CN111612148A - Near-earth space communication system deployment method based on similarity decomposition multi-objective evolution - Google Patents

Abstract

The invention discloses a communication system deployment method based on similarity decomposition multi-objective optimization, which mainly solves the problem of aircraft deployment in a communication system. The method comprises the following implementation steps: (1) inputting algorithm parameters; (2) shifting and initializing an initial population; (3) sequentially selecting and carrying out cross variation on individuals in the solution population to obtain offspring individuals; (4) updating the solution population by utilizing the offspring individuals; (5) judging whether to terminate: if the iteration times meet the preset times, executing the step (6), otherwise, turning to the step (3); (6) and selecting the optimal aircraft deployment position according to the principle of maximum module density. From the perspective of NSCS, the method provides a general model covering important aspects of NSCS. For NSCS, a general model is used to calculate traffic attenuation and calculate routing efficiency. From the perspective of an algorithm, a mobile initialization MOEA/D based on similarity is provided to optimize the deployment of NSCS.

Description

Near-earth space communication system deployment method based on similarity decomposition multi-objective evolution

Technical Field

The invention belongs to the field of communication system deployment, relates to a multi-objective evolutionary method based on similarity decomposition, and particularly relates to a multi-objective optimized evolutionary algorithm based on similarity decomposition, which can be used for communication system deployment, cellular network deployment or sensor network deployment.

Background

The near-earth space is about 20 to 100 kilometers above sea level and comprises stratosphere, intermediate layer and thermal layer. At this altitude, the number of aircraft is reduced, which makes the near-space aircraft safe. The temperature is between-90 degrees centigrade and 30 degrees centigrade, which is suitable for most equipment. Wind speed of 80000 feet is less than 70 knots 99% of the time, which makes the near space vehicle easier to maneuver. The environment of the near-ground space is therefore ideal for aircraft.

Near Space Communication Systems (NSCS) are an emerging alternative solution to the ever-increasing communication needs of the modern world. Near space platforms are widely used for synthetic aperture radar due to their high resolution and wide swath. The device is convenient to configure and low in cost, and can be used for earth observation. Near space communication systems are attracting increasing attention as compared to terrestrial wireless communication systems and satellite communication systems. The deployment of aircraft is critical to the performance of near-space communication systems. Due to limited aircraft resources, near space communication systems cannot provide high speed networks while maintaining a large coverage area. The management of the near space communication system needs to optimize the trade-off solution to select different scenarios. This problem is therefore a multi-objective optimization problem. There has been some research into the optimization of the deployment of NSCS with respect to network speed, energy consumption, ground target coverage and quality of detection. Deployment optimization is also solved as a multi-objective optimization problem.

The deployment optimization problem of the aircraft is similar to that of other fields such as a sensor network, a cellular network and a smart grid. Mohamed proposes a probability model of forest fires, converts the deployment problem into a k coverage maximization problem, and proposes a distributed algorithm to obtain multiple sensor deployments. The i.m.r uses Voronoi diagrams to minimize energy consumption while maximizing coverage. The goal of sensor deployment optimization varies for various applications, with coverage being one of the most important. L.r.c proposes weighted network deployment optimization to maximize mobile sensor coverage. Shibo proposes a situation where deployment based on rows is locally optimal and proposes a deterministic deployment algorithm to generate coverage of terrain.

For the deployment optimization of the NSCS, no literature currently considers the path loss, routing efficiency, user requirements, and security issues, which are the main concerns of modern communication systems. All works so far only consider the above four aspects. In this context, the proposed NSCS model is to first consider the above four main aspects and provide a general model for the deployment optimization of NSCS. This model can be easily extended by more details of modern communication systems. The deployment patterns common to many other deployment optimization problems were also first investigated. A multi-objective optimization algorithm based on similarity is also proposed to optimize the deployment of the NSCS.

Disclosure of Invention

In view of the above-mentioned shortcomings of the prior art, the present invention proposes a communication system deployment method based on similarity decomposition multi-objective optimization. From the perspective of NSCS, a general model covering important aspects of NSCS is proposed. From the perspective of an algorithm, a mobile initialization MOEA/D based on similarity is provided to optimize the deployment of NSCS.

