CN104994515A - Gateway deploying method in cyber physical system - Google Patents

Abstract

The invention provides a gateway deploying method in a cyber physical system and relates to a gateway deploying method. By the multi-interval perturbation technology, a problem is overcome that the current particle swam optimization algorithm easily falls into the local optimum trap when searching gateway deploying, thereby further balancing load among all gateway nodes and increasing network service quality. The method comprises steps of initializing gateway particles; calculating flight of particles; updating history optimal position and local optimal position of particles; carrying out multi-interval perturbation for the local optimal position of the particles; and repeating the above steps until the maximum number of iteration P can be reached. The gateway position represented by the local optimal position pg is the optimal deploying position coordinate of the gateway in a wireless sensor network. The deploying position of the gateway in the cyber physical system is resolved by the multi-interval perturbation particle swarm optimization algorithm, so compared with the particle swarm-based algorithm, the rate of convergence of the algorithm is increased by about 30%, load balance level is increased by 130% and network service quality is remarkably increased.

Description

A kind of gateway deployment method under information physical emerging system

Technical field

The present invention relates to a kind of gateway deployment method under information physical emerging system.

Background technology

Information physical emerging system (Cyber Physical System, CPS) as a kind of novel intelligent system, merged by the organic and degree of depth of calculating, communication and control technology, realize brought close and the coordination of computational resource and physical resource, following CPS will be widely used in the numerous areas such as national defense industry and daily life.CPS forms a heterogeneous distributing system by the computing system of the communication network of isomery, isomery, the control system of isomery and heterogeneous physical equipment.

In information physical emerging system, cordless communication network as the carrier connecting various sensing equipment and wireless terminal, allow the seamless acquisition of various perception data and user whenever and wherever possible visit information become possibility.Under normal circumstances, sensor node in cordless communication network transfers data to router (Wireless Router, WR) after, again by gateway device (Gateway, GW) accessing Internet (Internet), realize the information sharing under wide area network, as shown in Figure 1.

Because most of traffic aggregation of network is in gateway, better from the node quality-of-service close to gateway, and poor from the node quality-of-service away from gateway, the unjustness of presence service quality between each node.Therefore, gateway often becomes the bottleneck of network performance, and it is disposed whether rationally has larger impact to network performance.

From mathematical problem angle, gateway deployment problem under information physical emerging system is actually the geometry K central issue in two-dimensional finite plane, find the optimal location of gateway, the covering radius reducing gateway is one of key problem of gateway deployment in information physical emerging system.In addition, because gateway needs, for multiple routing node provides data forwarding service, easily to become the network performance bottleneck of whole system simultaneously, how the load of each gateway node of active balance is also one of problem needing in gateway deployment process to pay close attention to.

At present, gateway deployment method under existing information physical emerging system realizes based on genetic algorithm (GA) and adaptive particle swarm optimization algorithm (APSO) mostly, local optimum trap is easily absorbed in when finding gateway deployment position, cause gateway covering radius excessive, network service quality declines; In addition, when finding gateway deployment scheme, usually not considering the load balance between each gateway, making in network, to occur that part gateway node load is higher, and the problem that other gateway node load is lighter, cause the overall performance of whole network system to decline.

Summary of the invention

The present invention proposes a kind of based on the gateway deployment method under the information physical emerging system of disturbance particle group optimizing between multi-region, utilizes perturbation technique between multi-region to overcome existing particle swarm optimization algorithm and is easily absorbed in the problem of local optimum when finding gateway deployment; Meanwhile, under the prerequisite reducing network coverage radius, the load further between each gateway node of balance, improves network service quality.

The present invention is that to solve the problems of the technologies described above the technical scheme taked as follows:

As shown in Figure 1, at router node (v ₁, v ₂..., v _n) quantity is arrange K gateway node (u in the wireless sensor network of n ₁, u ₂..., u _k), d (v _i, v _j) represent from node v _ibe routed to v _jrequired minimum hop count, shortest distance matrix can be tried to achieve by Floyd algorithm.

