that the mean is μ and the variance is σ 2 The complex Gaussian distribution of .

[0153] The uth drone base station to the user The channel is a LoS channel, and the channel at the nth moment can be expressed as:

[0154]

[0155] In this embodiment, γ U =(c / (4πf)) 2 , where c = 3*10 8 m / s, carrier frequency f=4.9GHz. The wth ground base station to the user The channel is represented as:

[0156]

[0157] In this embodiment, Randomly generated by Matlab, γ G =(c / (4πf))α is the channel gain at 1m, α=2.6.

[0158] In a time slot, The signal received from the u-th drone base station is:

[0159]

[0160] In this embodiment The noise power is passed Calculate,σ 2 is the noise power spectral density, σ 2 =10 -20.4 W / Hz.

[0161] The signal received from the w-th ground base station can be expressed as:

[0162]

[0163] Since the energy supply for UAV communication is a major bottleneck issue, the energy efficiency of the UAV is first optimized, and then the transmission power of the ground base station is minimized while achieving the same throughput, thereby reducing the overall energy consumption.

[0164] In the second step, we build a drone base station energy efficiency optimization model based on the first step, determine the objective function, and list the optimization problems:

[0165] The transmission rate of drone service users can be expressed by (5)-(7). In this embodiment, the maximum transmission power of the drone base station P Ut =5W. The upper limit of the UAV's propulsion power can be expressed as:

[0166]

[0167] In this embodiment, c1=1.852*10 -3 , c2=4500,g=9.8m / s 2 .

[0168] The energy efficiency of the UAV during its flight cycle can be expressed as:

[0169]

[0170] In this embodiment, P Uc =160W.

[0171] In order to maximize the energy efficiency under the constraints of the actual UAV motion model and user service quality, the UAV flight trajectory Q u , speed V u , acceleration A u and power distribution P u Perform joint optimization, and the resulting optimization problem can be described as

[0172]

[0173]

[0174]

[0175]

[0176]

[0177]

[0178]

[0179]

[0180] v min ≤||v u [n]||≤v max , (10i)

[0181] ||a u [n]||≤a max , (10j)

[0182] In this embodiment, R th =1bit / s / Hz, d min =50m,v min =10m / s,v max =100m / s, a max =5m / s 2 .

[0183] In the third step, the optimization problem is solved by designing an algorithm based on the parameter settings in the first and second steps and the transformation of (11)-(26):

[0184] 3.1) Trajectory Optimization Problem

[0185]

[0186]

[0187]

[0188] ‖v u [n]‖≤v max , (20d)

[0189] V u [n]≥v min , (20e)

[0190]

[0191]

[0192]

[0193] ||a u[n]||≤a max , (10j)

[0194]

[0195] In this embodiment, the initial trajectory of the drone is set to a uniform circular trajectory with a radius of R / 2 and a center of [0,0,0], and the initial speed is calculated based on the initial trajectory and the period T.

[0196] 3.2) Power allocation optimization problem

[0197]

[0198]

[0199]

[0200]

[0201]

[0202] In this embodiment, in order to achieve the user service quality threshold limit of the optimization problem, during power initialization, the power allocated to each group of later-decoded users at each moment is 9 / 2000 of the total power, and the power allocated to earlier-decoded users is 1 / 2000 of the total power.

[0203] 3.3) Design an alternating optimization algorithm. The specific steps are as follows:

[0204] 3.3.1) Set the initial value r = 0, the drone trajectory Q u [n] r , UAV speed V u [n] r , power allocation P[n] r ,λ r And the error threshold e

[0205] 3.3.2) Repeat

[0206] 3.3.3) Use the Q obtained from the last optimization u [n] r , V u [n] r , P u [n] r Solve the optimization problem (20) and obtain the optimal flight trajectory Q u [n] * , V u [n] * , A u [n] *

[0207] 3.3.4) Update Q u [n] r+1 =Q u [n] * and V u [n] r+1 =V u [n] *

[0208] 3.3.5) Using Q u [n] r+1 and P u [n] r Solve the optimization problem (26) and get the optimal power allocation P u [n] *

[0209] 3.3.6) Update P u [n] r+1 =P u [n] * and λ r+1

[0210] 3.3.7) r = r + 1

[0211] 3.3.8) until |λ r+1 -λ r |<e

[0212] 3.3.9) Get the optimized energy efficiency λ r , trajectory Q u [n] * , V u [n] * , A u [n] * and power distribution P u [n] *

[0213] The fixed points that specify the decoding order in this embodiment are That is, the horizontal and vertical coordinate center point of the service area.

