A multi-time-scale integrated air-ground communication network coverage method

Abstract

The present invention provides an air-ground integrated communication network coverage method based on multiple time scales, which belongs to the field of communication network technology and solves the problem in the prior art of how to maximize the amount of transmitted information per unit energy consumption of unmanned aerial vehicles to achieve optimal energy efficiency. The method comprises initializing relevant parameters, association strategies, resource allocation and flight trajectories; solving sub-problem one of a given flight trajectory and resource allocation to obtain the association strategy; based on the result of sub-problem one, solving sub-problem two of a given flight trajectory and association strategy to obtain the resource allocation; based on the results of sub-problems one and two, solving sub-problem three of a given resource allocation and association strategy to obtain the flight trajectory; and repeatedly iterating the optimization process until the results converge to obtain the final association strategy, resource allocation and flight trajectory. The present invention jointly optimizes user association, resource allocation and the flight trajectory of an aerial base station in a multi-time-scale manner, thereby making the method more practical and greatly improving the application effect.

Description

Technical Field

[0001] The present invention belongs to the technical field of communication networks, and relates to a process of air-ground integrated communication network coverage, in particular to an air-ground integrated communication network coverage method based on multiple time scales. Background Art

[0002] Mobile communication systems have undergone tremendous changes since the 1980s. Demands within the communications industry continue to increase, new applications continue to emerge, and innovation is breaking through the bottlenecks of traditional communications. With the explosive growth of mobile data traffic, current communication networks are no longer sufficient to meet these new challenges. A new communication architecture is needed to break through the limitations of traditional data exchange. 5G networks ushered in the concept of the Internet of Everything, and 6G networks are expected to evolve into a platform for the intelligent Internet of Everything. By integrating satellite and near-Earth networks with terrestrial networks, 6G networks can achieve true global coverage, maintaining high availability and robustness even in the event of natural disasters. Consequently, integrated air-ground communication networks have attracted significant attention in academia.

[0003] Among existing communication network technology solutions, unmanned aerial vehicles (UAVs) serve as economical and convenient airborne platforms, carrying communication base stations to serve ground terminals. Because UAVs operate at low altitudes, their maneuverability and flexibility offer advantages such as greater coverage and faster deployment, making them suitable for a variety of scenarios, particularly in rural areas, emergency deployments, and disaster relief. UAVs can be categorized as rotary-wing and fixed-wing. Rotary-wing UAVs are small, easy to control, and can ascend and descend vertically, but their power is very limited. Compared to rotary-wing UAVs, fixed-wing UAVs can fly at high speeds and have greater payload capacity. While UAVs play a significant role in air-to-ground scenarios, energy-efficient UAV deployment remains a challenge due to their limited onboard energy. Maximizing the amount of information transmitted per unit of energy consumed during a UAV's flight time to achieve optimal energy efficiency is crucial.

[0004] The existing technology provides some communication network coverage methods, but each method has its limitations, as follows:

[0005] 1. Maximizing the system's energy efficiency by jointly optimizing regional scheduling, airborne access point flight trajectories, and transmit power. However, this approach uses a single time scale that is not practical and does not consider the limited backhaul capacity of UAVs.

[0006] 2. Under the conditions of limited cache capacity and flight time, the minimum achievable throughput for each ground user is maximized by jointly optimizing cache layout, UAV resource allocation, and trajectory. The same time scale is used for aerial base station trajectory planning and resource allocation, without considering the fact that 5G base stations have fixed time slot lengths in practice. Furthermore, system energy efficiency is not taken into account during the optimization, making it impossible to support green coverage of the integrated air-ground communication network.

[0007] 3. A communication system optimization method for multi-UAV assisted communication, the main steps of which are: (1) forming a time-division multiple access downlink communication link between multiple UAVs and user communication equipment; (2) establishing a target optimization problem with the optimization goal of maximizing the system communication rate; (3) assuming that the UAV position and transmission power are fixed, determining the optimal user equipment and UAV communication connection allocation using a continuous convex approximation algorithm and an optimization function with a penalty term; (4) adopting a block coordinate optimization method and using a continuous convex approximation algorithm with a locally stable solution to optimize the UAV's hovering position and UAV communication power allocation, thereby realizing the communication construction of multiple UAVs with limited backhaul link capacity; this method does not dynamically deploy UAVs, nor does it consider the optimization problem of system energy efficiency, and has limitations in terms of green coverage of the air-ground integrated communication network.

[0008] 4. A drone-assisted communication method based on resource allocation and trajectory optimization. This method first plans a circular drone flight trajectory based on the distribution of users. Secondly, it uses the block coordinate descent method to decompose the non-convex problem into two sub-problems. Based on the circular trajectory of the drone flight, the beamforming matrix of the drone communication is optimized. Under the fixed beamforming condition, the drone flight speed is optimized and adjusted. Finally, a two-layer alternating optimization algorithm is used to alternately optimize the beamforming and drone flight speed to maximize the total transmission rate of the system. However, this method does not consider drone trajectory planning, nor does it involve the optimization of energy efficiency.

