CN105634593A - Indoor visible light communication LED array layout optimization method based on genetic algorithm - Google Patents

Abstract

The invention relates to an indoor visible light communication LED array layout optimization method based on a genetic algorithm. The method comprises following steps: 1, constructing chromosomes according to the number of LED arrays and the length and the width of a room; 2, creating an initial population by using the chromosomes obtained in the previous step; 3, calculating the fitness value of each individual in the population, judging whether an algorithm ending condition is satisfied; 4, carrying out selection, intersection and mutation operations to each individual in the population, thus generating a new population; and 5, analyzing the optimized LED array layout according to the optimum individuals obtained by the genetic algorithm. According to the algorithm of the invention, the problem that the receiving powers received on the receiving plane in the existing LED layout are distributed unevenly can be solved; the optimized LED array layout enables the receiving powers on the receiving plane to be distributed evenly; and the reliability problem of the LED communication is solved.

Description

A kind of indoor visible light communication LED array layout optimization method based on genetic algorithm

Technical field

The invention belongs to indoor visible light communication technical field, relate to the LED array layout optimization method based on genetic algorithm in a kind of visible light communication.

Background technology

Progressively become the hot-candidate technology of following the short distance wireless communication technology with features such as its good confidentiality, frequency spectrum resource are abundant, harmless by visible light communication technology (VLC, VisibleLightCommunication). In indoor VLC system, due to the factor such as reflection characteristic of room-sized, the Multipath Transmission of signal and body surface, the difference at communication layer receive that position receives luminous power can there were significant differences, thus seriously restricting the communication performance of VLC system. Under indoor environment, how obtaining consistent power and illumination distribution in all of position that accepts, the reliability for improving communication is significant.

Indoor white light LEDs (LightEmittingDiode) array of source layout determines the spatial distribution of indoor light intensity and luminous power. So, during design white light LEDs visible light communication system, illumination and the double requirements communicated should be taken into account. Appropriate design array of source, makes the illuminance of indoor meet standard of illumination, makes optical power distribution present in indoor simultaneously and be uniformly distributed. Standard according to internationalization illumination, the illuminance of office requires between 300-1500 lux.

Summary of the invention

In view of this, it is an object of the invention to provide a kind of indoor visible light communication LED array layout optimization method based on genetic algorithm, this algorithm can solve the problem that under existing indoor LED array layout, and receiving plane receives the problem that power is uneven. LED array layout after optimization can reduce the reception power swing in receiving plane, improves the reliability of communication system.

For reaching above-mentioned purpose, the present invention provides following technical scheme: based on the LED array layout optimization method of genetic algorithm in a kind of visible light communication, this algorithm comprises the following steps:

(1) with the length in room for x-axis, wide set up coordinate system for y-axis, build gene. Length information L and width information W according to room come structural gene storehouse, and then build chromosome, and create initial population.

(2) arranging evolutionary generation is t, and initializes evolutionary generation t=0;

(3) according to formula F (i)=dif (i, j)/E{ [P_r-E(P_r)]²Calculate the fitness value of each individuality in population; Wherein, i is that the i-th in population is individual, and (i j) represents two individual difference value, E (P to dif_r) represent and reception power receiving point all in receiving plane are asked expectation.

(4) according to the selected probability of each individualityPerform to select operation.

(5) judge that whether current evolutionary generation t is more than maximum evolutionary generation t_max, or continuous t_contDo not change for the fitness value of optimum individual in population; Meet one condition of any of the above and then forward step (8) to; If being unsatisfactory for one condition of any of the above, then make evolutionary generation t=t+1.

(6) to the individuality in population according to crossover probability P_croPerform to intersect and operate;

(7) with the individuality in population according to mutation probability P_mutPerform mutation operation, return step (3);

(8) find out the individuality that in population, fitness value is maximum, decode the best coordinates of each LED array according to this individuality, be the LED array layout after optimization.

