KR20030053278A - Method for Optimizing Wireless Network using the optimization of domain - Google Patents

Abstract

PURPOSE: A wireless network optimizing method using domain optimization is provided to optimize a wireless network within practical time by dividing an area of the wireless network to domains and optimizing the domain. CONSTITUTION: An entire optimization-subjected area is divided into overlap domains(200). A suitability degree of every domain subjected for optimization is calculated(202). It is judged whether a suitability degree of the entire subject area is greater than a threshold value(204). A domain with the lowest suitability degree is designated as a primary domain(206). The primary domain is designated as a target domain(208). Setting of the target domain is optimized by using a gene algorithm(210). It is checked whether the suitability degree of the primary domain is greater than an average value of the entire domains(212). If the suitability degree of the primary domain is smaller than the average value of the entire domains, a domain with the lowest suitability degree among secondary domains is designated as a target domain(214). Setting of the target domain is optimized by using the gene algorithm(216).

Description

Method for Optimizing Wireless Network using the optimization of domain}

The present invention relates to a method for optimizing a wireless network. In particular, the wireless network area to be optimized is divided into sub-domains called domains, and the optimized domain is easily optimized within a realistic time by optimizing the divided domain using a genetic algorithm. A wireless network optimization method using domain optimization and a computer readable recording medium having recorded thereon a program for realizing the method.

In the future, wireless data communication networks will connect with current wired communication networks to provide services to users everywhere. In this case, in order to efficiently use a wireless communication resource such as frequency to provide a high quality service to a user, the wireless data network must be variably optimized during the communication service according to the amount of data traffic.

Wireless network optimization is performed by optimizing the parameters of each base station antenna in order to maximize communication service quality and maximize communication service provided.

That is, it is necessary to vary the direction, tilt and propagation strength of the antenna according to the distribution of the communication traffic to provide communication capacity for dynamically changing user traffic.

Therefore, the computational complexity required for optimization is determined by the number of parameters to be optimized. In the general case, the parameters to be optimized include parameters such as the direction of the antenna, tilt and propagation strength which determine signal strength.

Many automation tools have been developed and new methods have been proposed to automate wireless network optimization, but the optimization problem itself is a non-linear problem, and in general, non-linear problems cannot be solved in real time. .

Therefore, in order to solve the wireless network optimization problem, the conventional wireless network optimization method used a method of searching for a solution set using a heuristic method such as a genetic algorithm for the wireless network optimization. In order to improve the computation speed, C / PVM (Parallel Virtual Machine) parallel computation method was used by connecting many computers.

However, such a conventional method is generally a very efficient method for searching for a solution set by the algorithm itself. However, since the computational complexity of the fitness function essential for the genetic algorithm is too high for the optimization of the wireless network, the time is realistic. There was a problem that I could not get meaningful results.

The present invention has been devised to solve the above problems, and the wireless network area to be optimized is divided into sub-domains of domains, and the genetic domains are optimized to simplify the divided domains in a realistic time. It is an object of the present invention to provide a method for optimizing a wireless network using domain optimization and a computer-readable recording medium having recorded thereon a program for realizing the method.

1 is a diagram illustrating an embodiment of primary and secondary domains for optimizing a wireless network according to the present invention.

2A is a flowchart illustrating an embodiment of a wireless network optimization method using secondary domain optimization according to the present invention.

2B is a program list for a wireless network optimization method applying secondary domain optimization according to the present invention.

3A is a flowchart illustrating an embodiment of a method for optimizing a wireless network without applying secondary domain optimization according to the present invention.

3B is a program list for a wireless network optimization method without applying secondary domain optimization according to the present invention.

4A and 4B are comparative explanatory diagrams of coverage optimization efficiency according to calculation amount.

