A 5G base station site selection optimization model and solution algorithm for drone communication

Abstract

The present invention discloses a 5G base station site selection optimization model for drone communication and its solution algorithm. The model establishment process includes defining the state parameters and variables in drone communication, determining the objective function of the optimization model, and satisfying certain constraints. A solution algorithm for the 5G base station site selection optimization model for drone communication is established, including designing the chromosome characteristics of heuristic algorithms 1 and 2, designing the fitness function, the chromosome mutation operation in heuristic algorithm 1, the neighborhood generation operation in heuristic algorithm 2, the suboptimal solution receiving mechanism and other processes. The present invention models the 5G base station site selection problem for drone communication as a mixed integer nonlinear programming problem, and designs an efficient solution algorithm to solve it. The present invention also takes into account the communication guarantee of ground users, incorporates ground users into the model, and combines the communication needs of drones and the communication needs of ground users in the 5G base station site selection.

Description

Technical Field

[0001] The patent of this invention relates to a 5G base station site selection optimization model and its solution algorithm for drone communications. Technical Background

[0002] With the continuous advancement of science and technology, the intelligent construction of the Internet of Everything has begun to enter the public eye. The development of smart cities requires the development of intelligent unmanned equipment. Drones, as three-dimensional unmanned equipment for smart transportation, are a core element in building an unmanned network. Drones are beginning to be used in various industries, including logistics, power inspections, border inspections, forest plant protection inspections, and urban delivery.

[0003] Drones are commonly used for inspection and material delivery. During inspection operations, each drone must take off from a designated location and conduct inspections along a designated route. During these inspections, drones may experience signal loss, placing them in a blind flight mode. Once in a blind flight mode, drones cannot receive proper control from the ground control station and cannot transmit in-flight inspection data packets. This is unacceptable in actual inspection operations. In addition to inspections, drones also use onboard communication equipment to connect to cellular networks in logistics. Logistics drones take off from designated takeoff points and then, according to designated route nodes, deliver goods and collect packages. However, during the mid-range of a drone delivery route, drones often operate in remote locations, where communication cannot be effectively maintained. In actual logistics, drones often crash due to uncontrolled communication loss.

[0004] With the rise of 5G communication networks, high-bandwidth, low-latency, and slicable networks have become increasingly popular. Connected drones connect to cellular networks via onboard communication modules, enabling communication with the core network. Drone inspections, drone logistics, and drone dynamic reconnaissance all rely on drones' onboard communication modules to connect to network base stations, with data exchanged between the base stations and the core network. The location of 5G base stations will directly impact the reliability and quality of drone communication networks. Summary of the Invention

[0005] In response to the above-mentioned technical problems existing in the prior art, the purpose of the present invention is to provide a 5G base station site selection optimization model for drone communication and its solution algorithm.

[0006] The invention patent is based on the 5G base station site selection under UAV communication. Through the daily route data and route demand points of the UAV, the route map of the UAV and the base station candidates can be mathematically modeled. There are two main types of base station construction: one is the 5G communication equipment on the 4G base station; the other is to build a 5G communication base station separately at the candidate location. Since the construction costs between the two are different, the construction cost can be used in the modeling. . k = 0 indicates the construction type is 5G communication equipment on a 4G base station, and k = 1 indicates the construction type is a separate 5G communication base station. A model is built under constraints such as notification quality, communication latency, basic route coverage, and ground user coverage, and a heuristic algorithm is designed to solve the problem.

[0007] A 5G base station site selection optimization model and its solution algorithm for drone communication. The modeling process is as follows:

[0008] 1.1 Symbol Definition:

[0009] M={L1,L2,…,L m}: UAV flight route collection;

[0010] M'={1,2,…,g}: ground user communication demand set;

[0011] L m ={1,2,..,n}: route L m The set of communication demand points on ;

[0012] N = {1, 2, ..., n}: set of candidate base station locations;

[0013] f i : Communication demand of UAV communication point i (bits / s);

[0014] g i : Communication demand of ground user i (bits / s);

[0015] β: Communication signal quality assurance threshold;

[0016] γ: total investment cost of base station construction;

[0017] α l : UAV flight route l demand point coverage threshold;

[0018] α g : Ground user communication coverage threshold;

[0019] The communication power of demand point i on the UAV route l;

[0020] p h : Communication power of ground user h;

[0021] T: Communication delay guarantee threshold;

[0022] W 4G :4G communication base station allowed bandwidth;

[0023] W 5G :5G communication base station allowed bandwidth;

[0024] N0: ambient noise power;

[0025] CT q (G / G / 1): average waiting time in the communication base station processing queue;

[0026] The cost of building a k-type base station at candidate location j;

[0027] r i : The communication service revenue after demand point i is served;

[0028] 1.2 Variables:

[0029] This means that the communication demand point i on the UAV flight route l is served by base station j;

[0030] This means that the communication demand point i on the UAV flight route l is not served by base station j;

[0031] This means building a base station of type k at candidate location j;

[0032] This means that there is no base station of type k built at candidate location j;

[0033] z hj =1, it means that ground user h is served by base station j;

[0034] z hj =0, it means that the ground user h is not served by the base station j;

[0035] 1.3 Model objective function:

[0036] The objective function set in this modeling is to maximize the benefits generated after the demand service is met. The benefits generated after the demand is met mainly include two parts. One part is the communication demand on the drone route. Each drone route will generate n communication demand points, and each communication demand point will have a size f iThe communication demand, unit is bits / s. Another part of the revenue comes from the satisfaction of the needs of ground users. Each ground user h also has a communication demand g. h The unit is also bits / s. The specific expression of the objective function in the 5G base station site selection modeling for drone communication designed in this invention patent is as follows:

[0037] Objective function expression:

[0038]

[0039] The design goal is to maximize the revenue generated by serving the demand, so a max design is used. The first term in the objective function represents the set of demand for the drone route, and the second term represents the set of demand for ground users.