The technical scheme of the invention is as follows: the communication system deployment problem is regarded as a multi-objective problem, wherein the network speed and the coverage range are used as objective functions, the objective functions are simultaneously optimized by using an evolutionary multi-objective method based on similarity decomposition, and a better deployment solution position is searched, and the implementation steps comprise the following steps:

(1) inputting algorithm parameters: parameters of a near space communication system deployment method are based on other parameters of a similarity decomposition algorithm;

(2) initialization: generating an initial population with the size of N by using shift initialization, and initializing an output population into an empty set;

(3) constructing a network speed, wherein the coverage range is used as an objective function: 3a) network speed consists of three elements: user model, path loss and routing efficiency. (ii) a

3b) Function of coverage S:

wherein r is_cRadius of the aircraft covering circle, SA_jIs the area covered by the aircraft, and M is the number of the aircraft, it is noted that no coincidence is considered here, because the aircraft is deployed in blocks;

(4) optimizing an objective function:

4a) aggregating the two objective functions C and S in the step (3) into a single objective function by using a Chebyshev mathematical decomposition method as sub-objective functions, recording the number of the sub-objective functions as N, namely the population size is the same as the number of the sub-objective functions, and obtaining weight vectors corresponding to the two sub-objective functions C and S in the single objective function respectively;

4b) calculating an initial solution population decomposition target value: obtaining an initial solution population with the size of N by using the shift initialization method in the step (2), and calculating to obtain a single objective function value and sub-objective function values of each individual in the population;

4c) selecting a parent individual: selecting two parent individuals from the solution population, wherein one of the two parent individuals is a solution individual corresponding to the ith sub-objective function, and the other is a solution individual randomly selected from the solution population;

4d) cross mutation: carrying out uniform cross operation on the two selected parent individuals to obtain a child individual, and carrying out neighborhood variation operation on the child individual to obtain a new child;

4e) updating the sub population corresponding to the ith sub objective function solution individual: constructing a sub-population with the individual number of M for the ith sub-objective function solution individual according to the Euclidean distance minimum principle among weight parameters in each sub-objective function, and updating the sub-population corresponding to the ith objective sub-function solution individual by using new filial generations;

4f) repeating the steps 4c) to 4e) until the N subfunctions are executed, and obtaining a solution population (x)₁,y₁,...,x_M,y_M) I.e. coordinate position;

4g) judging whether to terminate: if the population evolution termination algebra meets the preset algebra gen, executing (5), otherwise, repeating the steps 4c) to 4 g);

(5) selecting the best deployment position: subjecting the final solution population (x) obtained in step 4g) to₁,y₁,...,x_M,y_M) As the optimal deployment location for the aircraft.

Compared with the prior art, the invention has the following advantages:

first, from the perspective of NSCS, a general model covering important aspects of NSCS is proposed. For NSCS, a general model is used to calculate traffic attenuation and calculate routing efficiency. We elaborate on the needs of the user, the NSCS handles various types of regions in different ways. Safety issues are also taken into account. These four aspects cover the major problems of modern communication systems from three directions: NSCS, communication procedures and users. To our knowledge, this is the first general NSCS model to consider these four aspects.

Secondly, from the perspective of an algorithm, a mobile initialization MOEA/D based on similarity is proposed to optimize the deployment of NSCS. The algorithm was the first to take advantage of group perception. Based on group perception, the similarity-based MOEA/D can effectively optimize the deployment of NSCS.

Drawings

FIG. 1 is a flow chart of the present invention.