Further, the initialization of population scheme described in step (1) comprises the following steps:

(1.1) each individuality (also referred to as chromosome) in population has following form:

CS=((x₁,y₁),(x₂,y₂),...,(x_N,y_N))

Wherein, N represents the LED array number in room on ceiling, (x_i,y_i) representing a gene, x and y represents abscissa and the vertical coordinate of LED array respectively, and the span of x and y is x_i��geneLibx,y_i�� geneLiby, meets following formula:

$geneLibx=\{0,1\×\frac{L}{\mathrm{\ξ}},2\×\frac{L}{\mathrm{\ξ}},...,\mathrm{\ξ}\×\frac{L}{\mathrm{\ξ}}\}$

$geneLiby=\{0,1\×\frac{W}{\mathrm{\μ}},2\×\frac{W}{\mathrm{\μ}},...,\mathrm{\μ}\×\frac{W}{\mathrm{\μ}}\}$

Wherein �� and �� represents the grid number needing the length L and width W in room to be subdivided into.

(1.2) when initializing population, according to (1.1), N is built_popuIndividuality. Wherein, the abscissa x in each genes of individuals position randomly selects from gene bank geneLibx, and the vertical coordinate y in gene position randomly selects from gene bank geneLiby.

Further, the fitness computational methods detailed process described in step (3) is:

(3.1) length in room is L, and width is W, is highly H, and receiving plane height is h. Receiving plane is divided into �� �� �� grid. The reception power on each mesh point is calculated according to below equation:

$P_r=P_t\frac{(m+1)E}{2{{\mathrm{\πD}}_d}^2}{\cos }^m\left(\mathrm{\φ}\right)T_s\left(\mathrm{\ψ}\right)g\left(\mathrm{\ψ}\right)cos\left(\mathrm{\ψ}\right)$

Wherein E is the PD surface area receiving the inside detector, D_dIt is the air line distance between access point and PD, T_s(��) gain of optical filter, �� is the angle of incidence of PD,Being LED emission angle, g (��) is optical concentrator gain. LED half-power angle ��_1/2Determining lambert's exponent m of light source, the conversion relation between them is: m=-ln2/ln (cos (��_1/2))��

(3.2) expectation is asked can to obtain E (P all of mesh point_r). (i, j) represents two individual difference value to dif, and j individuality has the highest fitness value in current population. Calculate according to the following formula dif (i, j):

$dif(i,j)=\underset{k=1}{\overset{N}{\Σ}}\sqrt{{(x_{ik}-x_{jk})}^2+{(y_{ik}-y_{jk})}^2}/(N\×\sqrt{L^2+W^2})$

(3.3) introduce individual variation degree function as the weighting function of fitness function, can effectively prevent the disappearance of effective gene in Evolution of Population process to be absorbed in local optimum.

Further, " the continuous t described in step (5)_contDo not change for the fitness value of optimum individual in population " detailed process be:

(5.1) design variable F_lastRepresent the preferably individual fitness value in previous generation population, F_crrtRepresent the best ideal adaptation angle value of this generation population.

(5.2) if F_crrtMore than F_last, then t is made_cont=0, F_last=F_crrt; Otherwise, t is made_cont=t_cont+1,F_last=F_crrt;

Further, " to the individuality in population according to crossover probability P in step (6)_croPerform intersect operation " detailed process be:

(6.1) from population, two chromosome CS1 and CS2 are selected at random, then according to following formula is to each Chromosome segment:

Segment1=(x₁,y₁)...(x_N/4,y_N/4)

Segment2=(x_N/4+1,y_N/4+1)...(x_N/2,y_N/2)

Segment3=(x_N/2+1,y_N/2+1)...(x_3N/4,y_3N/4)

Segment4=(x_3N/4,y_3N/4)...(x_N,y_N)

(6.2) one section of Segment is randomly selected, then the gene on exchange correspondence position. Altogether needing to carry out the above-mentioned number of times operated that intersects is N_popu*P_cro��

Further, " with the individuality in population according to mutation probability P in step (7)_mutPerform mutation operation " detailed process be:

(7.1) search volume (plane at ceiling place) of genetic algorithm is divided into N_ar*N_arIndividual uniform zonule, each region representation is Ar (k);

(7.2) a chromosome CS is randomly choosed_mut, and add up the chromosome CS in region Ar (k)_mutGene number be set to n (k);

(7.3) n (k) is sorted, Ar_maxTo there being n in region_maxIndividual gene, Ar_minTo there being n in region_minIndividual gene;

(7.4) random erasure one belongs to Ar_maxGene, and method described in step (1) generates one and belongs to Ar_minGene, and be inserted in chromosome corresponding position.