[0040] 1.4 Constraints:

[0041] In the patent of this invention, the constraints we designed mainly include two parts: drones and ground users. Among them, drone constraints mainly include drone route constraints and base station constraints. In route constraints, we first need to constrain that the demand point i on each route can only be served by one base station at most. The premise for each demand point to be provided with communication services is that a base station needs to be built at that point. Constraints (2)-(6) are basic constraints on the communication between drones and base stations, among which (2)-(3) are constraints on drone inspection needs, including that any communication demand point i on each drone route l can only be served by one base station at most; each base station candidate location can only build one type of base station; communication services can only be provided after the base station is built. The constraint formulas are expressed as follows (2)-(4):

[0042]

[0043]

[0044]

[0045] In addition, for ground users, constraints are also required. Each ground user h can only be served by at most one base station, so we can get constraint formula (5). The premise for ground users to be covered is that there are base stations to be built. Constraint (6) is to constrain the relationship between communication and base station construction. Constraints (5)-(6) are basic constraints between ground users and base stations, and are specifically expressed as follows:

[0046]

[0047]

[0048] In addition, during the drone communication process, since the service processing type designed on the base station side is (G / G / 1), each communication demand i corresponds to this communication data f i At each base station, request packets (i.e., communication data request signals, measured in bits) arrive continuously. Each request packet represents the communication request data for that demand point, measured in bits per second. The base station's processing speed is μ, assuming that all base stations have the same processing speed. If communication requests continuously arrive at base station j while maintaining a constant processing speed, a queue of communication requests will form at base station j. We assume the base station's queue length is infinite. This means there is no queue length requirement on the base station side. Even if the base station is fully loaded, communication data requests can still reach base station j, but they will need to be queued, with a first-in, first-out (FIFO) service policy. Due to the existence of queues, the service time for each communication request can be characterized. Each communication request also has a basic processing delay upper limit. If this delay exceeds the set maximum delay, the request is invalidated, resulting in packet loss for drones or ground users. Since each route and ground user has basic coverage requirements, if the packet loss rate on a route exceeds the set threshold, the route will become unusable. Constraint formulas (7) and (9) are the delay constraints for the queues of UAVs and ground users, as shown below:

[0049]

[0050]

[0051] In addition to the time delay constraint, the communication quality is also a constraint between the drone and the base station. Each communication demand point i on the route l will correspond to a transmission power The communication transmission power is directly related to the data request f, and we design it as a linear correlation. Assuming that the ambient noise N0 during the drone flight is a known quantity, the signal-to-interference ratio is used to reflect the communication quality between the drone and the base station. Similarly, when the ground user's communication transmission power is p h The communication quality between the UAV and the ground user and the base station is also constrained by the signal-to-interference ratio. Constraint formulas (8) and (10) are the communication quality constraints between the UAV, the ground user, and the base station, respectively. They are shown as follows:

[0052]

[0053]

[0054] During a drone inspection operation, each demand point i on the drone route may concurrently communicate with the same base station j. In this case, the transmission bandwidth of base station j is used simultaneously by both demand point i and ground user h. Since the maximum communication bandwidth of a base station cannot be changed, it is necessary to constrain the concurrent communication requests of base station j. The specific constraints are expressed as shown in the following constraint formulas (11) and (12).

[0055]

[0056]

[0057] Among them, constraint formula (11) is the bandwidth constraint for 5G communication equipment on 4G base stations, and constraint formula (12) is the bandwidth constraint for independently built 5G base stations.

[0058] For inspection drones, the premise for each route to operate effectively is that the demand points on each route l must meet a minimum coverage threshold. When the coverage threshold on a drone route l is less than α l When the route cannot be put into use normally. Similarly, for ground users, it is also necessary to ensure the minimum coverage threshold α g The specific constraint formulas are shown in (13)-(14) below.

[0059]

[0060]

[0061] Since the investment in base station construction is huge, the total investment of the project is often known before the base station is built. Therefore, the total investment needs to be constrained in the model, as shown in the following constraint formula (15).

[0062]

[0063] The constraint formulas of the mathematical model are (2)-(18). The specific model constraint formulas are shown as follows (2)-(18):

[0064]

[0065]

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072]

[0073]

[0074]

[0075]

[0076]

[0077]

[0078]

[0079]

[0080]

[0081] Expression (2) indicates that each demand point on each route of the drone is served by at most one base station;

[0082] Expression (3) indicates that each candidate site location can only build one type of base station;

[0083] Expression (4) indicates that only after the base station is built can the communication guarantee service be provided to the UAV;

[0084] Expression (5) indicates that the demand of each ground user can be served by at most one base station;

[0085] Expression (6) indicates that communication guarantee services can be provided to ground users only after the base station is built;

[0086] Expression (7) indicates that once the UAV demand point is served by the base station, the delay constraint needs to be met;

[0087] Expression (8) indicates that once the UAV demand point is served by the base station, the communication quality constraint is required;

[0088] Expression (9) indicates that once the ground user demand point is served by the base station, the delay constraint needs to be met;