Detailed Description

Referring to fig. 1, the implementation steps of the invention are as follows:

step 1, inputting algorithm parameters: parameters of a near space communication system deployment method are based on other parameters of a similarity decomposition algorithm;

step 2, population initialization: generating an initial population with the size of N by using shift initialization, and initializing an output population into an empty set;

(2a) the entire spatial coverage area is divided into P tiles, denoted as { B }₁,...,B_i,...,B_PRecording the center coordinates (X) of each block Bi_bi,Y_bi) The number of users per block is taken as the block size, and the blocks are different from each other as much as possible in terms of the number of users;

(2b) arranging the P blocks in descending order according to the number of users;

(2c) generating an initial solution for aircraft deployment:

(2c1) each newly deployed aircraft attempts to select the region where the previous aircraft did not select the highest number of users to deploy the aircraft, and the aircraft that was deployed later should not conflict with the aircraft that was previously deployed, and, for the initial solution k,

is the position coordinate of the jth aircraft in the initial solution k,

is the coordinate of the central point of the second block,

(X^B,Y^B) Is the X-axis length and the Y-axis length of each block,

α, β -U (0,1), i.e., α and β are initial random values between (0, 1);

(2c2) if the central position of the ith block Bi conflicts with the previously deployed aircraft, then the aircraft deployment position will move to the next unselected block;

(2c3) the aircraft is deployed until the deployment reaches the boundary of the spatial coverage area, or P blocks are deployed.

Step 3, constructing network speed, and taking coverage range as target function

(3a) Network speed is quantified by network capacity, and a network speed quantification function is represented as C:

wherein the CTU_lNetwork capacity used for a single user, L being the number of users;

the network capacity of a single user comprises three elements of a user model, path loss and routing efficiency, and the formula of the path loss is as follows:

wherein P is_rIs the received signal power, P, directly related to the network capacity acquired by the user_tIs the signal transmission power, λ_cIs the wavelength, G_tAnd G_rRespectively, the transmission gain and the reception gain of the antenna, x being the transmission distance; since the network capacity is P_rAccumulation over a period of time, thus P_rAs one of the quantitative factors of the network capacity;

user l is at location (x)_l,y_l,z_l) Is located at position (x)_j,y_j,z_j) The network capacity received by aircraft j is defined as follows:

RU_jl＝1-E_f(D_jl)³

E_fis an attenuation factor in the signal transmission process, D_jlIs the distance, RU, between user l and aircraft j_jlA reception ratio for a user;

actual network capacity that the user obtains from aircraft jCU_jlIs defined as follows:

CU_jl＝CE_jlRU_jl

wherein, CA_jTo total capacity of the aircraft, NA_jNumber of subscribers connected to aircraft j, CE_jlIs the network capacity, RU, of the aircraft to assign to a single user l_jlReception ratio for user l;

total network capacity CTU obtained by the user from all aircraft_lIs defined as follows:

S_jlthis is indicated as 1 if there is a connection between the aircraft j and the user l, or 0 otherwise.

(3d) Function of coverage S:

SA_j＝πr_c ²，

wherein r is_cRadius of the aircraft covering circle, SA_jIs the area covered by the aircraft, and M is the number of the aircraft, it is noted that no coincidence is considered here, because the aircraft is deployed in blocks;

the method as claimed in claim 1, wherein the step 4a) of using the chebyshev mathematical decomposition method to aggregate the two objective functions in the step (3) into a single objective function is performed by the following chebyshev mathematical decomposition formula:

where x is the solution of a function, f_i(x) Two objective functions C and S as described before can be represented,

is f_i(x) Maximum value of (a)_iThe weight parameter corresponding to the ith objective function, i ∈ (1,2), g (x), represents the total objective function value corresponding to the aggregation of the two objective functions, i.e. the total single objective function is minimized.

And 5, selecting the parent individuals.

Selecting two parent individuals from the solution population, wherein one of the two parent individuals is a solution individual corresponding to the ith sub-objective function, and the other is a solution individual randomly selected from the solution population;

and 6, crossing.

The purpose of the crossover is to distribute aircraft between potential zones. For two selected solution units

And

calculating the similarity between the two solutions;

the similarity calculation adopts Euclidean distance, and the Euclidean distance between individuals P and Q is solved as follows:

selecting two points with the nearest Euclidean distance between solution P and Q, and taking the random probability theta, when theta is smaller than the cross probability PB defined in advance_cWhen not executing the cross operation; when theta is greater than cross probability PB_cWhen the method is used, the cross operation is executed, and the method specifically comprises the following steps:

l is randomly chosen from 1, M,

respectively, are solving for P and QThe positions of the two closest european points,

is the descendant generated in solution P;

and 7, mutation.