[0009] 5. Existing technologies maximize the achievable throughput of smart vehicles in air-ground scenarios by jointly optimizing user association, base station / UAV transmission power allocation, and UAV trajectory, while considering different limitations and Quality of Service (QoS) constraints; however, this technology does not consider the power consumption of airborne base stations, nor does it address the optimization of energy efficiency in the air-ground network supporting airborne base stations.

[0010] In summary, in the existing technology, how to achieve the association between users and base stations, resource allocation, and trajectory planning of aerial base stations in an air-ground integrated scenario, and improve the system's energy efficiency while meeting user service quality, is a problem that still needs to be further solved. In solving this problem, using the same time scale leads to excessive algorithm complexity and impractical application, which is a bottleneck that needs to be overcome. Summary of the Invention

[0011] Therefore, in order to solve the limitations of the various methods mentioned in the background technology and make a breakthrough in the communication network coverage problem, the present invention proposes a new method; by integrating satellites and near-ground networks into the ground network, an integrated air-ground communication architecture is established, breaking through the traditional data exchange limitations; the present invention aims to maximize the energy efficiency of the entire system, jointly optimizes user associations, resource allocation and flight trajectories of aerial base stations, so that the method is more practical, and the multi-time scale approach can greatly improve the optimization effect of practical applications.

[0012] The present invention adopts the following technical solutions to achieve the purpose:

[0013] A multi-time-scale integrated air-ground communication network coverage method comprises the following steps:

[0014] S1. Initialize system parameters, including ground user parameters, air base station parameters, and other auxiliary parameters;

[0015] S2. Initialize the association strategy, resource allocation, and flight trajectory. During initialization, resource allocation is to distribute resources evenly to each user, and the flight trajectory is for the airborne base station to fly in a straight line along a specific direction starting from the initial position.

[0016] S3. Perform iterative optimization calculations to solve subproblem 1 given a flight trajectory and resource allocation, where the objective function of subproblem 1 is to maximize system efficiency, and obtain an associated strategy;

[0017] S4. Solve subproblem 2 for the given trajectory and association strategy. In solving subproblem 2, the association strategy and resource allocation use different time scales. Therefore, a multi-time-scale unification method is used in the solution process to solve the resource allocation based on the association strategy obtained in subproblem 1.

[0018] S5. Solve subproblem 3 for a given resource allocation and association strategy, and obtain a flight trajectory based on the association strategy obtained from subproblem 1 and the resource allocation obtained from subproblem 2.

[0019] S6. Repeat the iterative optimization calculation process of steps S3 to S5 until the solutions of subproblems 1, 2, and 3 converge, and obtain the final association strategy, resource allocation, and flight trajectory, completing the joint optimization of the air-ground integrated communication network coverage.

[0020] Furthermore, the specific initialization content of step S1 is as follows:

[0021] S1-1. Initialize the number of ground users K, the number of airborne base stations U, the number of time slots N for flight trajectory planning, and the time slot length δ for flight trajectory planning. t , the number of time slots for resource allocation M, the length of the time slot for resource allocation τ t ; and the set of aerial base stations is represented as The set of ground users is represented as

[0022] S1-2. Initialize the location of the ground user, the satellite, and the air base station, including:

[0023] The positions of ground users are randomly distributed, the satellite is at the initial position, the airborne base station is at the initial position, and the airborne base station is backhauled via the satellite; the horizontal coordinate of the kth ground user is The horizontal position of the satellite at the nth time slot is expressed as The satellite height is H, and the fixed height of the u-th aerial base station is Each ground user is associated with only one aerial base station;

[0024] S1-3, initialize other auxiliary parameters in the system, including W US 、W GU 、 Iteration number j = 0, v, κ1, κ2; where W US W represents the bandwidth of the backhaul link transmitting data during time slot n; GU represents the total bandwidth of the access link transmitting data during time slot n; They represent the maximum transmission power of the access link and the backhaul link in the nth time slot respectively; v is the flight speed of the UAV; k1 and k2 are fixed parameters related to air density, UAV weight, wing area, etc.

[0025] Specifically, during initialization, the flight trajectory is that the aerial base station starts from the initial position and flies in a straight line at 45 degrees to the southwest.