[0214] The fourth step is to establish the power minimization problem and design an algorithm to solve it based on the parameter settings in the first step and (28), (29):

[0215] 4.1) The power minimization problem can be expressed as

[0216]

[0217]

[0218]

[0219] In this embodiment It is calculated based on the average throughput of the drone flight cycle, that is,

[0220] According to (31) and (32), problem (30) can be transformed into a convex optimization problem

[0221]

[0222]

[0223]

[0224]

[0225] In this embodiment, the ground base station has a fixed power consumption P Gc =270W.

[0226] 4.2) Design an iterative algorithm to solve the optimization problem. The specific steps are as follows:

[0227] 4.2.1) Set the initial value r = 0, the base station power allocation A' w r , optimization target initial value μ r And the error threshold e

[0228] 4.2.2) Repeat

[0229] 4.2.3) Use A' obtained from the last optimization w r Solve the optimization problem (33) and get the optimal power allocation A' w *

[0230] 4.2.4) Update A' w r+1 =A' w * , μ r+1 =μ r

[0231] 4.2.5) r = r + 1

[0232] 4.2.6) until |μ r+1 -μ r |<e

[0233] 4.2.7) Get the optimized transmission power μ r and power distribution A' w *

[0234] In order to compare with the solution without drone assistance, this embodiment sets a comparison solution that only uses ground base station coverage and other small areas. In addition, the energy efficiency in the embodiments all considers bandwidth. The performance of the comparison scheme in different scenarios will be compared below. It should be noted that, unless otherwise specified, the simulation parameters all use the parameters in the above steps.

[0235] Figure 2 The optimized trajectory and user distribution of the drone-assisted communication scheme in an environment with large cells and not very dense user distribution are shown. At this time, the amount of information received by all users during the flight of two drone base stations for one week is 61.074Gbit, the propulsion consumes 41.16kJ of energy, and the drone base station transmits signals and consumes 33kJ of energy in the circuit. In order to achieve the same average throughput as a single drone base station, the total transmission energy consumption of the four ground base stations in one cycle is 740kJ, and the circuit consumes 108kJ of energy. Therefore, the overall energy efficiency is 199kbit / J. In the comparison scheme, to achieve the same average throughput under the conditions of the same user density and coverage area, 6 ground base stations are required to be covered by NOMA. The total transmission energy consumption in one cycle is 1095.6kJ, and the circuit consumes 162kJ of energy. At this time, the energy efficiency of the comparison scheme is 146kbit / J. The drone-assisted scheme can effectively reduce energy consumption and improve energy efficiency by 36%.

[0236] Figure 3 The optimized trajectory and user distribution of the drone-assisted communication scheme in an environment with small cells and dense user distribution are shown. The cell radius is set to R = 500m, the path loss index is set to α = 2.8, and the drone flight cycle is set to T = 50s. At this time, the amount of information received by all users during the flight of two drone base stations is 35.568Gbit, the propulsion consumes 22.18kJ of energy, and the drone base station transmits signals and consumes 16.5kJ of energy for the circuit. In order to achieve the same average throughput as a single drone base station, the total transmission energy consumption of the four ground base stations in one cycle is 315.6kJ, and the circuit consumes 54kJ of energy. Therefore, the overall energy efficiency is 261kbit / J. In the comparison scheme, to achieve the same average throughput under the conditions of the same user density and coverage area, 6 ground base stations are required to be covered by NOMA. The total transmission energy consumption in one cycle is 466.8kJ, and the circuit consumes 81kJ of energy. At this time, the energy efficiency of the comparison scheme is 195kbit / J. The drone-assisted scheme can effectively reduce energy consumption and improve energy efficiency by 34%.