The purpose of the mutation is to search for potential areas to find the best position to take full advantage of the aircraft. It is assumed that the intersection has been performed and that the aircraft 2 of individual B provides the desired direction vector in region B. Based on this direction vector, individual a can improve one object without deteriorating another object by searching for the area around it. Given the two individual P ' and Q ' resulting from the crossover, the airship sequences of Q ' are rearranged to obtain a common perception of each potential region and find a satisfactory sequence of aircraft.

For solutions P and Q, firstly, the direction vector and the vertical vector between the solutions P and Q are obtained through calculation:

is the direction vector of the connecting line of the Kth point of solution P and the Kth point of solution Q,

is a perpendicular vector perpendicular to the direction vector;

and then generating a solution position after mutation according to the direction vector and the vertical vector:

is the position of the solution after the variation,

α, γ, β, are random parameters between (0, 1).

Step 8, repeating the step 6 to the step 7 until the N subfunctions are executed, and obtaining a solution population (x)₁,y₁,...,x_M,y_M)

And 9, judging whether the operation is terminated.

And (5) if the population evolution termination iteration times meet the preset algebra gen, the range of the population evolution termination iteration times is 200-300, executing the step 10, and otherwise, repeating the steps 6-8.

Step 10, selecting the optimal deployment position.

Improved solution population (x) from step 10₁,y₁,...,x_M,y_M) And selecting the solution individual with the maximum module density value as the optimal solution.

Claims (5)

1. A deployment method for a near-earth space communication system, comprising the steps of:

(1) inputting parameters of a deployment method of the near-earth space communication system;

(2) generating an initial population with the size of N by using a shift initialization method, and initializing the generated initial population into an empty set;

(3) network speed is quantized, and coverage is taken as an objective function:

3a) network speed is quantified by network capacity, and a network speed quantification function is expressed as an objective function C:

wherein the CTU_lNetwork capacity used for a single user L, L being the number of users;

3b) function to determine coverage S:

SA_j＝πr_c ²，

wherein r is_cRadius of the aircraft covering circle, SA_jThe area covered by the aircraft j, M is the number of the aircraft j, and the aircraft are deployed in blocks;

(4) optimizing an objective function:

4a) aggregating the two objective functions C and S in the step (3) into a single objective function by using a Chebyshev mathematical decomposition method as sub-objective functions, recording the number of the sub-objective functions as N, namely the population size is the same as the number of the sub-objective functions, and obtaining weight parameters corresponding to the two sub-objective functions C and S in the single objective function respectively;

4b) calculating an initial solution population decomposition target value: obtaining an initial solution population with the size of N by using a shift initialization method, and calculating to obtain a single objective function value and sub-objective function values of each individual in the solution population;

4c) selecting a parent individual: selecting two parent individuals from the solution population, wherein one of the two parent individuals is a solution individual corresponding to the ith sub-objective function, and the other is a solution individual randomly selected from the solution population;

4d) cross mutation: carrying out uniform cross operation on the two selected parent individuals to obtain a child individual, and carrying out neighborhood variation operation on the child individual to obtain a new child;

4e) updating the sub population corresponding to the ith sub objective function solution individual: constructing a sub-population with the individual number of M for the ith sub-objective function solution individual according to the Euclidean distance minimum principle among weight parameters in each sub-objective function, and updating the sub-population corresponding to the ith objective sub-function solution individual by using new filial generations;

4f) repeating the steps 4c) to 4e) until the N subfunctions are executed, and obtaining a solution population (x)₁,y₁,...,x_M,y_M) I.e. coordinate position;

4g) judging whether to terminate: if the population evolution termination algebra meets the preset algebra gen, executing (5), otherwise, repeating the steps 4c) to 4 g);

(5) selecting the best deployment position: subjecting the final solution population (x) obtained in step 4g) to₁,y₁,...,x_M,y_M) As the optimal deployment location for the aircraft.