If router node v _iselect gateway node u _kas its service node, then claim v _ibe included in gateway u _kservices set U _kin, i.e. v _i∈ U _k.Now, u _kwith services set U _kultimate range between interior joint is be called gateway u _kcovering radius.Maximal cover radius in all gateway nodes be called gateway collection { u _k} _kcovering radius.

The covering radius of gateway collection is the important evaluation index of gateway deployment, and the less communication quality of covering radius is higher, and network service quality also can correspondingly improve.Therefore, in information physical emerging system, one of the important optimization target of gateway deployment is exactly make the covering radius of gateway collection minimum, shown in (1):

$\begin{array}{cc}\min & \underset{1\≤k\≤K}{\max }\underset{v_i\∈U_k}{\max }\left(d(v_i,u_k)\right)\\ s.t.& {\left\{u_k\right\}}_K\⋐R^2\end{array}---\left(1\right)$

In addition, for taking into full account the service quality of whole network, on the basis of covering radius optimizing gateway collection, the load level further between each gateway node of balance is needed.Therefore, in information physical emerging system, another optimization aim of gateway deployment is exactly make the load level gap between each node minimum, shown in (2):

min(load _max-load _min) (2)

Wherein, load _maxfor the most high capacity of gateway in whole information physical emerging system, load _minfor the minimum load of gateway in whole information physical emerging system, be the load capacity of each node of better quantitatively evaluating, the router node quantity usually adopting gateway node to serve is as its load capacity.

For solving the gateway Optimization deployment problem of information physical emerging system, the optimizing that the present invention proposes to use disturbance particle cluster algorithm between multi-region to carry out gateway location solves.For the gateway node of the K in information physical emerging system, its coordinate at two dimensional surface is: u _k=(a _k, b _k), k=1,2 ..., K is X=(x with the particle position of gateway coordinate composition ₁, x ₂..., x _m), particle rapidity is V=(v ₁, v ₂..., v _m), wherein m=2K, x _2k-1=a _k, x _2k=b _k, X is gateway group { u _k} _kvector representation form, the position of t particle is X _t=(x _{t, 1}, x _{t, 2}..., x _t,m), speed is Vt=(v _{t, 1}, v _{t, 2}..., v _t,m).

Its detailed process is as follows:

(1) gateway particle initialization

In network effective coverage [0, L], random generation T intended particle the initial position of composition first generation population, each particle in population represents one group of possible gateway location, produces the initial velocity of random each particle simultaneously the velocity amplitude scope of particle is set for [-v _max, v _max]=[-0.15 × L, 0.15 × L], inertia weight ω=0.729, Studying factors c ₁=c ₂=1.49, maximum iteration time P=500.

According to formula (3), t (t=1,2..., T) individual particle history optimal location p is set _tfor the initial position of this particle, particle global optimum position p is set according to formula (4) _goptimal location in all particle initial positions:

$p_t=X_t^1,t=\mathrm{1,2}...,T---\left(3\right)$

$p_g=\underset{X_t^1}{\arg \max }f\left(X_t^1\right),t=\mathrm{1,2}...,T---\left(4\right)$

Wherein, f () is adaptive value function, and its account form is such as formula shown in (10).

(2) particle flight calculating is carried out

For the p be made up of gateway location for any one object vector of population wherein: p=1,2 ..., P, t=1,2 ..., T.For t particle, calculate according to the speed flying speed that more new formula (5) carries out this particle of future generation ${V_t}^{p+1}=(v_{t,1}^{p+1},v_{t,2}^{p+1},...,v_{t,m}^{p+1}):$

$v_{t,j}^{p+1}=\mathrm{\ω}\×v_{t,j}^p+c_1\×\mathrm{rand}(p_t-x_{t,j}^p)+c_2\×\mathrm{rand}\×(p_g-x_{t,j}^p),j=\mathrm{1,2},...,m---\left(5\right)$

Wherein, rand is the random number between 0 ~ 1.