[0089] Expression (10) indicates that once the ground user demand point is served by the base station, the communication quality constraint is required;

[0090] Expression (11) represents the bandwidth upper limit constraint of each 4G base station;

[0091] Expression (12) represents the bandwidth upper limit constraint of each 5G base station;

[0092] Expression (13) represents the minimum demand point coverage constraint for each UAV route;

[0093] Expression (14) represents the minimum demand point coverage constraint for ground users;

[0094] Expression (15) represents the total base station investment cost constraint;

[0095] Expressions (16)-(18) represent variable type constraints;

[0096] In the queuing constraint at the base station side, the total arrival rate of communication data arriving at base station j is set to λ j , the service speed of the base station is μ. So the total arrival rate of base station j is λ j , including the demand on the UAV route and the demand of ground users. The constraints are shown below (19) - (22).

[0097]

[0098]

[0099]

[0100]

[0101] Expression (19) represents the queuing model of the base station;

[0102] Expression (20) represents the utilization rate of the base station;

[0103] Expression (21) represents the arrival rate of the base station;

[0104] Expression (22) represents the service rate of the base station;

[0105] 1.5 Solution algorithm:

[0106] Two heuristic algorithms are designed in this patent to solve the site selection and communication service decision of UAV communication base station. In the base station site selection problem, the site selection of the base station is determined by the decision variables To indicate that Indicates that k types of base stations are established at the candidate location j. The selection of each drone communication base station will affect the communication connection decision between the demand points on the subsequent drone route and the demand points of ground users. The connection status between the demand points on the drone route and the base station is determined by the decision variable To reflect. Indicates that the demand point i on the drone route l is served by base station j. The demand of ground users and the connection status between base stations are represented by the decision variable z hj To reflect, a hj = 1 indicates that user h on the ground is served by base station j. In this patent application, we design a genetic algorithm-based construction heuristic algorithm to solve the site selection of 5G base stations for drones. In each generation, a construction heuristic algorithm based on large neighborhood search is designed to solve the connection allocation decision between base stations and demand points.

[0107] 1.5.1 Heuristic Algorithm 1 Chromosome Characteristics

[0108] In heuristic algorithm 1, we designed a construction algorithm based on genetic algorithm to construct the initial solution, and designed the chromosome structure length to be 2n according to the number n of candidate base station locations and the number of types of base stations. According to the binary coding rules, the gene bits of each chromosome are expressed by the numbers 0 and 1. When the binary coding number on the gene bit on the chromosome is 1, it means that a base station of this type is built at the candidate position at that position. The coding sequence of the chromosome is: [(0,0),(0,0),(1,0),(1,0),(0,1),…,(1,0)]. Each gene bit in the chromosome has two pieces of information. The first piece of information indicates whether 5G communication equipment is installed on the 4G base station, and the second piece of information indicates whether a 5G base station is to be built. Only one of the two positions before and after each gene bit can have a binary number of 1. For example Figure 1 As shown, Figure 1 There are 10 candidate base station locations, of which 5G communication equipment on 4G base stations is built at candidate locations 2, 3, 7, and 8, and a 5G base station is built at location 6. The chromosome code is:

[0109] [(0,0),(1,0),(1,0),(0,0),(0,0),(0,1),(1,0),(1,0),(0,0),(0,0)]

[0110] 1.5.2 Heuristic Algorithm 2 Chromosome Characteristics

[0111] In the heuristic algorithm 2, the patent of this invention designs a construction heuristic algorithm based on large neighborhood search to solve the communication connection decision between demand points and base stations. After the site selection decision of each generation of base stations is determined, the demand points need to be allocated based on the decision of each generation of base stations. Figure 1 As shown in the figure, the number of drone routes is 3, and the number of demand points on each route is 4. There are 4 communication demand points for ground users, so there are a total of 16 communication demand points. The numbering method of the demand points on the drone route is: route number - demand number, where the numbering method of the demand points of ground users is The chromosome is encoded in a two-dimensional matrix, and each position is represented by binary, such as Figure 2 shown. Figure 2 Each row in the table represents the demand point served by the base station. For base stations that have not been built, the rows where the base stations are located are all zeros.

[0112] R=[[0,0,0,0,0,0,0,0,0,1,0,0,0,0,0] [0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,1][1,0,0,0,1,1,0,0,1,0,0,0,0,0,0,0] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0] [0,0,1,0,0,0,1,0,0,0,0,0,0,0,1,0] [0,0,0,0,0,0,0,0,0,0,0,0,1,1,0,0] [0,0,0,1,0,0,0,1,0,0,1,0,0,0,0,0] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]]

[0113] 1.5.3 Fitness Function

[0114] In the first stage, the decision is what type of base station to build at what location. Each generation of feasible solutions in heuristic algorithm 1 will correspond to the solution in heuristic algorithm 2. In this invention patent, each of our demand points corresponds to a communication demand f i and g h Each demand corresponds to a benefit, and the design goal is to maximize service benefits. In heuristic algorithms, the fitness function directly influences algorithm iterations. The larger the fitness function value, the more likely individuals are to be retained, and negative fitness function values ​​are not allowed. Therefore, the objective function of this design can be directly used as the fitness function F.