According to an aspect of the present invention, there is provided a wireless network optimization method applied to a wireless network optimization system, comprising: dividing an entire optimization target region into domains, and then calculating a goodness of fit for each of the divided domains. Stage 1; Until the goodness-of-fit for the entire optimization target region is greater than or equal to the predetermined threshold value, the domain having the lowest goodness of fit among the divided domains is designated as a primary domain and the primary domain is designated as a target domain. A second step of optimizing the designated target domain using a genetic algorithm after designating a target domain; And if the goodness of fit of the primary domain does not improve beyond the average fit value of the entire domain and exceeds the predetermined number of gene evolution generations during the optimization of the second stage, the lowest among the secondary domains of the base station currently being optimized. And selecting a secondary domain having a goodness of fit and optimizing one by one using a genetic algorithm, and returning to the second step if the goodness of fit of the primary domain is equal to or greater than the average of all domains.

Meanwhile, the present invention provides a method for optimizing a wireless network, comprising: a first step of dividing an entire optimization target region into domains and then calculating suitability for each of the divided domains; And assigning a target domain from a domain having the lowest goodness of fit among the divided domains until the fitness for the entire optimization target region is equal to or greater than a predetermined threshold value, and then optimizing the designated target domain using a genetic algorithm. It includes a second step.

On the other hand, the present invention, in order to optimize the wireless network, in a wireless network optimization system having a processor, the first function of dividing the entire optimization target region into domains, and then calculating the suitability for each of the divided domains ; Until the goodness-of-fit for the entire optimization target region is greater than or equal to the predetermined threshold value, the domain having the lowest goodness of fit among the divided domains is designated as a primary domain and the primary domain is designated as a target domain. A second function of optimizing the designated target domain using a genetic algorithm after designating a target domain; And when the goodness of fit of the primary domain does not improve beyond the average fitness value of the entire domain and exceeds the predetermined gene evolution generation number during the optimization of the second function, the lowest among the secondary domains of the base station currently being optimized. Select from secondary domains having goodness-of-fit and optimize them using genetic algorithms one by one, and if the goodness-of-fit of the primary domain is above the average of all domains, read the program for realizing the third function to return to the second function. Provides a record medium that can be.

On the other hand, the present invention, in order to optimize the wireless network, in a wireless network optimization system having a processor, after dividing the entire area to be optimized into domains, and calculating the suitability for each of the divided domains function; And assigning a target domain from a domain having the lowest goodness of fit among the divided domains until the fitness for the entire optimization target region is equal to or greater than a predetermined threshold value, and then optimizing the designated target domain using a genetic algorithm. A computer readable recording medium having recorded thereon a program for realizing a second function is provided.

The present invention minimizes the computational complexity required to achieve meaningful results by dividing and optimizing the optimization problem into smaller problems. The optimization method corresponding to the present invention is directed to optimization objectives for commonly used wireless networks. All are available, and the amount of computation does not change significantly depending on the optimization goal chosen.

Hereinafter, the description will be made based on coverage optimization, which is an optimization target of the most commonly used wireless network.

In this case, the objective function can be defined as follows.

Maximize

f (x ₁ , x ₂ , x ₃ , ..., x _n-1 , x _n )

Subject to:

∀i, x _i ≥l _i , x _i ≤ k _i

Here, x _i denotes a parameter of an antenna to be optimized, f () denotes a function of receiving parameters to be optimized and returning coverage of the target area, and n is determined according to the target antenna and the number of parameters to be optimized.

The objective function above is to find a combination of parameters that maximizes the function f (). l_iAnd k_iIs the parameter x_iIs determined by the type and other realistic restrictions. For example, the tilt angle of the antenna may be limited according to the installation position of the antenna. The size of the solution set of the optimization problemorBecomes Where R_iIs a parameter x_iPrecision,AIs the number of antennas installed in the base station, M is the number of base stations, P is the number of parameters of the antenna to be optimized.

In the present invention, it is assumed that the position and initial antenna parameters of each base station are already determined. The optimization method corresponding to the present invention first begins by dividing the wireless network into smaller sub-areas called "domains."

The domain includes a base station to be subjected to an optimization, and the suitability of the domain is determined by the parameters of the base station as the target, and based on the service coverage value of the domain area, The parameter will be determined. At this time, the base station where the antennas to be optimized are installed is defined as the "target base station" of the domain. Therefore, in performing the wireless network optimization, the number of domains becomes equal to the number of base stations to be optimized.