[0026] Furthermore, the specific contents of step S3 include:

[0027] S3-1. Specify the association strategy: specify that an air base station can be a maximum of k thProvide communication services to terrestrial users, of which k th ≤K; introduce binary variable a ku [n] indicates whether the kth ground user is connected to the uth air base station in the nth time slot. If connected, then a ku [n]=1, otherwise a ku [n] = 0; the following association strategy is obtained:

[0028]

[0029] a ku [n]∈{0, 1}

[0030] S3-2. Define the objective function of subproblem 1. The objective function of subproblem 1 is the same as the objective functions of subproblems 2 and 3, and is the objective function for maximizing system efficiency, as follows:

[0031]

[0032] S3-3. Solve sub-problem 1, including:

[0033] The optimization problem of subproblem 1 is:

[0034] maxη EE

[0035] stC1:s[0]=i

[0036] C2:

[0037] C3:

[0038] C4:

[0039] C5:

[0040] Relax the binary associated variables into continuous variables and rewrite subproblem 1 as follows:

[0041] maxη EE

[0042] stC1:s[0]=i

[0043] C2:

[0044] C3:

[0045] C4:

[0046] C5:

[0047] Introducing the penalty function F(a ku [n])=a ku [n](a ku [n]-1), the penalty function is a convex function, and subproblem 1 is further rewritten as:

[0048] maxη EE +κF(a ku [n])

[0049] stC1:s[0]=i

[0050] C2:

[0051] C3:

[0052] C4:

[0053] C5:

[0054] Among them, K>0 is a penalty factor, and the objective function η EE +κF(a ku [n]) is the difference of concave functions, that is, η EE -(-k F(a ku [n])); In the j+1th iteration, the F(a ku [n]) is replaced by the first-order Taylor expansion as follows:

[0055]

[0056] And concluded:

[0057]

[0058] According to the above process, subproblem 1 is transformed into the following linear programming:

[0059]

[0060] stC1:s[0]=i

[0061] C2:

[0062] C3:

[0063] C4:

[0064] C5:

[0065] At this point, the CVX tool can be used to solve subproblem one by substituting the given flight trajectory and resource allocation. The solution variable of subproblem one is the association strategy.

[0066] Furthermore, the specific contents of step S4 include:

[0067] S4-1. Specify resource allocation: In the resource allocation of the air base station, the air base station uses a fixed time slot length τ t Update for the unit; use b ku [m] represents the bandwidth ratio allocated to the kth terrestrial user in the access link in time slot m. The value is a discrete value between 0 and 1. Each subcarrier is allocated to only one terrestrial user. If the number of subcarriers is large enough, then b ku [m] is approximately continuous between 0 and 1. Based on the above provisions, the resource allocation is as follows:

[0068]

[0069]

[0070] S4-2. Establish and apply the time scale identity model; in the optimization process of sub-problem 2, the association strategy a ku [n] and resource allocation b ku [m] Different time scales are used, and the model and association strategy are unified according to the time scale under different conditions. ku [n] characteristics, and the resource allocation after time scale homogenization

[0071] S4-3. Solve sub-problem 2, including:

[0072] The optimization problem of subproblem 2 is:

[0073] max η FE

[0074] stC1:s[0]=i

[0075] C2:

[0076] C3:

[0077] C4:

[0078] C5:

[0079] C6:

[0080] Unified time scalea ku[n] and b ku [m], b ku [m] is converted into

[0081] The objective function at this time is as follows:

[0082]

[0083] At this point, the association strategy and the given flight trajectory obtained from subproblem one can be substituted into the CVX tool to solve subproblem two, where the solution variable for subproblem two is resource allocation.

[0084] Furthermore, in step S4-2, the time scale homogenization model and the association strategy a under different conditions are used. ku [n] characteristics, and the resource allocation after time scale homogenization There are five specific situations:

[0085] Case 1: If the median of resource allocation can reflect the characteristics of resource allocation, the time slot with smaller length τ t With a larger time slot length δ t Divide, that is, c = δ t / τ t , we get the corresponding relationship between two different time scale lengths; then we can use each c b ku The value of [m] is obtained by taking its median

[0086] Case 2: If the resource allocation is representative for a certain period of time or the memory is insufficient, the time slot length τ is smaller. t With a larger time slot length δ t Divide, that is, c = δ t / τ t , we get the corresponding relationship between two different time scale lengths; then we can use each c b ku The value of [m] is obtained by taking out some representative values and averaging them.

[0087] Case 3: If the resource allocation time is representative, the time slot with smaller length τ t With a larger time slot length δ t Divide, that is, c = δ t / τ t , we get the corresponding relationship between two different time scale lengths; then we convert b ku The value of [m] is averaged every c times.

[0088] Case 4: If the maximum value of resource allocation can reflect the characteristics of resource allocation, the time slot length τ with a smaller valuet With a larger time slot length δ t Divide, that is, c = δ t / τ t , we get the corresponding relationship between two different time scale lengths; then we convert b ku The value of [m] is obtained by taking the maximum value of each c

[0089] Case 5: If resource allocation is random, the time slot length τ is smaller. t With a larger time slot length δ t Divide, that is, c = δ t / τ t , we get the corresponding relationship between two different time scale lengths; then we can use each c b ku The value of [m] is obtained by randomly taking several values and averaging them.