The value scope if the flying speed of particle outpaces, be then limited on boundary condition according to formula (6):

$v_{t,j}^{p+1}=\left\{\begin{array}{ccc}-v_{\max }& \mathrm{if}& v_{t,j}^{p+1}<-v_{\max }\\ v_{\max }& \mathrm{if}& v_{t,j}^{p+1}>v_{\max }\end{array}\right.---\left(6\right)$

The calculating of this particle position of future generation is carried out according to location updating formula (7)

$x_{t,j}^{p+1}=x_{t,j}^p+v_{t,j}^{p+1},j=\mathrm{1,2},...,m---\left(7\right)$

(3) more new particle history optimal location and particle global optimum position

According to the particle position after renewal particle history optimal location and particle global optimum position is calculated according to formula (8) and formula (9):

$p_t=\left\{\begin{array}{ccc}{p}_t& \mathrm{if}& f\left(p_t\right)\&le;f\left(X_t^{p+1}\right)\\ X_t^{p+1}& \mathrm{if}& f\left(p_t\right)>f\left(X_t^{p+1}\right)\end{array}\right.t=\mathrm{1,2},...,T---\left(8\right)$

$p_g=\left\{\begin{array}{ccc}{p}_g& \mathrm{if}& f\left(p_g\right)\&le;f\left(X_t^{p+1}\right)\\ X_t^{p+1}& \mathrm{if}& f\left(p_g\right)>f\left(X_t^{p+1}\right)\end{array}\right.t=\mathrm{1,2},...,T---\left(9\right)$

Wherein, f () is adaptive value function, and the gateway deployment method that the present invention proposes needs to consider gateway covering radius and gateway load equilibrium level, and therefore adopted adaptive value function account form is such as formula shown in (10):

$f\left({\left\{u_k\right\}}_K\right)=\underset{1\&le;k\&le;K}{\max }\underset{v_i\&Element;U_k}{\max }\left(d(v_i,u_k)\right)+\mathrm{\&lambda;}\&times;({\mathrm{load}}_{\max }-{\mathrm{load}}_{\min })---\left(10\right)$

{ u _k} _kbe of equal value with X, be all the variable representing gateway node position, only one is aggregate form, a number vector form, and aggregate form is in order to physical meaning is clear, and vector form is in order to mathematical notation is clear;

Wherein, λ is scale coefficient, because the importance of gateway covering radius index exceedes gateway load equilibrium level, so scale coefficient value is λ=0.1.

D (v _i, u _k) be sensor node v _ito gateway node u _kjumping figure.For a kth gateway u _k, its node location is (a _k, b _k), k=1,2 ..., K.Apart from this gateway node distance be less than communication radius sensor node composition set be Θ _k, this gateway node is to Θ _kthe jumping figure of middle arbitrary node is 1.Node v _ito Θ _kdistance be then node v _ito gateway node u _kjumping figure can calculate by formula (11):

$d(v_i,u_k)=\left\{\begin{array}{cc}1& v_i\&Element;{\mathrm{\&Theta;}}_k\\ \underset{v_l\&Element;{\mathrm{\&Theta;}}_k}{\min }\left(d(v_i,v_l)\right)+1& v_i\&NotElement;{\mathrm{\&Theta;}}_k\end{array}\right.---\left(11\right)$

The router node quantity that the load level of each gateway node is served according to it characterizes, and router node pays the utmost attention to the gateway node of selection within its coverage distance as service node.For the router node outside all gateway node coverage distances, then select the gateway node selecting jumping figure minimum by formula (11) as its service node.The load of whole information physical emerging system gateway node is (load ₁, load ₂..., load _k), then load _max=max (load ₁, load ₂..., load _k), load _min=min (load ₁, load ₂..., load _k).