However, in reality, it is not easy to find an area where service coverage is determined by only one single base station. This is because, in general, the signal strength that determines whether a communication service at any one point is affected by signals from neighboring base stations. To solve this problem, use "nested domains".

Thus, in performing wireless network optimization, one target base station has one "primary domain" and one or more "secondary domains", where the primary domain overlaps one or more secondary domains. In this case, the primary domain is a domain of the target base station, and the secondary domain is a domain of base stations adjacent to the primary domain. Therefore, the number of secondary domains of either base station is the number of base stations adjacent to the target base station.

In the present invention, a domain for optimizing a base station to be optimized is defined as a convex hull including an adjacent base station. You can use a more complex method like "minimum maximal independent set" to determine the domain, but the method of determining the domain depends on the goal you want to optimize.

The wireless network optimization evaluates the coverage (i.e. serviceability ratio) of all domains to be optimized using an evaluation function, averages all domains, and then selects a parameter of a target base station of a domain whose coverage of the domain area is below the average. By changing them.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram illustrating an embodiment of primary and secondary domains for optimizing a wireless network according to the present invention.

In the figure, the shaded area surrounded by the solid line around the base station B (i, i) is the primary domain 11 of the base station B (i, i).

The primary domain region 11 of B (i, i) is a convex hull containing all adjacent base stations. Another hexagonal shaded area surrounded by a dotted line is the secondary domain 12 of B (i, i) and at the same time the primary domain of B (i-1, i + 1). In the figure, the area 13 shown in bold represents the overlapping area between the primary domain 11 and the secondary domain 12 of B (i, i).

After calculating the service coverage of all domains subject to optimization, if the primary domain 11 of base station B (i, i) has the lowest goodness of fit among the domains, then B (i i) Optimization of the parameters will be performed first.

The wireless network optimization system to which the present invention is applied uses a genetic algorithm to optimize each domain. Using genetic algorithms, the goodness of fit of the primary domain is above the mean of all primary domains, or until the number of evolutionary generations of the gene reaches a predefined threshold.

If the goodness-of-fit has not improved above the average and the number of evolutionary generations has reached a threshold, this is determined to be due to an improperly optimized (i.e. not optimized) parameter of the neighboring base station, and the secondary domains of the current target base station. We select from the secondary domains with the lowest goodness of fit and optimize them using genetic algorithms one by one until the goodness of fit of the primary domain is above average. This wireless network optimization algorithm is as shown in FIG.

2A is a flowchart illustrating an embodiment of a wireless network optimization method using secondary domain optimization according to the present invention.

The wireless network optimization method of FIG. 2A optimizes one domain at a time. In this case, the domain to be optimized is defined as a target domain.

The location and configuration parameters of each base station are assumed to have an initial value determined before the optimization method is performed.

After dividing the entire optimization target region into overlapping domains (200), the fitness of all the domains targeted for each optimization is calculated (202). Determine (204).

As a result of the determination, if the goodness of fit of the entire target area is greater than the threshold value, it ends. do.

That is, the domain having the lowest goodness of fit among the domains is designated as the primary domain (206), the primary domain is designated as the target domain (Target Domain) (208), and the designated target The setup of a target domain is optimized using a genetic algorithm (210).

Then, it is checked whether the goodness of fit of the primary domain is equal to or greater than the average value of the entire domain (212), and if it is equal to or greater than the average suitability value of the entire domain (hereinafter, simply referred to as the average value), it returns to "204". If it is less than the average value of the domain, the domain having the lowest goodness of fit among the secondary domains of the primary domain is designated as the target domain (214), and the target domain is set to the target algorithm. After optimizing using 216, it returns to "212" again.

The above description is briefly described as follows. First, the entire optimization target region is divided into domains, and then the fitness is calculated for each divided domain (step 1). Then, when the fitness for the entire optimization target region becomes equal to or greater than a predetermined threshold value. Until, the domain with the lowest goodness of the divided domains to the primary domain (primary domain) and the primary domain as the target domain (Target domain) and then designated target domain using a genetic algorithm to optimize (Step 2). If the goodness-of-fit of the primary domain does not improve above the average goodness-of-fit of the entire domain and exceeds a certain number of gene evolution generations, the second domain having the lowest goodness of fit among the secondary domains of the base station currently being optimized is selected. By optimizing using a genetic algorithm one by one, if the goodness of fit of the primary domain is above the average of the entire domain, the process returns to the second process (third process).