[0115]

[0116] 1.5.4 Chromosome Mutation Operation in Heuristic Algorithm 1

[0117] After completing the population initialization operation, the chromosome mutation operation needs to be performed. The mutation of each gene position of the chromosome in the base station site selection decision of the chromosome mutation gene fragment in the patent design of this invention can not only change the location of the base station construction, but also change the type of base station built at the same location through mutation. The mutation process is shown in the figure below. Figure 3For positional variation, the same gene bit changes from 1 to 0, or vice versa. Similarly, for base station type variation, the type variation must be performed on the same base station candidate. Each base station candidate has two gene bits: the first bit indicates a 4G base station, and the second bit indicates a 5G base station. Only one position between the two can have a value equal to 1.

[0118] 1.5.5 Neighborhood Generation Operation in Heuristic Algorithm 2

[0119] In the demand point allocation heuristic algorithm, the neighborhood generation operation is exchange. For the chromosome in heuristic algorithm 2, each row represents the demand point served by each base station. Each column represents the base station that provides communication services to the demand point under the current demand point. Since it is set that each communication demand point can only be provided with communication services by at most one base station, there can only be at most one position equal to 1 on each column of the chromosome in algorithm 2, and the rest of the positions are 0. The exchange operation mainly operates on each row in the chromosome, randomly selects any two rows on the chromosome, and then randomly selects the exchange fragments, and exchanges the genes of the same length in the two rows to generate a new chromosome. The mutation process is as follows Figure 4 shown.

[0120] 1.5.6 Suboptimal Solution Acceptance Mechanism

[0121] During the iterative process of the heuristic algorithm, solutions worse than the current optimal solution will be generated. These solutions are not discarded blindly, but a certain probability is used to accept the suboptimal solution. The algorithm randomly generates the probability P rand and the receiving probability P a The formula is shown below (23). When the probability of random generation P a ≥P rand , accept the current suboptimal solution.

[0122]

[0123] During the iteration process, the global optimal solution needs to be updated at all times, and the global optimal solution is updated using the following formula (24).

[0124]

[0125] Algorithm termination criterion design:

[0126] In a solution algorithm, it's necessary to design a termination condition for the algorithm iterations. Algorithms often have a limited solution time. Companies need to obtain viable solutions within a short period of time, so the algorithm's total runtime must be limited to T. Therefore, one of the termination conditions in this solution algorithm is that the runtime is less than T. Furthermore, the algorithm also requires a maximum number of solution iterations, which is V in this design.

[0127] The beneficial effects achieved by the present invention are:

[0128] This patent uses actual drone flight route data to extract drone route data requirements. Using the route requirements from historical data, 5G communication base stations are optimally deployed to ensure smooth drone operation. Furthermore, the patent describes two types of 5G base station site selection: the first involves building a single 5G base station at a candidate location to achieve 5G base station coverage; the second involves installing 5G communication equipment on a 4G base station at the candidate location to achieve 5G network coverage. However, in this construction method, the network bandwidth of the 5G communication equipment and the bandwidth of the 5G base station differ. By introducing different site construction costs and route coverage, communication guarantees are achieved for the entire drone communication network. In this patent, the 5G base station site selection problem for drone communication is modeled as a mixed-integer nonlinear programming problem and an efficient solution algorithm is designed. Constraints are considered, including the signal quality of drone communication requests and the processing queue latency of the 5G base station. 5G base station bandwidth constraints are used to implement concurrency control of the base stations, thereby ensuring the drone's latency requirements. In addition, we also consider the communication guarantee of ground users, incorporate ground users into the model, and combine the communication needs of drones and ground users in the site selection of 5G base stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0129] Figure 1 Schematic diagram of site selection for 5G base stations for drone communications;

[0130] Figure 2 Assign chromosome schematics to communication points;

[0131] Figure 3 Schematic diagram of chromosome variation for base station site selection;

[0132] Figure 4 Generate a schematic diagram for the heuristic 2 neighborhood;

[0133] Figure 5 The following is a schematic diagram of the case;

[0134] Figure 6 This is a flowchart of the algorithm. DETAILED DESCRIPTION

[0135] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to examples and drawings. The contents mentioned in the embodiments are not intended to limit the present invention.

[0136] A modeling residual solution algorithm for optimizing the site selection of 5G base stations for user drone communications. The modeling process is as follows:

[0137] 2.1 Symbol Definition:

[0138] M={L1,L2,…,L m}: UAV flight route collection;

[0139] M'={1,2,…,g}: ground user demand set;

[0140] L m ={1,2,..,n}: route L m The set of communication demand points on ;

[0141] N = {1, 2, ..., n}: set of candidate base station locations;

[0142] f i : Communication demand of UAV communication point i (bits / s);

[0143] g i : Communication demand of ground user i (bits / s);

[0144] β: Communication signal quality assurance threshold;

[0145] γ: total investment cost of base station construction;

[0146] α l : UAV flight route l demand point coverage threshold;

[0147] α g : Ground user communication coverage threshold;

[0148] The communication power of demand point i on the UAV route l;

[0149] p h : Communication power of ground user h;

[0150] T: Communication delay guarantee threshold;

[0151] W 4G :4G communication base station allowed bandwidth;

[0152] W 5G :5G communication base station allowed bandwidth;

[0153] N0: ambient noise power;

[0154] CT q (G / G / 1): average waiting time in the communication base station processing queue;

[0155] The cost of building a k-type base station at candidate location j;

[0156] r i: The communication service revenue after demand point i is served;

[0157] 2.2 Variables:

[0158] The communication demand point i on the UAV flight route l is served by base station j;

[0159] Build a base station of type k at candidate location j;

[0160] z hj =1: terrestrial user h is served by base station j;