[0237] Figure 4The optimal trajectory of the drone base station 1 under different cycles is shown. It can be seen from the figure that after the proposed algorithm optimizes the trajectory, the flight trajectory of the drone is concentrated near the center of the service area to effectively improve the user throughput. In addition, the optimized trajectory is very smooth and has a large turning radius, which can effectively avoid the additional propulsion consumption caused by excessive acceleration, reflecting the effectiveness of the trajectory optimization algorithm.

[0238] Figure 5 The effect of the drone flight cycle on energy efficiency is shown. From the results, it can be seen that the optimal flight cycle of the drone is different for cells of different radii. The smaller the radius of the cell, the greater the impact of the flight cycle on energy efficiency. The appropriate flight cycle can be reasonably selected according to the radius of the cell.

[0239] Figure 6 The effects of different trajectory optimization and user pairing methods are shown. Among them, the circular trajectory is the trajectory of flying with the radius with the minimum propulsion power consumption. It can be seen that compared with the circular trajectory with the minimum power consumption and the random user pairing method without considering the distance, the proposed trajectory optimization and user pairing method can effectively improve the energy efficiency of the UAV.

[0240] The above-described embodiments merely express the implementation methods of the present invention, but they cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A drone-assisted green stereoscopic coverage method for ultra-large-scale machine scenes, characterized in that: The following steps are involved: The first step is to build a system model: 1.1) In order to achieve dense coverage and avoid deploying ultra-high-density ground base stations, drone base stations are used to serve users at the edge of the cell; the hexagonal area covered by the equivalent four ground base stations and two drone base stations is regarded as a whole, which is equivalent to covering four cells; it is assumed that the drone base station, the ground base station and the user are all single-antenna, and the ground base station and the user are all stationary; the drone base station and the ground base station serve users at the edge of the cell and the center of the cell respectively; the drone base station and the ground base station both use NOMA. To ensure the SIC effect, each NOMA pair consists of two users, sharing the same resource block with a bandwidth of B, and different NOMA pairs are allocated resource blocks of different frequencies; it is assumed that the sets of drone base stations and ground base stations are u∈{1,2} and w∈{1,2,3,4} respectively, and the set of NOMA pairs they serve is expressed as That is, both the UAV base station and the ground base station occupy I resource blocks to serve the same number of 2I users, and the radius of each macro unit is R; in order to ensure that the UAV base station and the ground base station serve the same number of users, the service range of the ground base station is Discretize the flight time T of the drone into N t =T / N is the time slot with a step length, and the set of moments can be expressed as n∈{0,1,..,N}; Assuming that the drone base station flies at a fixed height H, the two-dimensional coordinates of the u-th drone base station can be defined as q u [n] = [x u [n],y u [n]]; In addition, the location of the wth ground base station is L w =[x w ,y w ,z w ],user The coordinates of Where l∈{U,G} indicates whether the user is served by a drone base station or a ground base station, m indicates the base station number, and j∈{1,2} indicates the user in the NOMA pair. express The two-dimensional position of 1.2) In the system model represents a matrix of dimension N×M; ||a|| represents the bi-norm of vector a; It means that the mean is μ and the variance is σ 2 The complex Gaussian distribution of The uth drone base station to the user The channel is a LoS channel, and the channel at the nth moment can be expressed as: where γ U is the channel gain at 1m; from the wth ground base station to the user The channel is subject to Rayleigh fading, which can be expressed as: in γ G is the channel gain at 1 m, α represents the channel loss exponent; The interference between NOMA pairs is eliminated by using different resource blocks. In one time slot, The signal received from the u-th drone base station is: in, It is given by the u-th drone base station in the n-th time slot The power distribution, Indicates the unit power transmission signal, represents additive white Gaussian noise; The signal received from the w-th