The wireless network optimization system to which the present invention is applied uses a genetic algorithm to optimize each domain. Using genetic algorithms, the goodness of fit of the primary domain is continued until it has a mean value above all primary domains or until the number of gene evolution generations reaches a predefined threshold.

If the goodness-of-fit has not improved above the average and the number of evolutionary generations has reached a threshold, this is determined to be due to an improperly optimized (i.e. not optimized) parameter of the neighboring base station, and the secondary domains of the current target base station. We select from the secondary domains with the lowest goodness of fit and optimize them using genetic algorithms one by one until the goodness of fit of the primary domain is above average.

This process is performed repeatedly until the goodness of fit of the whole region is equal to or greater than the threshold value. These termination conditions are taken as examples and in practice the optimization termination can be determined by other conditions such as the number of iterations or the elapsed time.

2B is a program list for a wireless network optimization method applying secondary domain optimization according to the present invention.

S is the fitness calculated by the evaluation function used for genetic algorithm application, S _Threshol is the fitness threshold, and S _Average is the average fitness of the whole domains. The subscript in S indicates the subject of the evaluation function. For example, S _TargetDomain represents the goodness of fit of the target domain ( TargetDoamin) .

N is the number of gene evolutionary generations, N _{TestedGenerations} is the number of evolutionary generations evaluated to date, N _Threshold is the threshold for the number of evolutionary generations, and S.min () returns the domain with the lowest fit among the domains to be optimized Indicates.

3A is a flowchart illustrating an embodiment of a method for optimizing a wireless network without applying secondary domain optimization according to the present invention.

If the primary domain's goodness of fit is not above average due to the optimization of the target base station, because of improperly set base stations in the nested secondary domain, the domain that targets this base station will also have a low goodness of fit value. Will also be selected and optimized. Therefore, it is possible to optimize the wireless network with almost the same efficiency more simply than the method of FIG. 2A.

Here, it is also assumed that the position and setting parameters of each base station are determined to have an initial value before the optimization is performed.

After dividing the entire optimization target region into domains (300), the fitness of all the domains targeted for each optimization is calculated (302), and it is determined whether the fitness of the entire optimization target region is greater than or equal to the threshold (304). .

As a result of the determination, if the goodness of fit of the entire optimization target region is equal to or greater than the threshold value, the process ends.

As a result of the determination, if the goodness of fit of the entire optimization target region is lower than the threshold value, optimization for the setting variables of the domain having the lowest goodness of fit is first performed.

That is, the domain having the lowest goodness of fit among the domains is designated as the target domain (Target Domain) (306), and the setting of the target domain (Target Domain) is optimized using the genetic algorithm (308), and then "304". Go back.

The above description is briefly described as follows. After dividing the entire optimization target region into domains, a goodness of fit is calculated for each of the divided domains. Then, until the goodness-of-fit for the entire optimization target region is greater than or equal to a certain threshold, the domain having the lowest goodness of fit among the divided domains is designated as the target domain, and then optimized using the genetic algorithm. It is.

The wireless network optimization system to which the present invention is applied uses a genetic algorithm to optimize each domain.

Using genetic algorithms, the goodness of fit of the target domain is greater than or equal to the mean value of the target domains or until the number of evolutionary generations of the gene reaches a predefined threshold.

In the present invention, when the fitness is not improved beyond the average value after domain optimization and the number of gene evolution generations reaches a threshold, the optimization for this domain is abandoned, and the other domain having the lowest fitness after the domain is selected again for optimization. do.

Also in this case, the above process is repeated until the goodness of fit of the entire region is greater than or equal to the threshold value. Such termination conditions are exemplified in the present invention, and in practice, optimization termination may be determined by other conditions such as the number of repetitions or the elapsed time.