[0161] The modeling and solution algorithm formula for 5G base station site optimization for drone communication is as follows:

[0162]

[0163]

[0164]

[0165]

[0166]

[0167]

[0168]

[0169]

[0170]

[0171]

[0172]

[0173]

[0174]

[0175]

[0176]

[0177]

[0178]

[0179]

[0180]

[0181]

[0182]

[0183]

[0184] Expression (2) indicates that each demand point on each route of the drone is served by at most one base station;

[0185] Expression (3) indicates that each candidate site location can only build one type of base station;

[0186] Expression (4) indicates that only after the base station is built can the communication guarantee service be provided to the UAV;

[0187] Expression (5) indicates that the demand of each ground user can be served by at most one base station;

[0188] Expression (6) indicates that only after the base station is built can communication guarantee services be provided to ground users;

[0189] Expression (7) indicates that once the UAV demand point is served by the base station, the delay constraint needs to be met;

[0190] Expression (8) indicates that once the UAV demand point is served by the base station, the communication quality constraint is required;

[0191] Expression (9) indicates that once the ground user demand point is served by the base station, the delay constraint needs to be met;

[0192] Expression (10) indicates that once the ground user demand point is served by the base station, the communication quality constraint is required;

[0193] Expression (11) represents the bandwidth upper limit constraint of each 4G base station;

[0194] Expression (12) represents the bandwidth upper limit constraint of each 5G base station;

[0195] Expression (13) represents the minimum demand point coverage constraint for each UAV route;

[0196] Expression (14) represents the minimum demand point coverage constraint for ground users;

[0197] Expression (15) represents the total base station investment cost constraint;

[0198] Expressions (16)-(18) represent variable type constraints;

[0199] Expression (19) represents the queuing model of the base station;

[0200] Expression (20) represents the utilization rate of the base station;

[0201] Expression (21) represents the arrival rate of the base station;

[0202] Expression (22) represents the service rate of the base station;

[0203] 2.3 Constructing a solution algorithm:

[0204] 2.3.1 Heuristic Algorithm 1 Chromosome Characteristics

[0205] In Heuristic Algorithm 1, we designed a genetic algorithm-based construction algorithm to construct the initial solution. Based on the number n of candidate base station locations and the number of base station types, we designed a chromosome structure length of 2n. According to binary encoding rules, each chromosome's gene bit is represented by the digits 0 and 1. When the binary coded digit in a chromosome gene bit is 1, it indicates that a base station of that type will be built at that candidate location. The chromosome coding sequence is: [(0,0),(0,0),(1,0),(1,0),(0,1),…,(1,0)]. Each chromosome gene bit contains two pieces of information: the first indicates whether 5G communication equipment is installed on the 4G base station, and the second indicates whether a 5G base station will be built. Only one of the two positions in each chromosome can have a binary digit of 1.

[0206] 2.3.2 Heuristic Algorithm 2 Chromosome Characteristics

[0207] In the heuristic algorithm 2, the patent of this invention designs a construction heuristic algorithm based on large neighborhood search to solve the communication connection decision between demand points and base stations. After each generation of base station site selection decision is determined, it is necessary to allocate demand points based on each generation of base station decision. The chromosome is encoded in a two-dimensional matrix, and each position is represented in binary, such as Figure 2 shown. Figure 2 Each row in represents the demand point served by the base station. For base stations that were not built in the first phase, the columns where the base stations are located are all 0.

[0208] 2.3.3 Fitness Function

[0209] In the first stage, the decision is what type of base station to build at what location. Each generation in heuristic algorithm 1 corresponds to the solution in heuristic algorithm 2. In the present invention, each of our demand points corresponds to a communication demand f i and g hEach demand corresponds to a benefit, and the design goal is to maximize service benefits. In heuristic algorithms, the fitness function directly influences algorithm iterations. The larger the fitness function value, the more likely individuals are to be retained, and negative fitness function values ​​are not allowed. Therefore, the objective function of this design can be directly used as the fitness function F.

[0210]

[0211] 2.3.4 Chromosome mutation operation in heuristic algorithm 1

[0212] After completing the population initialization operation, the chromosome mutation operation needs to be performed. The mutation of each gene position of the chromosome in the base station site selection decision of the chromosome mutation gene fragment in the patent design of this invention can not only change the location of the base station construction, but also change the type of base station built at the same location through mutation. The mutation process is shown in the figure below. Figure 3 For positional variation, the same gene bit changes from 1 to 0, or vice versa. Similarly, for base station type variation, the type variation must be performed on the same base station candidate. Each base station candidate has two gene bits: the first bit indicates a 4G base station, and the second bit indicates a 5G base station. Only one position between the two can have a value equal to 1.

[0213] 2.3.5 Neighborhood Generation Operation in Heuristic Algorithm 2

[0214] In the demand point allocation heuristic algorithm, the neighborhood generation operation is exchange. For the chromosome in heuristic algorithm 2, each row represents the demand point served by each base station. Each column represents the base station that provides communication services to the demand point under the current demand point. Since it is set that each communication demand point can only be provided with communication services by at most one base station, there can only be at most one position equal to 1 on each column of the chromosome in algorithm 2, and the rest of the positions are 0. The exchange operation mainly operates on each row in the chromosome, randomly selects any two rows on the chromosome, and then randomly selects the exchange fragments, and exchanges the genes of the same length in the two rows to generate a new chromosome. The mutation process is as follows Figure 4 shown.