[0090] Furthermore, the specific contents of step S5 include:

[0091] The optimization problem of subproblem three is:

[0092] maxη EE

[0093] stC1:

[0094] C2:

[0095] C3:s[0]=i

[0096] C4:

[0097] C5:

[0098] C6:

[0099] At this time, the objective function of sub-problem three is as follows:

[0100]

[0101]

[0102] where η EE and is non-convex, let

[0103]

[0104] Introducing slack variables:

[0105] (H u 2+||q u [n]-w k [n]|| 2 )≤Q ku [n]

[0106]

[0107] get:

[0108]

[0109]

[0110] Subproblem 3 can therefore be rewritten as:

[0111] max η EE

[0112] stC1:

[0113] C2:

[0114] C3:s[0]=i

[0115] C4:

[0116] C5:

[0117] C6:

[0118] C7:

[0119] C8:

[0120] Using the first-order Taylor expansion, we can obtain the lower bound of f(x) at a given feasible point xj:

[0121]

[0122] make x=Qku[n], and we get at the j+1th iteration:

[0123]

[0124]

[0125]

[0126] After the above process, sub-problem 3 can be further rewritten as:

[0127] maxη EE

[0128] stC1:

[0129] C2:

[0130] C3:s[0]=i

[0131] C4:

[0132] C5:

[0133] C6:

[0134] C7:

[0135] C8:

[0136] At this point, the association strategy obtained from subproblem one and the resource allocation obtained from subproblem two can be substituted into subproblem three using the CVX tool. The variable for solving subproblem three is the flight trajectory.

[0137] In summary, due to the adoption of this technical solution, the beneficial effects of the present invention are as follows:

[0138] The present invention integrates satellites and near-Earth networks into the ground network, establishing an integrated air-ground communication architecture and breaking through the traditional limitations of data exchange. When jointly optimizing the flight trajectory of aerial base stations, user associations, and resource allocation, the same time scale used in existing research cannot be achieved in practice. Therefore, an innovative optimization method based on multiple time scales is proposed, making the method of the present invention more practical.

[0139] The present invention establishes an integrated air-ground communication network system model, which includes a satellite and multiple airborne base stations. The airborne base stations can be backhauled via satellite. The satellites and multiple airborne base stations in the jointly deployed system provide communication services for ground users in the area. At the same time, the communication power consumption and flight power consumption of the airborne base stations are taken into consideration, with the goal of maximizing the energy efficiency of the entire system. In this way, user association, resource allocation, and the flight trajectory of the airborne base stations are jointly optimized, making the communication network coverage method of the present invention complete and effective, fully considering all situations, and having extremely significant practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0140] Figure 1 A schematic diagram of a communication network architecture involved in the method of the present invention;

[0141] Figure 2 is the total throughput changing with time slots;

[0142] Figure 3 This is a graph showing the relationship between system energy efficiency and the number of ground users. DETAILED DESCRIPTION

[0143] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0144] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0145] A multi-time-scale integrated air-ground communication network coverage method comprises the following steps:

[0146] S1. Initialize the system parameters, including ground user parameters, air base station parameters and other auxiliary parameters;

[0147] S2. Initialize the association strategy, resource allocation, and flight trajectory. During initialization, resource allocation is to evenly distribute resources to each user, and the flight trajectory is for the aerial base station to fly in a straight line in a specific direction starting from the initial position. In this embodiment, during initialization, the flight trajectory is for the aerial base station to fly in a straight line 45 degrees southwest from the initial position.

[0148] S3. Perform iterative optimization calculations to solve subproblem 1 given the flight trajectory and resource allocation. The objective function of subproblem 1 is to maximize system efficiency, and the associated strategy is obtained by solving the problem.

[0149] S4. Solve subproblem 2 for a given trajectory and association strategy. In the solution of subproblem 2, the association strategy and resource allocation use different time scales. Therefore, the multi-time-scale unification method is used in the solution process to solve the resource allocation based on the association strategy obtained in subproblem 1.

[0150] S5. Solve subproblem 3 for the given resource allocation and association strategy. Based on the association strategy obtained from subproblem 1 and the resource allocation obtained from subproblem 2, solve the flight trajectory.

[0151] S6. Repeat the iterative optimization calculation process of steps S3 to S5 until the solutions of subproblems 1, 2, and 3 converge, and obtain the final association strategy, resource allocation, and flight trajectory, completing the joint optimization of the air-ground integrated communication network coverage.

[0152] The schematic diagram of the system architecture for communication network coverage using this method is as follows Figure 1 shown.

[0153] In this embodiment, the specific initialization content of step S1 is as follows:

[0154] S1-1. Initialize the number of ground users K, the number of airborne base stations U, the number of time slots N for flight trajectory planning, and the time slot length δ for flight trajectory planning. t , the number of time slots for resource allocation M, the length of the time slot for resource allocation τ t ; and the set of aerial base stations is represented as The set of ground users is represented as

[0155] S1-2. Initialize the location of the ground user, the satellite, and the air base station, including:

[0156] The positions of ground users are randomly distributed, the satellite is at the initial position, the airborne base station is at the initial position, and the airborne base station is backhauled via the satellite; the horizontal coordinate of the kth ground user is The horizontal position of the satellite at the nth time slot is expressed as The satellite height is H, and the fixed height of the u-th aerial base station is Each ground user is associated with only one aerial base station;

[0157] S1-3, initialize other auxiliary parameters in the system, including W US 、W GU 、 The number of iterations j = 0, v, κ1, κ2. Where W US W represents the bandwidth of the backhaul link transmitting data during time slot n; GU represents the total bandwidth of the access link transmitting data during time slot n; They represent the maximum transmission power of the access link and the backhaul link in the nth time slot respectively; v is the flight speed of the UAV; k1 and k2 are fixed parameters related to air density, UAV weight, wing area, etc.