(4) disturbance between multi-region is carried out to particle global optimum position

After carrying out particle history optimal location and particle global optimum location updating, in order to ensure that population is unlikely to be absorbed in rapidly local optimum process at gateway location searching process, between introducing multi-region, disruption and recovery is to particle global optimum position p _gbe optimized.Concrete grammar be whole iterative process is divided in early days, stage that mid-term, later stage three are different, make particle can by wave process close to global optimum by add different normal deviates period in different iterative computation.

In the stage in early days, because particle distance gateway global optimum positional distance is comparatively far away, therefore introduce larger normal deviate; At mid-term stage, introduce medium normal deviate; During the late stages of developmet, because particle distance gateway global optimum positional distance is comparatively near, less normal deviate is therefore introduced.Concrete perturbation scheme is such as formula shown in (12):

$p_g^{\&prime;}=\left\{\begin{array}{cc}N(p_g,{\mathrm{\&sigma;}}_1)& \mathrm{if}\frac{p}{P}\&le;{\mathrm{\&theta;}}_1\\ N(p_g,{\mathrm{\&sigma;}}_2)& \mathrm{if}{\mathrm{\&theta;}}_1<\frac{p}{P}\&le;{\mathrm{\&theta;}}_2\\ N(p_g,{\mathrm{\&sigma;}}_3)& \mathrm{if}\frac{p}{P}>{\mathrm{\&theta;}}_2\end{array}\right.---\left(12\right)$

Wherein, Discontinuous Factors σ ₁=r × 0.1, σ ₂=r × 0.01, σ ₃=r × 0.001, r is the coverage distance of gateway node.Interval parameter θ ₁=0.1, θ ₁=0.3.

Carrying out between multi-region after disturbance, according to the particle global optimum position p' after formula (10) calculation perturbation _gcorresponding adaptive value functional value f (p' _g), if f is (p' _g) <f (p _g), Ze Geng new particle global optimum position p _g=p' _g, otherwise, keep particle global optimum position p _gconstant.

(5) step (2) is repeated to step (4) until iterations arrives P=500, particle global optimum position p _grepresentative gateway location is the optimum deployed position coordinate of gateway in wireless sensor network.

The invention has the beneficial effects as follows:

The present invention utilizes perturbation technique between multi-region to overcome to utilize existing particle swarm optimization algorithm to be easily absorbed in the deficiency of local optimum when finding gateway deployment, simultaneously, under the prerequisite reducing network coverage radius, load between each gateway node of further balance, substantially increases network service quality.

Carry out the validity (see Fig. 2 and Fig. 3) of experimental verification method proposed by the invention by the following method, in an experiment, it is in the deployment region of 100*100 that router node is randomly dispersed in length.Network node quantity is T=100, and the router node maximum node number of degrees are 7, and minimum is that the deployment number of 1. gateways is divided into K=3, and experiment carries out 100 times.Algorithm the convergence speed is compared according to the minimum adaptive value of contemporary population and the minimum adaptive value of the 500th generation population and is obtained, and namely converges to P=500 the earliest and represents convergence rate (this experiment is corresponding with technical scheme in embodiment) for the age value of the minimum adaptive value of population.

Experimental result shows, by the deployed position of gateway in disturbance PSO Algorithm information physical emerging system between multi-region, can compare than the existing algorithm based on population, when covering radius system, convergence of algorithm speed improves about 30%, load balance level improves 130%, and therefore the method can significantly improve network service quality.

Accompanying drawing explanation

Fig. 1 is gateway deployment schematic diagram under information physical emerging system (in figure GW gateway device, WR represents router); Fig. 2 is that the router node under information physical emerging system disposes graph of a relation, and in figure, circle represents router node position, and line represents to there is correspondence between router node; Fig. 3 uses the information physical emerging system deployment architecture figure obtained after disturbance particle cluster algorithm carries out gateway deployment optimization between multi-region, wherein solid dot represents router node, the gateway node that box indicating is disposed, dotted line represents the correspondence between router, and solid line represents the correspondence between gateway node and router.