ground base station can be expressed as: Where P Gt is the maximum transmission power of the ground base station, yes The power allocation factor of the message; First, the energy efficiency of the UAV is optimized, and then the transmission power of the ground base station is minimized while achieving the same throughput, thereby reducing the overall energy consumption; In the second step, we build a drone base station energy efficiency optimization model based on the first step, determine the objective function, and list the optimization problems: Without loss of generality, assume that the first user in a NOMA pair is is allocated more transmission power, and its message is decoded first; according to the received signal of (3) and the decoding characteristics of NOMA, In decoding When other signals are regarded as interference, the achievable rate at the nth moment can be expressed as: Conditions should be met and Where P Ut It is the UAV transmission power limit; in order to ensure It can be decoded at both users in a NOMA pair, and the achievable rate needs to satisfy: decoding hour, First, the decoded signal is eliminated by SIC. The decoding rate can be expressed as: In addition, fixed-wing UAVs are used as air base stations to serve users. Since the UAVs fly periodically in the air, the upper limit of their propulsion power can be expressed as: in and represents the two-dimensional velocity and acceleration of the u-th UAV at the n-th moment, c1 and c2 are parameters related to the UAV itself, and g represents the gravitational acceleration; According to (6), (7) and (8), the energy efficiency of the UAV during its flight cycle can be expressed as: Among them, P Uc The circuit consumes fixed power; In order to maximize the energy efficiency under the constraints of the actual UAV motion model and user service quality, the UAV flight trajectory Q u , speed V u , acceleration A u and power distribution P u After joint optimization, the optimization problem (10) can be described as: v min ≤||v u [n]||≤v max , (10i) ||a u [n]||≤a max , (10j) Among them, R th represents the user service quality threshold, d min is the minimum distance to avoid collision between drones, v min is the minimum speed of the fixed-wing UAV to maintain the analysis, v max and a max is the maximum speed and acceleration of the UAV; (10b) is the user service quality limit, (10e) avoids collision between two UAVs, (10f) and (10g) satisfy the physical model of UAV flight, and (10h) shows that the UAV is in periodic motion; The third step is to design an algorithm to solve the drone optimization problem: Divide the complex problem shown in (10) into sub-problems, and use the block coordinate descent BCD method to effectively solve the problem and converge to the local optimum; 3.1) First, given the power allocation, the trajectory of the UAV is optimized. The trajectory optimization problem can be formulated as: st(10b),(10e)-(10j), in is the optimal value obtained from the last power optimization; And through transformation, the original trajectory optimization problem (11) is transformed into a concave-convex fractional programming problem (20) with convex set constraints. Specifically: the trajectory optimization problem shown in formula (11) is a non-convex fraction. First, it is transformed into a form with a concave denominator and a convex numerator, and then solved by the Dinkelbach method; First, Transformed into concave form by first-order Taylor expansion, we get: in is the trajectory obtained from the last optimization, It can be expressed as: Then, for a given transmit power, It can be expressed as: Similarly, for a given transmit power It can be expressed as: in It can be expressed as: Then, by introducing the variable V u [n]≤‖v u [n]‖, the denominator can be transformed into: The additional variables introduced can be The first-order Taylor expansion of is approximately transformed into: In addition, the limit (10e) is still non-convex, which can be transformed into in According to the above transformation, the original trajectory optimization problem (11) is transformed into a concave-convex fractional programming problem with convex set constraints. This problem can be solved by the Dinkelbach method, so it is modified to problem (20): ‖v u [n]‖≤v max ,(20d) V u [n]≥v min ,(20e) (10f),(10g),(10h),(10j),(19), where λ r is a constant, and its value is updated after each iteration; 3.2) Next, the power allocation is optimized according to the UAV trajectory obtained in problem (20). Since the propulsion power is related to the UAV trajectory at this time and is fixed, the power allocation optimization problem can be expressed as problem (21): st(10b),(10c),(10d) in Obtained through the last trajectory optimization; The power allocation