[0158] In this embodiment, the specific contents of step S3 include:

[0159] S3-1. Specify the association strategy:

[0160] Since the ground users may be very dense in reality, it is impossible for the airborne base station to provide communication services to all users in the coverage area at the same time. Therefore, it is stipulated that an airborne base station can provide communication services to a maximum of k users at the same time. th Provide communication services to terrestrial users, of which k th ≤K; introduce binary variable a ku [n] indicates whether the kth ground user is connected to the uth air base station in the nth time slot. If connected, then a ku [n]=1, otherwise a ku [n] = 0; the following association strategy is obtained:

[0161]

[0162] a ku [n]∈{0, 1}

[0163] S3-2. Define the objective function of subproblem 1. The objective function of subproblem 1 is the same as that of subproblems 2 and 3, and is the objective function for maximizing system efficiency, as follows:

[0164]

[0165] S3-3. Solve sub-problem 1, including:

[0166] The optimization problem of subproblem 1 is:

[0167] maxη EE

[0168] stC1:s[0]=i

[0169] C2:

[0170] C3:

[0171] C4:

[0172] C5:

[0173] Subproblem 1 is non-convex. Relax the binary associated variables into continuous variables and rewrite subproblem 1 as follows:

[0174] max η EE

[0175] stC1:s[0]=i

[0176] C2:

[0177] C3:

[0178] C4:

[0179] C5:

[0180] It is easy to see that the above problem is convex. However, since the associated variable is relaxed to a continuous value between 0 and 1, it cannot be guaranteed to converge to 0 or 1; therefore, a penalty function F(a ku [n])=a ku [n](a ku [n]-1), the penalty function is a convex function, and subproblem 1 is further rewritten as:

[0181] maxη EE +κF(a ku [n])

[0182] stC1:s[0]=i

[0183] C2:

[0184] C3:

[0185] C4:

[0186] C5:

[0187] Among them, k>0 is a penalty factor, and the objective function η EE +κF(a ku [n]) is the difference of concave functions, that is, η EE -(-kF(a ku [n])); Therefore, the problem becomes a DC programming problem (DCP); In order to transform the problem into a convex problem, in the j+1th iteration, the F(a ku [n]) is replaced by the first-order Taylor expansion as follows:

[0188]

[0189] And concluded:

[0190]

[0191] According to the above process, subproblem 1 is transformed into the following linear programming:

[0192]

[0193] stC1:s[0]=i

[0194] C2:

[0195] C3:

[0196] C4:

[0197] C5:

[0198] At this point, the CVX tool can be used to solve subproblem one by substituting the given flight trajectory and resource allocation. The solution variable of subproblem one is the association strategy.

[0199] Furthermore, the specific contents of step S4 include:

[0200] S4-1. Regulations on resource allocation:

[0201] In the resource allocation of the air base station, the air base station uses a fixed time slot length τ t Update for the unit; use b ku [m] represents the bandwidth ratio allocated to the kth terrestrial user in the access link in time slot m. The value is a discrete value between 0 and 1. Each subcarrier is allocated to only one terrestrial user. If the number of subcarriers is large enough, then b ku [m] is approximately continuous between 0 and 1. Based on the above provisions, the resource allocation is as follows:

[0202]

[0203]

[0204] S4-2. Establish and apply the time scale identity model; in the optimization process of sub-problem 2, the association strategy a ku [n] and resource allocation b ku [m] Different time scales are used, and the model and association strategy are unified according to the time scale under different conditions. ku [n] characteristics, and the resource allocation after time scale homogenization

[0205] In practice, resource allocation is updated in the drone's onboard 5G base station, which has a fixed-length update slot. Because the resource allocation update slot length is very small, using the 5G base station's slot length for both flight trajectory and resource allocation optimization would exponentially increase computational complexity. Therefore, this embodiment proposes a timescale homogenization model to address multi-timescale scenarios.

[0206] S4-3. Solve sub-problem 2, including:

[0207] The optimization problem of subproblem 2 is:

[0208] max η FE

[0209] stC1:s[0]=i

[0210] C2:

[0211] C3:

[0212] C4:

[0213] C5:

[0214] C6:

[0215] When calculating energy efficiency, a ku [n] and b ku [m] is a different time scale. Therefore, it is necessary to convert it into a unified time scale, thereby unifying the time scale a ku [n] and b ku [m], specifically b ku [m] is converted into

[0216] It is easy to see that the objective function at this time is as follows:

[0217]

[0218] The objective function in the above equation is a concave function, and the constraints in subproblem 2 are also convex, so this problem is a convex programming problem. At this point, we can use the CVX tool to solve subproblem 2 by substituting the associated strategy and the given flight trajectory from subproblem 1. The solution variable for subproblem 2 is the resource allocation.