(7.5) repeat to be above operation N_popu*P_mutSecondary, namely complete mutation operation.

The beneficial effects of the present invention is: the algorithm of the present invention can solve the problem that the reception power problem pockety received in receiving plane under existing LED layout, LED array layout after optimization so that the reception power in receiving plane is evenly distributed, can solve the integrity problem of LED communication.

Accompanying drawing explanation

In order to make the purpose of the present invention, technical scheme and beneficial effect clearly, the present invention provides drawings described below to illustrate:

Fig. 1 is the LED array layout optimization method flow chart based on genetic algorithm;

Fig. 2 is typical visible ray indoor communication system;

Fig. 3 is typical rectangular LED arrays layout and power distribution thereof;

Fig. 4 is the LED array layout after optimizing and power distribution thereof.

Detailed description of the invention

Below in conjunction with accompanying drawing, the preferred embodiments of the present invention are described in detail:

Fig. 2 is typical visible ray indoor communication system, Fig. 1 is the LED array layout optimization method flow chart based on genetic algorithm, as it can be seen, a kind of indoor visible light communication LED array layout optimization method based on genetic algorithm provided by the invention, this algorithm comprises the following steps:

(1) typical indoor visible light communication system is as in figure 2 it is shown, the length information L=5 rice in room, width information W=5 rice, is highly 3 meters, and receiving plane is 0.85 meter high. Shown in Fig. 2, build coordinate axes, be alongst designated x-axis, be designated y-axis along width.

(2) utilizing the information in step (1) to build chromosome, each individuality (also referred to as chromosome) in population has following form:

CS=((x₁,y₁),(x₂,y₂),...,(x_N,y_N))

Wherein, N represents the LED array number in room on ceiling, N=16 in this example. (x_i,y_i) representing a gene, x and y represents abscissa and the vertical coordinate of LED array respectively, and the span of x and y is x_i��geneLibx,y_i�� geneLiby, meets following formula:

$geneLibx=\{0,1\×\frac{L}{\mathrm{\ξ}},2\×\frac{L}{\mathrm{\ξ}},...,\mathrm{\ξ}\×\frac{L}{\mathrm{\ξ}}\}$

$geneLiby=\{0,1\×\frac{W}{\mathrm{\μ}},2\×\frac{W}{\mathrm{\μ}},...,\mathrm{\μ}\×\frac{W}{\mathrm{\μ}}\}$

Wherein �� and �� represents the grid number needing the length L and width W in room to be subdivided into, and it can affect the precision of the LED array layout searched, it is contemplated that is actually needed, and making them here respectively is all 50.

(3) when initializing population, according to step (2), N is built_popuIndividuality. The size of population influences whether that can genetic algorithm converge to optimal solution and converge to the speed of optimal solution. Consider to be actually needed, here by N_popuIt is set to 200. Wherein, the abscissa x in each genes of individuals position randomly selects from gene bank geneLibx, and the vertical coordinate y in gene position randomly selects from gene bank geneLiby.

(4) according to the selected probability of each individualityPerform to select operation.

(5) arranging evolutionary generation is t, and initializes evolutionary generation t=0.

(6) length in room is L, and width is W, is highly H, and receiving plane height is h. Receiving plane is divided into �� �� �� grid. The division of sizing grid will influence whether the search precision of genetic algorithm, it is contemplated that the room of this example is of a size of 5 meters * 5 meters, therefore room is divided into 50*50 grid. The reception power on each mesh point is calculated according to below equation:

$P_r=P_t\frac{(m+1)E}{2{{\mathrm{\πD}}_d}^2}{\cos }^m\left(\mathrm{\φ}\right)T_s\left(\mathrm{\ψ}\right)g\left(\mathrm{\ψ}\right)cos\left(\mathrm{\ψ}\right)$

Wherein E is the PD surface area receiving the inside detector, D_dIt is the air line distance between access point and PD, T_s(��) gain of optical filter, �� is the angle of incidence of PD,Being LED emission angle, g (��) is optical concentrator gain. LED half-power angle ��_1/2Determining lambert's exponent m of light source, the conversion relation between them is: m=-ln2/ln (cos (��_1/2)). Shown in the parameter table specific as follows used in this example:

The parameter used in table 1 VISIBLE LIGHT SYSTEM

Symbol	Explanation	Value
			L*W*H	Room-sized	5*5*3m
h	Receiving plane height	0.85m
			N	LED array number	16
Pt	LED array transmit power	452mW
			��1/2	Half-power angle	80deg.
�C	LED chip number in LED array	7��7
			�C	Interval between LED chip	0.01m
Ts(��)	Filter gain	1.0
			��c	The angle of visual field	55deg.
A	Detector surface is amassed	1.0cm2
			�C	Receiver photoelectric transformation efficiency	0.53A/W
n	PD lens reflex coefficient	1.5
			I(0)	Central illumination intensity	23.81cd

(7) expectation is asked can to obtain E (P all of mesh point_r). (i, j) represents two individual difference value to dif, and j individuality has the highest fitness value in current population. Calculate according to the following formula dif (i, j):

$dif(i,j)=\underset{k=1}{\overset{N}{\Σ}}\sqrt{{(x_{ik}-x_{jk})}^2+{(y_{ik}-y_{jk})}^2}/(N\×\sqrt{L^2+W^2})$

(8) introduce individual variation degree function as the weighting function of fitness function, can effectively prevent the disappearance of effective gene in Evolution of Population process to be absorbed in local optimum.

(9) according to formula F (i)=dif (i, j)/E{ [P_r-E(P_r)]²Calculate the fitness value of each individuality in population; Wherein, i is that the i-th in population is individual, and (i j) represents two individual difference value, E (P to dif_r) represent and reception power receiving point all in receiving plane are asked expectation.

(10) design variable F_lastRepresent the preferably individual fitness value in previous generation population, F_crrtRepresent the best ideal adaptation angle value of this generation population.

(11) if F_crrtMore than F_last, then t is made_cont=0, F_last=F_crrt; Otherwise, t is made_cont=t_cont+1,F_last=F_crrt;

(12) judge that whether current evolutionary generation t is more than maximum evolutionary generation t_max=100, or continuous t_contIn=15 generation populations, the fitness value of optimum individual does not change; Meet one condition of any of the above and then forward step (20) to; If being unsatisfactory for one condition of any of the above, then make evolutionary generation t=t+1. t_maxAnd t_contGenetic algorithm will be affected converge to the speed of optimal solution, it is contemplated that problem scale, here they will be respectively set to 100 and 15.

(13) from population, two chromosome CS1 and CS2 are selected at random, then according to following formula is to each Chromosome segment:

Segment1=(x₁,y₁)...(x_N/4,y_N/4)

Segment2=(x_N/4+1,y_N/4+1)...(x_N/2,y_N/2)

Segment3=(x_N/2+1,y_N/2+1)...(x_3N/4,y_3N/4)

Segment4=(x_3N/4,y_3N/4)...(x_N,y_N)

(14) one section of Segment is randomly selected, then the gene on exchange correspondence position. Altogether needing to carry out the above-mentioned number of times operated that intersects is N_popu*P_cro. In order to meet the needs of genetic algorithm convergence rate, the span of crossover probability is generally between 0.6��0.9. Take crossover probability P in this example_cro=0.6.

(15) search volume (plane at ceiling place) of genetic algorithm is divided into N_ar*N_arIndividual uniform zonule, each region representation is Ar (k). Dividing of region can be determined according to the size of search volume. Search volume in this example is 5 meters * 5 meters, therefore the zoning in setting search space is 5*5;

(16) a chromosome CS is randomly choosed_mut, and add up the chromosome CS in region Ar (k)_mutGene number be set to n (k);

(17) n (k) is sorted, Ar_maxTo there being n in region_maxIndividual gene, Ar_minTo there being n in region_minIndividual gene;

(18) random erasure one belongs to Ar_maxGene, and method described in step (1) generates one and belongs to Ar_minGene, and be inserted in chromosome corresponding position.

(19) repeat to be above operation N_popu*P_mutSecondary, namely complete mutation operation. In order to be able to make genetic algorithm stably search a feasible solution, mutation probability can not obtain too big, is typically in less than 0.01. This example takes mutation probability P_mut=0.01. Return step (8).