Embodiment

Gateway deployment method under a kind of information physical emerging system described in present embodiment, uses disturbance particle cluster algorithm between multi-region to carry out the optimizing of gateway location; Definition is for the gateway node of the K in information physical emerging system, and its coordinate at two dimensional surface is: (a _k, b _k), k=1,2 ..., K is X=(x with the particle position of gateway coordinate composition ₁, x ₂..., x _m), particle rapidity is V=(v ₁, v ₂..., v _m), wherein m=2K, x _2k-1=a _k, x _2k=b _k, the position of t particle is X _t=(x _{t, 1}, x _{t, 2}..., x _t,m), speed is V _t=(v _{t, 1}, v _{t, 2}..., v _t,m);

The implementation procedure of described method is as follows:

Step one, the initialization of gateway particle

In network effective coverage [0, L], random generation T intended particle the initial position of composition first generation population, each particle in population represents one group of possible gateway location, produces the initial velocity of random each particle simultaneously the velocity amplitude scope of particle is set for [-v _max, v _max]=[-0.15 × L, 0.15 × L], inertia weight is ω, and Studying factors is c ₁, c ₂, maximum iteration time is P;

According to formula (3), t (t=1,2..., T) individual particle history optimal location p is set _tfor the initial position of this particle, particle global optimum position p is set according to formula (4) _goptimal location in all particle initial positions:

$p_t=X_t^1,t=\mathrm{1,2}...,T---\left(3\right)$

$p_g=\underset{X_t^1}{\arg \max }f\left(X_t^1\right),t=\mathrm{1,2}...,T---\left(4\right)$

Wherein, f () is adaptive value function, and its account form is such as formula shown in (10);

{ u _k} _kbe represent all gateway node set, X is each node coordinate particular location in gateway node set;

Step 2, carry out particle flight calculating

The p be made up of gateway location is expressed as any one object vector of population: wherein: p=1,2 ..., P, t=1,2 ..., T; For t particle, calculate according to the speed flying speed that more new formula (5) carries out this particle of future generation ${V_t}^{p+1}=(v_{t,1}^{p+1},v_{t,2}^{p+1},...,v_{t,m}^{p+1}):$

$v_{t,j}^{p+1}=\mathrm{\&omega;}\&times;v_{t,j}^p+c_1\&times;\mathrm{rand}(p_t-x_{t,j}^p)+c_2\&times;\mathrm{rand}\&times;(p_g-x_{t,j}^p),j=\mathrm{1,2},...,m---\left(5\right)$

Wherein, rand is the random number between 0 ~ 1;

The value scope if the flying speed of particle outpaces, be then limited on boundary condition according to formula (6):

$v_{t,j}^{p+1}=\left\{\begin{array}{ccc}-v_{\max }& \mathrm{if}& v_{t,j}^{p+1}<-v_{\max }\\ v_{\max }& \mathrm{if}& v_{t,j}^{p+1}>v_{\max }\end{array}\right.---\left(6\right)$

The calculating of this particle position of future generation is carried out according to location updating formula (7)

$x_{t,j}^{p+1}=x_{t,j}^p+v_{t,j}^{p+1},j=\mathrm{1,2},...,m---\left(7\right)$

Step 3, more new particle history optimal location and particle global optimum position

According to the particle position after renewal particle history optimal location and particle global optimum position is calculated according to formula (8) and formula (9):

$p_t=\left\{\begin{array}{ccc}{p}_t& \mathrm{if}& f\left(p_t\right)\&le;f\left(X_t^{p+1}\right)\\ X_t^{p+1}& \mathrm{if}& f\left(p_t\right)>f\left(X_t^{p+1}\right)\end{array}\right.t=\mathrm{1,2},...,T---\left(8\right)$