[0219] In the above step S4-2, when using multiple time scales to build a model, it is necessary to consider how to resolve the conflicts brought about by different time scales in the optimization problem. This is the challenge brought about by multiple time scales. In sub-problem 2, the association strategy a ku [n] and resource allocation b ku If the time scales used by [m] are different, then time scale conflicts will arise when calculating throughput and energy efficiency. Therefore, this embodiment innovatively proposes a method for unifying multiple time scales, which not only takes into account the characteristics of resource allocation but also reduces the complexity of the algorithm.

[0220] Identification model and association strategy based on time scale under different conditions ku [n] characteristics, and the resource allocation after time scale homogenization There are five specific cases, and the processing method is selected according to the technical background:

[0221] Case 1: If the median of resource allocation can reflect the characteristics of resource allocation, the time slot with smaller length τ t With a larger time slot length δ t Divide, that is, c = δ t / τ t , we get the corresponding relationship between two different time scale lengths; then we can use each c b ku The value of [m] is obtained by taking its median

[0222] Case 2: If the resource allocation is representative for a certain period of time or the memory is insufficient, the time slot length τ is smaller. t With a larger time slot length δ t Divide, that is, c = δ t / τ t , we get the corresponding relationship between two different time scale lengths; then we can use each c b ku The value of [m] is obtained by taking out some representative values and averaging them.

[0223] Case 3: If the resource allocation time is representative, the time slot with smaller length τ t With a larger time slot length δ t Divide, that is, c = δ t / τ t , we get the corresponding relationship between two different time scale lengths; then we convert b ku The value of [m] is averaged every c times.

[0224] Case 4: If the maximum value of resource allocation can reflect the characteristics of resource allocation, the time slot length τ with a smaller value t With a larger time slot length δ t Divide, that is, c = δ t / τ t , we get the corresponding relationship between two different time scale lengths; then we convert b ku The value of [m] is obtained by taking the maximum value of each c

[0225] Case 5: If resource allocation is random, the time slot length τ is smaller. t With a larger time slot length δ t Divide, that is, c = δ t / τ t , we get the corresponding relationship between two different time scale lengths; then we can use each c b ku The value of [m] is obtained by randomly taking several values and averaging them.

[0226] In this embodiment, the specific contents of step S5 include:

[0227] The optimization problem of subproblem three is:

[0228] max η EE

[0229] stC1:

[0230] C2:

[0231] C3:s[0]=i

[0232] C4:

[0233] C5:

[0234] C6:

[0235] At this time, the objective function of sub-problem three is as follows:

[0236]

[0237] where η EE and is non-convex, let

[0238]

[0239] Introducing slack variables:

[0240] (H u 2 +||q u [n]-w k [n]|| 2 )≤Q ku [n]

[0241]

[0242] get:

[0243]

[0244]

[0245] Subproblem 3 can therefore be rewritten as:

[0246] maxη EE

[0247] stC1:

[0248] C2:

[0249] C3:s[0]=i

[0250] C4:

[0251] C5:

[0252] C6:

[0253] C7:

[0254] C8:

[0255] It is easy to see that the problem is still non-convex. Since when x>0 It is convex, and the first-order Taylor expansion is used to expand the given feasible point x. j The lower bound of f(x) is obtained at:

[0256]

[0257] make x=Qku[n], and we get at the j+1th iteration:

[0258]

[0259]

[0260]

[0261] After the above process, sub-problem 3 can be further rewritten as:

[0262] maxη EE

[0263] stC1:

[0264] C2:

[0265] C3:s[0]=i

[0266] C4:

[0267] C5:

[0268] C6:

[0269] C7:

[0270] C8:

[0271] At this point, the association strategy obtained from subproblem one and the resource allocation obtained from subproblem two can be substituted into subproblem three using the CVX tool. The variable for solving subproblem three is the flight trajectory.

[0272] After the above-mentioned method process of this embodiment, the final association strategy, resource allocation and flight trajectory are obtained, and the joint optimization of the air-ground integrated communication network coverage can be completed.