(20) find out the individuality that in population, fitness value is maximum, decode the best coordinates of each LED array according to this individuality, be the LED array layout after optimization.

(21) the reception power in common indoor LED array layout and receiving plane thereof is as shown in Figure 3. As seen from the figure, under this layout, the reception power distribution in receiving plane is extremely uneven, presents the feature that middle high surrounding is low. LED array layout and power thereof after genetic algorithm optimization are distributed as shown in Figure 4. In the diagram, hence it is evident that visible, the power distribution that receives after optimization has less undulatory property and more balanced, and this is significant to the reliability improving communication.

What finally illustrate is, preferred embodiment above is only in order to illustrate technical scheme and unrestricted, although the present invention being described in detail by above preferred embodiment, but skilled artisan would appreciate that, in the form and details it can be made various change, without departing from claims of the present invention limited range.

Claims (6)

1. the indoor visible light communication LED array layout optimization method based on genetic algorithm, it is characterised in that: the method comprises the following steps:

(1) determine the length and width in room, and build the gene information needed for genetic algorithm with this; The some LED array numbers arranged according to above-mentioned gene information and ceiling build chromosome, initialize species information;

(2) the receiving plane height in room is set, and receiving plane is divided into several grids, calculate the reception power on each grid, and obtain expectation and the variance of all grids reception power in whole receiving plane respectively;

(3) expectation described in step (2) and variance is utilized to calculate fitness value F (i) of each individuality in population; According to the selected probability of each individualityPerform to select operation;

(4) judge that whether current evolutionary generation t is more than maximum evolutionary generation t_max, or continuous t_contDo not change for the fitness value of optimum individual in population; Meet one condition of any of the above and then forward step (7) to; If being unsatisfactory for one condition of any of the above, then evolutionary generation t is made to add 1;

(5) individuality in population is performed intersection operation according to crossover probability; Perform mutation operation with the individuality in population according to mutation probability, return step (3);

(6) find out the individuality that in population, fitness value is maximum, decode the best coordinates of each LED array according to this individuality, be the LED array layout after optimization.

2. a kind of indoor visible light communication LED array layout optimization method based on genetic algorithm according to claim 1, it is characterised in that: in described step (1), with the length in room for x-axis, wide set up coordinate system for y-axis, build gene; Length information L and width information W according to room come structural gene storehouse, and then build chromosome, and create initial population; Arrange and initialize evolutionary generation t.

3. a kind of indoor visible light communication LED array layout optimization method based on genetic algorithm according to claim 1, it is characterized in that: the fitness computational methods detailed process described in step (3) is: according to formula F (i)=dif (i, j)/E{ [P_r-E(P_r)]²Calculate the fitness value of each individuality in population; Wherein, i is that the i-th in population is individual, and (i j) represents two individual difference value, E (P to dif_r) represent and reception power receiving point all in receiving plane are asked expectation.

4. a kind of indoor visible light communication LED array layout optimization method based on genetic algorithm according to claim 1, it is characterised in that: " the continuous t described in step (4)_contDo not change for the fitness value of optimum individual in population " detailed process be:

(4.1) design variable F_lastRepresent the preferably individual fitness value in previous generation population, F_crrtRepresent the best ideal adaptation angle value of this generation population;

(4.2) if F_crrtMore than F_last, then t is made_cont=0, F_last=F_crrt; Otherwise, t is made_cont=t_cont+1,F_last=F_crrt��

5. a kind of indoor visible light communication LED array layout optimization method based on genetic algorithm according to claim 1, it is characterized in that: the detailed process " performing to intersect to operate according to crossover probability to the individuality in population " in step (5) is: select two chromosomes at random from population according to crossover probability, and to the geometrical feature according to room to Chromosome segment, then one section of chromosome therein is selected, the gene on allele that exchange is corresponding.

6. a kind of indoor visible light communication LED array layout optimization method based on genetic algorithm according to claim 1, it is characterized in that: the detailed process " performing mutation operation with the individuality in population according to mutation probability " in step (5) is search space partition region to genetic algorithm, and add up the gene number on each region, find out and comprise certain region that gene number is maximum, and one of them gene of random erasure; Find out and comprise certain region that gene number is minimum, insert a gene belonging to this region.