optimization problem (21) is expressed as problem (26): (10c),(10d) In the above way, the original trajectory and power allocation joint optimization problem is transformed into two convex optimization sub-problems, which are solved by the algorithm in 3.3); the constant λ in algorithm 3.3) r+1 Updated to: based on previous optimization values: 3.3) Design of alternating optimization algorithm Through the above method, subproblems (20) and (26) can be transformed into convex problems and solved using CVX tools. In order to solve problem (10), an algorithm is proposed to alternately optimize the two subproblems to solve the original problem. The specific steps are as follows: 3.3.1) Set the initial value r = 0, the drone trajectory Q u [n] r , UAV speed V u [n] r , power allocation P[n] r ,λ r And the error threshold e 3.3.2) Repeat 3.3.3) Use the Q obtained from the last optimization u [n] r , V u [n] r , P u [n] r Solve the optimization problem (20) and obtain the optimal flight trajectory Q u [n] * , V u [n] * , A u [n] * 3.3.4) Update Q u [n] r+1 =Q u [n] * and V u [n] r+1 =V u [n] * 3.3.5) Using Q u [n] r+1 and P u [n] r Solve the optimization problem (26) and get the optimal power allocation P u [n] * 3.3.6) Update P u [n] r+1 =P u [n] * and λ r+1 3.3.7)r＝r+1 3.3.8) until |λ r+1 -λ r |<e 3.3.9) Get the optimized energy efficiency λ r , trajectory Q u [n] * , V u [n] * , A u [n] * and power distribution P u [n] * During the flight of the drone, the user and decoding order of each NOMA pair are given; the decoding order is determined by the user and the fixed point The distance between them determines; Step 4: Optimize ground base station power allocation and minimize transmission power: 4.1) First, establish the ground base station optimization problem; due to the assumption that The information is decoded first, so It should be the user with better channel conditions in the NOMA pair, that is So The decoding rate at the two users can be expressed as: And because Can get In order for both users to successfully decode, After eliminating interference through SIC, exist The decoding rate at is: Under the constraint that the average throughput of users is the same as that of the users served by the drone, the power minimization problem (30) can be expressed as: in is the threshold to achieve the average throughput of the users served by the drone, and (30b) indicates that the drone base station and the ground base station achieve the same service quality; The original optimization problem (30) can be transformed into a convex problem (33) as follows: In addition, the fixed power consumption of the ground base station is P Gc ; 4.2) Design iterative algorithms to solve optimization problems Through the above method, problem (30) is transformed into a convex problem (33) and solved using the CVX tool. In order to solve problem (33), an iterative algorithm is proposed to solve the original problem. The specific steps are as follows: 4.2.1) Set the initial value r = 0, the base station power allocation A' w r , optimization target initial value μ r And the error threshold e 4.2.2) Repeat 4.2.3) Use A' obtained from the last optimization w r Solve the optimization problem (33) and get the optimal power allocation A' w * 4.2.4) updateA' w r+1 =A' w * ,m r+1 =μ r 4.2.5)r＝r+1 4.2.6) until |μ r+1 -μ r |<e 4.2.7) Get the optimized transmission power μ r and power distribution A' w * .

2. The method for green stereoscopic coverage in a large-scale machine scene assisted by a drone according to claim 1 is characterized in that: In the third step 3.2), the specific steps of transforming the power allocation optimization problem (21) into a convex optimization subproblem are: The numerator of the power allocation optimization problem (21) is non-convex. It can be transformed using the idea of ​​continuous convex approximation. Given the flight trajectory of the UAV, It can be transformed into: in and They can be expressed as: in It can be obtained through the last power allocation optimization; Therefore, given the flight trajectory satisfy: Similarly, when the trajectory is fixed It can be expressed as: Therefore, the power allocation optimization problem (21) can be expressed as: (10c),(10d).

3. The method for green stereoscopic coverage in a large-scale machine scene assisted by a drone according to claim 1 is characterized in that: In the third step 4.1), the specific steps of transforming the original optimization problem (30) into a convex problem (33) are as follows: In order to transform the non-convex restriction into a convex set, the same method as (22) can be used to obtain: in right Simplifying, we can get: According to (31) and (32), the original optimization problem (30) is transformed into a convex problem (33):

Images