[0273] This example also provides comparative data between this method and a single-time-scale algorithm and a random algorithm. The single-time-scale algorithm uses the same time scale for airborne base station trajectory optimization and resource allocation, with a time slot length of 0.5 seconds. The random algorithm randomly associates airborne base stations with users and plans trajectories while evenly distributing resources. The parameters for the simulation scenario in this example are as follows:

[0274] Table 1 Simulation parameters

[0275] parameter value parameter value Number of flight slots N 60 Flight slot length 0.5s Number of resource allocation time slots M 3000 Resource allocation slot length 0.01s Number of drones 2 Number of users 50 Total bandwidth of access link 1 30Mhz Total bandwidth of access link 2 30Mhz Drone 1 height 1.2km Drone 2 height 1km Total backhaul link bandwidth 50Mhz Satellite transmit power 62dBm Flight speed 50m / s Maximum number of connected users 10 Satellite initial position [-345, 0]km Satellite altitude 2000km Drone 1 initial position [1, 0.7]km Drone 2 initial position [1, 1]km <![CDATA[κ1]]> <![CDATA[9.26×10 -4 kg / m]]> <![CDATA[κ2]]> <![CDATA[2.250kg·m 3 / s 4 ]]>

[0276] The simulation experiment scenario of this embodiment is a network scenario with a simulation area of 1000m×1000m, in which a large number of ground users are randomly distributed in the ground area. In this simulation, the multi-time scale normalization method uses the average of each c of the three cases. Initially, b1(n, k) and b2(n, k) are set to matrices with N rows and K columns, and the values are both 1 / K. UAV 1 flies along the straight line y=x-300 from the initial position, and UAV 2 flies along the straight line y=x from the initial position. The moving speed of each aerial base station is v=50m / s. Perform multiple iterations to obtain the final result. The performance comparison of the multi-time scale algorithm, the single time scale algorithm and the random algorithm is given below:

[0277] Figure 2 The figure shows a line graph of the total system throughput as time slots increase when ground users are randomly distributed. The number of ground users is 50, the number of airborne base stations is 2, and the number of satellites is 1. While the total time length remains constant, it is clear that increasing the number of time slots in the system also increases the system throughput. The figure shows that the multi-time-scale algorithm achieves the highest total system throughput, followed by the single-time-scale algorithm, and the random algorithm achieves the lowest. Furthermore, the multi-time-scale algorithm achieves an 11.7% improvement in total system throughput when the number of time slots is 40 compared to the single-time-scale algorithm. Compared to the random algorithm, the system throughput improves by 301.8% when the number of time slots is 40.

[0278] Because the flight duration of an aerial base station is fixed, the energy consumption of the flight is the same across all algorithms. Furthermore, because the energy consumption of the 5G base station is fixed, the total energy consumption remains the same. The multi-time-scale algorithm achieves higher system throughput and superior performance, given the same energy consumption. This algorithm utilizes different time scales for aerial base station resource allocation and trajectory planning, enabling better service for ground users.

[0279] Figure 3 The figure shows a line graph showing the change in total system throughput with each increase in the number of users when the number of ground users is randomly distributed. The number of aerial base stations is 2, the number of satellites is 1, and the total time length remains unchanged. By varying only the number of ground users to 30, 40, 50, 60, and 70, the impact of changes in the number of ground users within the system on system energy efficiency is determined. As can be seen from the figure, the multi-time-scale algorithm achieves the highest system energy efficiency, followed by the single-time-scale algorithm, and the random algorithm achieves the lowest. Furthermore, compared to the single-time-scale algorithm, the multi-time-scale algorithm improves system energy efficiency by 10.2% when the number of users is 50. Compared to the random algorithm, the system energy efficiency improves by 305.3% when the number of users is 50. The single-time-scale algorithm is not practical, while the multi-time-scale algorithm not only achieves higher performance but can also be applied in practice. This demonstrates the superiority of the multi-time-scale algorithm of the present invention.