[0215] 2.3.6 Suboptimal Solution Acceptance Mechanism

[0216] During the iterative process of the heuristic algorithm, solutions worse than the current optimal solution will be generated. These solutions are not discarded blindly, but a certain probability is used to accept the suboptimal solution. The algorithm randomly generates the probability P rand and the receiving probability P a The formula is shown below (23). When the probability of random generation P a ≥P rand, accept the current suboptimal solution.

[0217]

[0218] During the iteration process, the global optimal solution needs to be updated at all times, and the global optimal solution is updated using the following formula (24).

[0219]

[0220] Algorithm termination criterion design:

[0221] In a solution algorithm, it's necessary to design a termination condition for the algorithm iterations. Algorithms often have a limited solution time. Companies need to obtain viable solutions within a short period of time, so the algorithm's total runtime must be limited to T. Therefore, one of the termination conditions in this solution algorithm is that the runtime is less than T. Furthermore, the algorithm also requires a maximum number of solution iterations, which is C in this design.

[0222] Implementation Case 1

[0223] This embodiment 1 is used as an example to verify the scientificity and effectiveness of the above-mentioned modeling and solution algorithm for 5G base station site selection optimization for drone communication (the algorithm flow diagram is shown in FIG. Figure 6 shown):

[0224] (1) According to Table 1 and Table 2 below, the initial parameters of the algorithm are set, and the algorithm population size is set to 200. The number of termination iterations is 200, and the exchange probability P in the algorithm is c =0.8, mutation probability p m= 0.1. The number of candidate locations in the base station site selection parameters is set to 10. The maximum allowable bandwidth for 4G base stations is 1GE, and the maximum allowable bandwidth for 5G base stations is 10GE. The ambient noise power M0 = 105W. Based on the communication data for the demand points in Table 3 below, the communication demand for each communication demand point is measured in bits per second. The chromosome structure length is designed to be 2n based on the number n of candidate base station locations and the number of base station types. According to binary encoding rules, each chromosome's gene bit is represented by the digits 0 and 1. When the binary code digit in a chromosome gene bit is 1, it indicates that a base station of that type will be built at that candidate location. The chromosome coding sequence is: [(0,0), (0,0), (1,0), (1,0), (0,1), …, (1,0)]. Each chromosome gene bit contains two pieces of information: the first indicates whether 5G communication equipment is installed on the 4G base station, and the second indicates whether a 5G base station will be built. Only one of the two positions in each chromosome can have a binary digit of 1. As shown below, there are 10 candidate base station locations, of which 5G communication equipment on 4G base stations is built at candidate locations 2, 3, 7, and 8, and a 5G base station is built at location 6. The chromosome code is:

[0225] [(0,0),(1,0),(1,0),(0,0),(0,0),(0,1),(1,0),(1,0),(0,0),(0,0)]

[0226] (2) Figure 1 As shown in the figure, the number of drone routes is 3, and the number of demand points on each route is 4. The communication demand points of ground users are 4, so there are 16 communication demand points in total. The numbering method is route number-demand number, where The chromosome is encoded in a two-dimensional matrix, and each position is represented by binary, such as Figure 5 Each column represents the demand point served by the base station. For base stations that were not built in the first phase, the columns containing the base stations are all zeros.

[0227] R=[[0,0,0,0,0,0,0,0,0,1,0,0,0,0,0] [0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,1][1,0,0,0,1,1,0,0,1,0,0,0,0,0,0,0] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0] [0,0,1,0,0,0,1,0,0,0,0,0,0,0,1,0] [0,0,0,0,0,0,0,0,0,0,0,0,1,1,0,0] [0,0,0,1,0,0,0,1,0,0,1,0,0,0,0,0] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]]

[0228] (3) After completing the population initialization operation, it is necessary to perform chromosome mutation operation. The mutation of each gene position of the chromosome in the base station site selection decision in the chromosome mutation gene fragment in the patent design of this invention can not only change the location of the base station construction, but also change the type of base station built at the same location through mutation. The process diagram is as follows Figure 3 For positional variation, the same gene bit changes from 1 to 0, or from 0 to 1. Similarly, for positional variation, positional variation needs to be performed at the same base station candidate point. Each base station candidate point has two gene bits, the first one represents a 4G base station, and the second one represents a 5G base station. Only one gene bit between the two can have a value equal to 1.

[0229] (4) In the iterative process of the heuristic algorithm, solutions worse than the current optimal solution will be generated. For these solutions, the design does not always discard them but adopts a certain probability to accept the suboptimal solution. The algorithm randomly generates the probability P rand and the receiving probability P a , as shown in formula (23). When the probability of random generation P a ≥P rand , accept the current suboptimal solution.

[0230]

[0231] During the iteration process, the global optimal solution needs to be updated at all times, and the global optimal solution is updated using formula (24).

[0232]

[0233] (5) According to the rules, the chromosomes in the population are mutated and exchanged to form a new population through mutation and exchange.

[0234] (6) Iteration: Set a suitable evolutionary generation. When the number of iterations is greater than the preset evolutionary generation or the convergence condition is reached, the optimization process ends.