3B is a program list for a wireless network optimization method without applying secondary domain optimization according to the present invention.

As described in FIG. 2B, S is a fitness calculated by an evaluation function used for genetic algorithm application, S _Threshol is a fitness threshold, and S _Average is an average goodness of fit of all domains. The subscript in S indicates the subject of the evaluation function. For example, S _TargetDomain represents the goodness of fit of the target domain ( TargetDoamin) .

N is the number of gene evolutionary generations, N _{TestedGenerations} is the number of evolutionary generations evaluated to date, and N _Threshold is the threshold for the number of evolutionary generations.

4A and 4B are comparative explanatory diagrams of coverage optimization efficiency according to calculation amount.

In solving the optimization problem by genetic algorithms, the large size of the population generally improves the algorithm's ability to search the solution set, which is advantageous for finding a search space with high evolutionary fit, but requires more computation. If the computational complexity of the evaluation function, such as wireless network optimization, is high, it can greatly affect the computation time required for optimization, and it can also act as a factor in reducing the convergence speed in convergence to the optimal solution.

In the optimization method corresponding to the present invention, it is not continuous parameter optimization but discontinuous parameter optimization. Since the optimization for one domain is continued after the optimization of the other domain, the suitability of the domain is affected and the correct solution is initially obtained. Hard to find.

In this case, since the solution set search by the genetic algorithm uses random search characteristics rather than emergence by the genetic algorithm, it is not meaningful to search for a more accurate solution by using various entities in the beginning. Thus, in this case, reducing the number of initial populations is efficient and helps to shorten the computation time.

However, as the optimization proceeds, the change of the parameters of the target base stations in the other domains converges, so that the base station optimization by the small group of groups no longer progresses. Therefore, by increasing the size of the population gradually according to the speed of convergence, the optimization of the wireless network through more precise base station optimization is minimized, thereby minimizing the time required for unnecessary object evolution and goodness-of-fit check, thereby making the optimization speed faster.

In the figure, in order to evaluate the proposed optimization algorithm, the result of the optimization by applying the proposed optimization method and the previous optimization method, a simple genetic algorithm for the base station parameters of the entire area, and the entire area using the random search method The results of the optimization were compared by applying the base station parameters of.

4A and 4B show the result of comparing the optimization efficiency based on the service coverage and the calculation amount used as the target in the present invention, since the objective of the present invention is to minimize the computational complexity required to obtain a meaningful result.

For the simulation according to the present invention, 25 base stations are uniformly distributed in a free space region of 10 x 10 km ² , each base station is composed of 3 sectors, and the antenna direction and tilt can be adjusted. It is assumed that all antennas belonging to the same base station have the same parameters. The initial value for the parameters of the base station was determined by generating a random number. In addition, a free space propagation model is used for the simulation, and the coverage of the wireless communication service is determined by the signal energy per bit, that is, the E _b / N ₀ threshold.

For the optimization of coverage, the genetic algorithm developed for the optimization proposed above was used, and the evaluation function of the genetic algorithm was the percentage of coverage for the domain region.

The optimization of the wireless network using the genetic algorithm to be compared with the present invention used the simple genetic algorithm of [3 ~ 4,7], the population of the genetic algorithm is 25 population, the mutation rate (mutation rate) is 0.05, crossover (crossover) rate) was used. In addition, each parameter to be optimized is encoded with a gene with 8-bit precision. The criterion of goodness of fit is the percentage of coverage for the entire area.

In FIG. 4A, the optimization efficiency was compared in the short term, and in FIG. 4B, the optimization efficiency was compared in the longer term. In the comparison result, the method 42 applying the conventional genetic algorithm shows an improved result than the random search 43, but the present invention 41 has a very small amount of calculation compared to the conventional method 42. You can see that it performs excellent optimization.

As described above, the method of the present invention may be implemented as a program and stored in a recording medium (CD-ROM, RAM, floppy disk, hard disk, magneto-optical disk, etc.) in a computer-readable form.

The present invention described above is capable of various substitutions, modifications, and changes without departing from the spirit of the present invention for those skilled in the art to which the present invention pertains, and the above-described embodiments and accompanying It is not limited by the drawings.