Claims

1. A multi-time-scale integrated air-ground communication network coverage method, characterized in that: The steps include: S1. Initialize system parameters, including ground user parameters, air base station parameters, and other auxiliary parameters; S2. Initialize the association strategy, resource allocation, and flight trajectory. During initialization, resource allocation is to evenly distribute resources to each user, and the flight trajectory is for the airborne base station to fly in a straight line along a specific direction starting from the initial position. S3. Perform iterative optimization calculations to solve subproblem 1 given a flight trajectory and resource allocation, where the objective function of subproblem 1 is to maximize system efficiency, and obtain an associated strategy; S4. Solve subproblem 2 for the given trajectory and association strategy. In solving subproblem 2, the association strategy and resource allocation use different time scales. Therefore, a multi-time-scale unification method is used in the solution process to solve the resource allocation based on the association strategy obtained in subproblem 1. S5. Solve subproblem 3 for a given resource allocation and association strategy, and obtain a flight trajectory based on the association strategy obtained from subproblem 1 and the resource allocation obtained from subproblem 2. S6. Repeat the iterative optimization calculation process of steps S3 to S5 until the solutions of subproblems 1, 2, and 3 converge, and the final association strategy, resource allocation, and flight trajectory are obtained, completing the joint optimization of the air-ground integrated communication network coverage. The specific initialization content of step S1 is as follows: S1-1. Initialize the number of ground users , the number of aerial base stations , the number of time slots for flight trajectory planning N , time slot length for flight trajectory planning , the number of time slots for resource allocation M , the time slot length for resource allocation ; And the set of aerial base stations is represented as ={1,2,..., }, the set of ground users is expressed as ={1,2,..., }; S1-2. Initialize the location of the ground user, the satellite, and the air base station, including: The positions of the ground users are randomly distributed, the satellite is at an initial position, the airborne base station is at an initial position, and the airborne base station is backhauled via the satellite; k The horizontal coordinates of a ground user are ; Satellite in the n The horizontal position of a time slot is expressed as ; Satellite altitude is H , No. u The fixed height of the aerial base station is ;Each ground user is associated with only one aerial base station; S1-3, initialize other auxiliary parameters in the system, including 、 、 、 , number of iterations j =0, 、 ; The specific contents of step S3 include: S3-1, specify the association strategy: specify that an air base station can be at most Provide communication services to terrestrial users, including ;Introduce binary variables , indicating the k Is the ground user in the n The time slot is connected to the u Air base station, if connected, then = 1, otherwise = 0; the following association strategy is obtained: S3-2. Define the objective function of subproblem 1. The objective function of subproblem 1 is the same as the objective functions of subproblems 2 and 3, and is the objective function for maximizing system efficiency, as follows: S3-3. Solve sub-problem 1, including: The optimization problem of subproblem 1 is: Relax the binary associated variables into continuous variables and rewrite subproblem 1 as follows: Introducing penalty function F ( ) = ( -1 ), the penalty function is a convex function, and subproblem 1 is further rewritten as: in, is a penalty factor, the objective function F ( ) is the difference of concave functions, that is F ( )); in the j In the +1 iteration, the target F ( ) is replaced by the first-order Taylor expansion as follows: ( ) And concluded: ( ) According to the above process, subproblem 1 is transformed into the following linear programming: At this point, the CVX tool can be used to solve subproblem 1 by substituting the given flight trajectory and resource allocation. The solution variable of subproblem 1 is the association strategy. The specific contents of step S4 include: S4-1. Specify resource allocation: In the resource allocation of the air base station, the air base station uses a fixed time slot length. Update for the unit; use Indicates time slot The internal access link is The bandwidth ratio allocated to each terrestrial user is a discrete value between 0 and 1. Each subcarrier is only allocated to one terrestrial user. If the number of subcarriers is large enough, is approximately continuous between 0 and 1; based on the above provisions, the resource allocation is as follows: S4-2. Establish and apply the time scale identity model; in the optimization process of sub-problem 2, the association strategy and resource allocation Different time scales are used, and the model and association strategy are unified according to the time scale under different conditions. The characteristics of the time scale are obtained after the resource allocation is unified ; S4-3. Solve sub-problem 2, including: The optimization problem of subproblem 2 is: Unified time scale and ,Will As in step S4-2, the method is converted into ; The objective function at this time is as follows: At this point, we can use the CVX tool to solve subproblem 2 by substituting the associated strategy and given flight trajectory obtained from subproblem 1. The variable to be solved for subproblem 2 is resource allocation. In step S4-2, the time scale is unified according to different conditions and the correlation strategy is The characteristics of the time scale are obtained after the resource allocation is unified , specifically divided into the following five situations: Case 1: If the median of resource allocation can reflect the characteristics of resource allocation, the time slot with smaller length With a larger time slot length Divide, that is c= , we can get the corresponding relationship between two different time scale lengths; then we can c indivual Take the median value of ; Case 2: If the resource allocation is representative for a certain period of time or the memory is insufficient, the time slot with a smaller length will be With a larger time slot length Divide, that is c= , we can get the corresponding relationship between two different time scale lengths; then we can c indivual The value of the representative value is taken out and then averaged to get ; Case 3: If the resource allocation time is representative, the time slot with the smaller length will be With a larger time slot length Divide, that is c= , we can get the corresponding relationship between two different time scale lengths; then The value of each c Take the average once to get ; Case 4: If the maximum value of resource allocation can reflect the characteristics of resource allocation, the time slot length is smaller. With a larger time slot length Divide, that is c= , we get the corresponding relationship between two different time scale lengths; then The value of each c The best one ; Case 5: If resource allocation is random, the time slot with smaller length will be With a larger time slot length Divide, that is c = , we can get the corresponding relationship between two different time scale lengths; then we can c indivual The value of is obtained by randomly picking several values and then averaging them. .

2. The multi-time-scale integrated air-ground communication network coverage method according to claim 1, characterized in that: During initialization, the flight trajectory is that the air base station starts from the initial position and flies 45 degrees southwest. Fly in a straight line.

3. The multi-time-scale integrated air-ground communication network coverage method according to claim 1, characterized in that: The specific contents of step S5 include: The optimization problem of subproblem three is: At this time, the objective function of sub-problem three is as follows: in and is non-convex, let Introducing slack variables: get: Subproblem 3 can therefore be rewritten as: Using the first-order Taylor expansion, at a given feasible point Obtained The lower bound of : make , , get j At +1 iteration: After the above process, sub-problem 3 can be further rewritten as: At this point, the association strategy obtained from subproblem one and the resource allocation obtained from subproblem two can be substituted into subproblem three using the CVX tool. The variable for solving subproblem three is the flight trajectory.

Images