[0235] Table 1 Basic parameters of the algorithm

[0236]

[0237] Table 2 Basic parameters of base stations

[0238]

[0239] Table 3 Demand parameters

[0240]

Claims

1. A 5G base station site selection optimization model for drone communication, characterized by The establishment of a model for optimizing the site selection of 5G base stations for drone communications includes the following steps: 1) Define the state parameters and variables in drone communication. The parameter symbols are defined as follows: M={L1,L2,…,L m }: UAV flight route collection; M'={1,2,…,g}: ground user communication demand set; L m ={1,2,..,n}: route L m The set of communication demand points on ; N = {1, 2, ..., n}: set of candidate base station locations; f i : Communication demand of UAV communication point i (bits / s); g i : Communication demand of ground user i (bits / s); β: Communication signal quality assurance threshold; γ: total investment cost of base station construction; α l : UAV flight route l demand point coverage threshold; α g : Ground user communication coverage threshold; The communication power of demand point i on the UAV route l; p h : Communication power of ground user h; T: Communication delay guarantee threshold; W 4G :4G communication base station allowed bandwidth; W 5G :5G communication base station allowed bandwidth; N0: ambient noise power; CT q (G / G / 1): average waiting time in the communication base station processing queue; The cost of building a k-type base station at candidate location j; r i : The communication service revenue after demand point i is served; The communication demand point i on the UAV flight route l is served by base station j; Build a base station of type k at candidate location j; z hj =1: terrestrial user h is served by base station j; The variable type constraints are as follows: 2) Determine the objective function of the optimization model: The objective function set in the modeling means maximizing the benefits generated after the demand service is met. The benefits generated after the demand is met mainly include two parts. The first part is the communication demand set of the drone route. Each drone route will generate n communication demand points, and each communication demand point will have a size of f i The communication demand of ground users is expressed in bits / s. The second part is the communication demand set of ground users. Each ground user h also has a communication demand g. h , the unit is also bits / s; therefore, the specific expression of the objective function in the 5G base station site selection modeling of UAV communication is shown in formula (1): The objective function above is to maximize the benefits generated after the demand is served; 3) Determine the constraints, including the following constraint process 1. Set basic constraints on the communication between drones and base stations. Each base station candidate location can only be built with one type of base station, and communication services can only be provided after the base station is built.

2. Set basic constraints on communications between ground users and base stations; 3. After the communication request data of the drone and the ground user arrive at the base station, they are queued for processing. The service rule of the base station is set to FIFO mode, and the delay constraints for the queues of drones and ground users are set; 4. Constrain the communication quality between drones, ground users and base stations; 5. The base station's transmission bandwidth is used simultaneously by drone route demand points and ground users. Since the base station's maximum communication bandwidth remains unchanged, constraints are imposed on the base station's concurrent communication requests. This means that bandwidth constraints are imposed on 5G communication devices on 4G base stations, as well as on independently constructed 5G base stations.

6. For inspection drones, the effective operation of each route requires that the demand points on each route meet a minimum coverage threshold. Similarly, a minimum coverage threshold must be guaranteed for ground users. Therefore, a minimum demand point coverage constraint is imposed on each drone route, as well as a minimum demand point coverage constraint on ground users.

7. Constraints on total base station investment costs; 8. In the queuing constraints on the base station side, set the constraints of the base station queuing model, base station utilization, base station arrival rate and base station service rate.

2. A 5G base station site selection optimization model for drone communication according to claim 1, characterized in that In step 3), the basic constraints set for communication between the drone and the base station are shown in formulas (2)-(4): Expression (2) indicates that each demand point on each route of the drone is served by at most one base station; Expression (3) indicates that each candidate site location can only build one type of base station; Expression (4) indicates that communication guarantee services can be provided to drones only after the base station is built.

3. A 5G base station site selection optimization model for drone communication according to claim 1, characterized in that In step 3), the basic constraints set for communication between ground users and base stations are shown in formulas (5)-(6): Expression (5) indicates that the demand of each ground user can be served by at most one base station; Expression (6) indicates that communication guarantee services can be provided to ground users only after the base station is built.

4. A 5G base station site selection optimization model for drone communication according to claim 1, characterized in that In step 3), the delay constraints for the queuing of UAVs and ground users are set as shown in formulas (7) and (9): Expression (7) indicates that once the UAV demand point is served by the base station, the delay constraint needs to be met; Expression (9) indicates that once a ground user demand point is served by a base station, the delay constraint must be met.

5. A 5G base station site selection optimization model for drone communication according to claim 1, characterized in that In step 3), the communication quality constraints between the UAV, ground users and base stations are as shown in formulas (8) and (10): Expression (8) indicates that once the UAV demand point is served by the base station, the communication quality constraint needs to be met; Expression (10) indicates that once the ground user demand point is served by the base station, the communication quality constraint must be met.

6. A 5G base station site selection optimization model for drone communication according to claim 1, characterized in that In step 3), the bandwidth constraints for the 5G communication equipment on the 4G base station and the bandwidth constraints for the independently constructed 5G base station are shown in formulas (11) and (12), respectively: Expression (11) represents the bandwidth upper limit constraint of each 4G base station; Expression (12) represents the bandwidth upper limit constraint of each 5G base station.

7. A 5G base station site selection optimization model for drone communication according to claim 1, characterized in that In step 3), the minimum demand point coverage constraint for each UAV route and the minimum demand point coverage constraint for ground users are shown in formulas (13) and (14), respectively: Expression (13) represents the minimum demand point coverage constraint for each UAV route; Expression (14) represents the minimum demand point coverage constraint for ground users.

8. A 5G base station site selection optimization model for drone communication according to claim 1, characterized in that In step 3), the total base station investment cost constraint is shown in formula (15): Expression (15) represents the total base station investment cost constraint.