$p_g=\left\{\begin{array}{ccc}{p}_g& \mathrm{if}& f\left(p_g\right)\&le;f\left(X_t^{p+1}\right)\\ X_t^{p+1}& \mathrm{if}& f\left(p_g\right)>f\left(X_t^{p+1}\right)\end{array}\right.t=\mathrm{1,2},...,T---\left(9\right)$

Wherein, f () is adaptive value function, and described gateway deployment method considers gateway covering radius and gateway load equilibrium level, and the adaptive value function account form of employing is such as formula shown in (10):

$f\left({\left\{u_k\right\}}_K\right)=\underset{1\&le;k\&le;K}{\max }\underset{v_i\&Element;U_k}{\max }\left(d(v_i,u_k)\right)+\mathrm{\&lambda;}\&times;({\mathrm{load}}_{\max }-{\mathrm{load}}_{\min })---\left(10\right)$

Wherein, λ is scale coefficient;

D (v _i, u _k) be sensor node v _ito gateway node u _kjumping figure; For a kth gateway u _k, its node location is (a _k, b _k), k=1,2 ..., K; Apart from this gateway node distance be less than communication radius sensor node composition set be Θ _k, this gateway node is to Θ _kthe jumping figure of middle arbitrary node is 1; Node v _ito Θ _kdistance be then node v _ito gateway node u _kjumping figure by formula (11) calculate:

$d(v_i,u_k)=\left\{\begin{array}{cc}1& v_i\&Element;{\mathrm{\&Theta;}}_k\\ \underset{v_l\&Element;{\mathrm{\&Theta;}}_k}{\min }\left(d(v_i,v_l)\right)+1& v_i\&NotElement;{\mathrm{\&Theta;}}_k\end{array}\right.---\left(11\right)$

The router node quantity that the load level of each gateway node is served according to it characterizes, and router node pays the utmost attention to the gateway node of selection within its coverage distance as service node; For the router node outside all gateway node coverage distances, then select the gateway node selecting jumping figure minimum by formula (11) as its service node; Define the load of whole information physical emerging system gateway node for (load ₁, load ₂..., load _k), then load _max=max (load ₁, load ₂..., load _k), load _min=min (load ₁, load ₂..., load _k);

Step 4, disturbance between multi-region is carried out to particle global optimum position

After carrying out particle history optimal location and particle global optimum location updating, between introducing multi-region, disruption and recovery is to particle global optimum position p _gbe optimized; Concrete grammar be whole iterative process is divided in early days, stage that mid-term, later stage three are different, make particle can by wave process close to global optimum by add different normal deviates period in different iterative computation;

In the stage in early days, introduce larger normal deviate; At mid-term stage, introduce medium normal deviate; During the late stages of developmet, less normal deviate is introduced; Concrete perturbation scheme is such as formula shown in (12):

$p_g^{\&prime;}=\left\{\begin{array}{cc}N(p_g,{\mathrm{\&sigma;}}_1)& \mathrm{if}\frac{p}{P}\&le;{\mathrm{\&theta;}}_1\\ N(p_g,{\mathrm{\&sigma;}}_2)& \mathrm{if}{\mathrm{\&theta;}}_1<\frac{p}{P}\&le;{\mathrm{\&theta;}}_2\\ N(p_g,{\mathrm{\&sigma;}}_3)& \mathrm{if}\frac{p}{P}>{\mathrm{\&theta;}}_2\end{array}\right.---\left(12\right)$

Wherein, Discontinuous Factors σ ₁=r × 0.1, σ ₂=r × 0.01, σ ₃=r × 0.001, r is the coverage distance of gateway node; Interval parameter θ ₁=0.1, θ ₂=0.3;

Carrying out between multi-region after disturbance, according to the particle global optimum position p' after formula (10) calculation perturbation _gcorresponding adaptive value functional value f (p' _g), if f is (p' _g) <f (p _g), Ze Geng new particle global optimum position p _g=p' _g, otherwise, keep particle global optimum position p _gconstant;

Step 5, repetition step 2 are to step 4 until iterations arrives maximum iteration time P, particle global optimum position p _grepresentative gateway location is the optimum deployed position coordinate of gateway in wireless sensor network;

In step one, inertia weight ω=0.729, Studying factors c ₁=c ₂=1.49, maximum iteration time P=500;

In step 3, the importance based on gateway covering radius index exceedes gateway load equilibrium level, and scale coefficient value is λ=0.1.