2. The method of claim 2, wherein the initialization using shifting of step (2) comprises:

(2a) the entire spatial coverage area is divided into P rectangular blocks, denoted as { B }₁,...,B_i,...,B_PRecording the center coordinates (X) of each rectangular block Bi_bi,Y_bi) Taking the number of users of each rectangular block as the size of the rectangular block;

(2b) arranging the P rectangular blocks in a descending order according to the number of users;

(2c) generating an initial solution for aircraft deployment:

(2c1) each newly deployed aircraft attempts to select the region where the previous aircraft did not select the highest number of users to deploy the aircraft, and the aircraft that was deployed later should not conflict with the aircraft that was previously deployed, and, for the initial solution k,

is the position coordinate of the jth aircraft in the initial solution k,

is the coordinate of the central point of the second rectangular block,

(X^B,Y^B) Is the X-axis length and the Y-axis length of each block,

α, β -U (0,1), i.e., α and β are initial random values between (0, 1);

(2c2) if the center position of the ith rectangular block Bi conflicts with the previously deployed aircraft, the aircraft deployment position will move to the next unselected rectangular block;

(2c3) the aircraft is deployed until the deployment reaches the boundary of the spatial coverage area, or the P tiles are deployed.

3. The method of claim 1, wherein the network capacity comprises three elements of a user model, a path loss and a routing efficiency, and the formula of the path loss is:

wherein P is_rIs the received signal power, P, directly related to the network capacity acquired by the user_tIs the signal transmission power, λ_cIs the wavelength, G_tAnd G_rRespectively, the transmission gain of the antenna and the reception gain of the antenna, and x is the transmission distance;

user l is at location (x)_l,y_l,z_l) Is located at position (x)_j,y_j,z_j) The network capacity received by aircraft j is defined as follows:

RU_jl＝1-E_f(D_jl)³

E_fis an attenuation factor in the signal transmission process, D_jlIs the distance, RU, between user l and aircraft j_jlA reception ratio for a user;

actual network capacity CU obtained by a user from aircraft j_jlIs defined as follows:

CU_jl＝CE_jlRU_jl

wherein, CA_jTo total capacity of the aircraft, NA_jNumber of subscribers connected to aircraft j, CE_jlIs the network capacity, RU, of the aircraft to assign to a single user l_jlReception ratio for user l;

user' sTotal network capacity CTU obtained from all aircraft_lIs defined as follows:

S_jlthis is indicated as 1 if there is a connection between the aircraft j and the user l, or 0 otherwise.

4. The method according to claim 1, wherein the step 4a) of using the chebyshev mathematical decomposition method to aggregate the two objective functions in the step (3) into a single objective function is performed by the following chebyshev mathematical decomposition formula:

where x is the solution of a function, f_i(x) Representing the objective functions C and S of the system,

is f_i(x) Maximum value of (a)_iThe weight parameter corresponding to the ith objective function, i ∈ (1,2), g (x), represents the total objective function value corresponding to the aggregation of the two objective functions C and S, i.e. minimizing the total single objective function.

5. The method of claim 1, wherein the step (4d) of cross-mutating comprises:

and (3) crossing: for two selected solutions P:

and Q:

calculating the similarity between the two solutions;

the similarity calculation adopts Euclidean distance, and the Euclidean distance between individuals P and Q is solved as follows:

selecting two points with the nearest Euclidean distance between solution P and Q, and taking the random probability theta, when theta is smaller than the cross probability PB defined in advance_cWhen not executing the cross operation; when theta is greater than cross probability PB_cWhen the method is used, the cross operation is executed, and the method specifically comprises the following steps:

l is randomly chosen from 1, M,

the positions of two points with the shortest Euclidean distance between solution P and Q respectively,

is the descendant generated in solution P;

(4b3) mutation:

for solutions P and Q, firstly, the direction vector and the vertical vector between the solutions P and Q are obtained through calculation:

is the direction vector of the connecting line of the Kth point of solution P and the Kth point of solution Q,

is a perpendicular vector perpendicular to the direction vector;

and then generating a solution position after mutation according to the direction vector and the vertical vector:

is the position of the solution after the variation,

α, γ, β, are random parameters between (0, 1).

Images