As described above, the present invention has an effect of automating a wireless network optimization problem in which a problem is too complicated in the current technology and an unrealistic time is required for the optimization calculation.

In addition, the present invention, as a specialized genetic algorithm for wireless network optimization, there is an effect that can perform very excellent optimization with a very small amount of calculation than the existing genetic algorithm.

In addition, the present invention has the effect of minimizing the computational complexity required to achieve meaningful results by optimizing the optimization problem into smaller problems.

In addition, according to the present invention, since the change in the parameters of the target base station in the other domain converges as the optimization proceeds, the base station optimization by the small number of individual groups no longer progresses, and thus the size of the population according to the convergence speed is increased. By gradually increasing the wireless network optimization, the optimization speed is minimized by minimizing the time required for unnecessary object evolution and fitness check.

Claims (7)

In the wireless network optimization method applied to the wireless network optimization system, Dividing an entire optimization target region into domains, and then calculating a goodness of fit for each of the divided domains; Until the goodness-of-fit for the entire optimization target region is greater than or equal to the predetermined threshold value, the domain having the lowest goodness of fit among the divided domains is designated as a primary domain and the primary domain is designated as a target domain. A second step of optimizing the designated target domain using a genetic algorithm after designating a target domain; And If the goodness of fit of the primary domain does not improve beyond the average fit of all domains and exceeds the predetermined number of gene evolution generations during the optimization process of the second stage, the lowest goodness of fit among the secondary domains of the base station currently being optimized A third step of returning to the second step if the fitness of the primary domain is higher than the average of all domains by optimizing one by one from a secondary domain having Wireless network optimization method using a domain optimization comprising a. In the wireless network optimization method applied to the wireless network optimization system, Dividing an entire optimization target region into domains, and then calculating a goodness of fit for each of the divided domains; And Until the goodness-of-fit for the entire optimization target region is greater than or equal to a predetermined threshold, the target domain is designated from the domain having the lowest goodness-of-fit value among the divided domains and then optimized using the genetic algorithm. 2nd step Wireless network optimization method using a domain optimization comprising a. The method of claim 2, The optimization process of the target domain of the second step, If the fitness of the optimized target domain does not improve beyond the average fitness value within a predetermined number of gene evolution generations, the optimization of the target domain is abandoned, and the domain having the lowest fitness after the target domain is designated as the target domain. Wireless network optimization method using domain optimization, characterized in that. The method according to claim 1 or 2, The goodness of fit of the domain, Wireless network optimization method using domain optimization, characterized in that determined by the parameters of the base station to be optimized. The method according to claim 1 or 2, The division of the entire optimization target region of the first step is The wireless network optimization method using domain optimization, characterized in that to divide the entire optimization target region into an overlapping domain. In order to optimize the wireless network, in a wireless network optimization system having a processor, A first function of dividing an entire optimization target region into domains, and then calculating a goodness of fit for each of the divided domains; Until the goodness-of-fit for the entire optimization target region is greater than or equal to the predetermined threshold value, the domain having the lowest goodness of fit among the divided domains is designated as a primary domain and the primary domain is designated as a target domain. A second function of optimizing the designated target domain using a genetic algorithm after designating a target domain; And During the optimization of the second function, if the fitness of the primary domain does not improve above the average fitness of the entire domain and exceeds the predetermined number of gene evolution generations, the lowest fitness of the secondary domains of the base station currently being optimized A third function of returning to the second function if the fitness of the primary domain is equal to or greater than the average of all domains by optimizing one by one from a secondary domain having A computer-readable recording medium having recorded thereon a program for realizing this. In order to optimize the wireless network, in a wireless network optimization system having a processor, A first function of dividing an entire optimization target region into domains, and then calculating a goodness of fit for each of the divided domains; And Until the goodness-of-fit for the entire optimization target region is greater than or equal to a predetermined threshold, the target domain is designated from the domain having the lowest goodness-of-fit value among the divided domains and then optimized using the genetic algorithm. 2nd function A computer-readable recording medium having recorded thereon a program for realizing this.