Claims (3)

1. the gateway deployment method under information physical emerging system, described method uses disturbance particle cluster algorithm between multi-region to carry out the optimizing of gateway location; Definition is for the gateway node of the K in information physical emerging system, and its coordinate at two dimensional surface is: (a _k, b _k), k=1,2 ..., K is X=(x with the particle position of gateway coordinate composition ₁, x ₂..., x _m), particle rapidity is V=(v ₁, v ₂..., v _m), wherein m=2K, x _2k-1=a _k, x _2k=b _k, the position of t particle is X _t=(x _{t, 1}, x _{t, 2}..., x _t,m), speed is V _t=(v _{t, 1}, v _{t, 2}..., v _t,m);

The implementation procedure of described method is as follows:

Step one, the initialization of gateway particle

In network effective coverage [0, L], random generation T intended particle the initial position of composition first generation population, each particle in population represents one group of possible gateway location, produces the initial velocity of random each particle simultaneously the velocity amplitude scope of particle is set for [-v _max, v _max]=[-0.15 × L, 0.15 × L], inertia weight is ω, and Studying factors is c ₁, c ₂, maximum iteration time is P;

According to formula (3), t (t=1,2..., T) individual particle history optimal location p is set _tfor the initial position of this particle, particle global optimum position p is set according to formula (4) _goptimal location in all particle initial positions:

$p_t=X_t^1,t=\mathrm{1,2}...,T---\left(3\right)$

$p_g=\underset{X_t^1}{\arg \max }f\left(X_t^1\right),t=\mathrm{1,2}...,T---\left(4\right)$

Wherein, f () is adaptive value function, and its account form is such as formula shown in (10);

{ u _k} _kbe represent all gateway node set, X is each node coordinate particular location in gateway node set;

Step 2, carry out particle flight calculating

The p be made up of gateway location is expressed as any one object vector of population: wherein: p=1,2 ..., P, t=1,2 ..., T; For t particle, calculate according to the speed flying speed that more new formula (5) carries out this particle of future generation $V_t^{p+1}=(v_{t,1}^{p+1}{,v}_{t,2}^{p+1},...,v_{t,m}^{p+1}):$

$v_{t,j}^{p+1}=\mathrm{\&omega;}\&times;v_{t,j}^p+c_1\&times;\mathrm{rand}\&times;(p_t-x_{t,j}^p)+c_2\&times;\mathrm{rand}\&times;(p_g-x_{t,j}^p),j=\mathrm{1,2}...,m---\left(5\right)$

Wherein, rand is the random number between 0 ~ 1;

The value scope if the flying speed of particle outpaces, be then limited on boundary condition according to formula (6):

$v_{t,j}^{p+1}=\left\{\begin{array}{cc}-v_{\max }& \mathrm{if}{v}_{t,j}^{p+1}<-v_{\max }\\ v_{\max }& \mathrm{if}{v}_{t,j}^{p+1}>v_{\max }\end{array}\right.---\left(6\right)$

The calculating of this particle position of future generation is carried out according to location updating formula (7)

$x_{t,j}^{p+1}=x_{t,j}^p+v_{t,j}^{p+1},j=\mathrm{1,2},...,m---\left(7\right)$

Step 3, more new particle history optimal location and particle global optimum position

According to the particle position after renewal particle history optimal location and particle global optimum position is calculated according to formula (8) and formula (9):