9. A 5G base station site selection optimization model for drone communication according to claim 1, characterized in that In step 3), in the queuing constraint at the base station side, the total arrival rate of communication data arriving at base station j is set to λ j , the service speed of the base station is μ, and the constraints of the base station queuing model, base station utilization, base station arrival rate and base station service rate are set as shown in formulas (19)-(22): Expression (19) represents the queuing model of the base station; Expression (20) represents the utilization rate of the base station; Expression (21) represents the arrival rate of the base station; Expression (22) represents the service rate of the base station.

10. The algorithm for solving the 5G base station site selection optimization model for UAV communication according to claim 1, characterized in that The solution algorithm for the 5G base station site selection optimization model for drone communications includes the following steps: Design two heuristic algorithms to solve the site selection and communication service decision of UAV communication base stations; In the base station location problem, the location of the base station is determined by the decision variables To indicate that Indicates that k types of base stations are established at the candidate location j; the selection of each UAV communication base station will affect the communication connection decision between the demand points on the subsequent UAV route and the demand points of ground users. The connection status between the demand points on the UAV route and the base station is determined by the decision variable To reflect, Indicates that the demand point i on the drone route l is served by base station j; the demand of ground users and the connection status between base stations are represented by decision variables z hj To reflect, z hj =1 means that user h on the ground is served by base station j; Therefore, a constructive heuristic algorithm based on genetic algorithm is designed to solve the site selection of drone 5G base stations. In each generation, a constructive heuristic algorithm based on large neighborhood search is designed to solve the connection allocation decision between base stations and demand points: 1) Design heuristic algorithm 1 chromosome characteristics A construction algorithm based on a genetic algorithm is designed to construct an initial solution. The chromosome structure length is designed to be 2n according to the number n of candidate base station locations and the number of base station types. According to the binary coding rules, the gene bits of each chromosome are expressed by the numbers 0 and 1. When the binary code number on the gene bit on the chromosome is 1, it indicates that a base station of that type is built at the candidate location at that location. When the binary code number on the gene bit on the chromosome is 0, it indicates that a base station of that type has not been built at that location. Each gene bit in the chromosome has two pieces of information. The first piece of information indicates whether 5G communication equipment is installed on the 4G base station, and the second piece of information indicates whether a 5G base station is to be built. Only one of the two positions on each gene bit can have a binary number of 1. 2) Design heuristic algorithm 2 chromosome characteristics A constructive heuristic algorithm based on large neighborhood search is designed to solve the communication connection decision between demand points and base stations. After the site selection decision of each generation of base stations is determined, the demand points are allocated based on the decisions of each generation of base stations. The chromosomes are encoded in a two-dimensional matrix, and each position is represented in binary. Each row of the two-dimensional matrix represents the demand point served by each base station, and each column represents the base station providing communication service to the demand point under the current demand point. Since each communication demand point is set to be provided with communication services by at most one base station, there can only be one position equal to 1 in each column of the chromosome in heuristic algorithm 2, and the rest of the positions are 0; in addition, for base stations that have not been built, the rows where the base stations are located are all 0; 3) Design fitness function In the first stage, decisions are made on the location and type of base stations to be built. Each generation of feasible solutions in heuristic algorithm 1 corresponds to the solution in heuristic algorithm 2. Each demand point corresponds to a communication demand f i and g h Each demand corresponds to a benefit. The goal of designing the fitness function is to maximize the service benefit. In the heuristic algorithm, the fitness function directly affects the iteration of the algorithm. The larger the fitness function value, the easier it is for the individual to be retained. The fitness function value is not allowed to be negative. Therefore, the objective function can be directly used as the fitness function F. Its calculation formula is as follows: 4) Chromosome mutation operation in heuristic algorithm 1 Chromosome mutation gene fragments are the mutations of each gene position of the chromosome in the base station site selection decision. Through mutation, not only the location of the base station construction can be changed, but also the type of base station built at the same location can be changed. For positional variation, the same gene position changes from 1 to 0, or from 0 to 1; Position mutation needs to be performed on the same base station candidate point. Each base station candidate point has two gene bits. The first bit represents a 4G base station, and the second bit represents a 5G base station. Only one position between the two can have a value equal to 1. 5) Neighborhood generation operation in heuristic algorithm 2 In the demand point allocation heuristic algorithm, neighborhood generation is achieved through exchange. For the chromosome in heuristic algorithm 2, each row represents the demand point served by each base station, and each column represents the base station providing communication services to the demand point under the current demand point. Each communication demand point can only be provided with communication services by at most one base station. For each column of the chromosome in heuristic algorithm 2, at most one position can be equal to 1, and the rest of the positions are 0. The exchange operation mainly operates on each row of the chromosome. It randomly selects any two rows on the chromosome, and then randomly selects the exchange fragments. The genes of the two rows with equal lengths are exchanged to generate a new chromosome. 6) Suboptimal solution reception mechanism In the iterative process of the heuristic algorithm, solutions worse than the current optimal solution will be generated. These solutions are not abandoned blindly, but a certain probability is used to accept the suboptimal solution. The algorithm randomly generates the probability P rand and the receiving probability P a The calculation formula is as follows (23), when the probability of random generation P a ≥P rand When , accept the current suboptimal solution: During the iteration process, the global optimal solution needs to be updated at all times. The global optimal solution is updated using the following formula (24): Finally, design the termination condition of the algorithm iteration, set the maximum number of iterations to C, and the total running time of the solution algorithm needs to be limited to T.

Images