$p_t=\left\{\begin{array}{ccc}{p}_t& \mathrm{if}& f\left(p_t\right)\&le;f\left(X_t^{p+1}\right)\\ X_t^{p+1}& \mathrm{if}& f\left(p_t\right)>f\left(X_t^{p+1}\right)\end{array}\right.t=\mathrm{1,2},...,T---\left(8\right)$

$p_g=\left\{\begin{array}{ccc}{p}_g& \mathrm{if}& f\left(p_g\right)\&le;f\left(X_t^{p+1}\right)\\ X_t^{p+1}& \mathrm{if}& f\left(p_g\right)>f\left(X_t^{p+1}\right)\end{array}\right.t=\mathrm{1,2},...,T---\left(9\right)$

Wherein, f () is adaptive value function, and described gateway deployment method considers gateway covering radius and gateway load equilibrium level, and the adaptive value function account form of employing is such as formula shown in (10):

$f\left({\left\{u_k\right\}}_K\right)=\underset{1\&le;k\&le;K}{\max }\underset{v_i\&Element;U_k}{\max }\left(d(v_i,u_k)\right)+\mathrm{\&lambda;}\&times;({\mathrm{load}}_{\max }-{\mathrm{load}}_{\min })---\left(10\right)$

Wherein, λ is scale coefficient;

D (v _i, u _k) be sensor node v _ito gateway node u _kjumping figure; For a kth gateway u _k, its node location is (a _k, b _k), k=1,2 ..., K; Apart from this gateway node distance be less than communication radius sensor node composition set be Θ _k, this gateway node is to Θ _kthe jumping figure of middle arbitrary node is 1; Node v _ito Θ _kdistance be then node v _ito gateway node u _kjumping figure by formula (11) calculate:

$d(v_i,u_k)=\left\{\begin{array}{cc}1& v_i\&Element;{\mathrm{\&Theta;}}_k\\ \underset{v_l\&Element;{\mathrm{\&Theta;}}_k}{\min }\left(d(v_i,v_l)\right)+1& v_i\&NotElement;{\mathrm{\&Theta;}}_k\end{array}\right.---\left(11\right)$

The router node quantity that the load level of each gateway node is served according to it characterizes, and router node pays the utmost attention to the gateway node of selection within its coverage distance as service node; For the router node outside all gateway node coverage distances, then select the gateway node selecting jumping figure minimum by formula (11) as its service node; Define the load of whole information physical emerging system gateway node for (load ₁, load ₂..., load _k), then load _max=max (load ₁, load ₂..., load _k), load _min=min (load ₁, load ₂..., load _k);

Step 4, disturbance between multi-region is carried out to particle global optimum position

After carrying out particle history optimal location and particle global optimum location updating, between introducing multi-region, disruption and recovery is to particle global optimum position p _gbe optimized; Concrete grammar be whole iterative process is divided in early days, stage that mid-term, later stage three are different, make particle can by wave process close to global optimum by add different normal deviates period in different iterative computation;

In the stage in early days, introduce larger normal deviate; At mid-term stage, introduce medium normal deviate; During the late stages of developmet, less normal deviate is introduced; Concrete perturbation scheme is such as formula shown in (12):

$p_g^{\&prime;}=\left\{\begin{array}{cc}N(p_g,{\mathrm{\&sigma;}}_1)& \mathrm{if}\frac{p}{P}\&le;{\mathrm{\&theta;}}_1\\ N(p_g,{\mathrm{\&sigma;}}_2)& \mathrm{if}{\mathrm{\&theta;}}_1<\frac{p}{P}\&le;{\mathrm{\&theta;}}_2\\ N(p_g,{\mathrm{\&sigma;}}_3)& \mathrm{if}\frac{p}{P}>{\mathrm{\&theta;}}_2\end{array}\right.---\left(12\right)$

Wherein, Discontinuous Factors σ ₁=r × 0.1, σ ₂=r × 0.01, σ ₃=r × 0.001, r is the coverage distance of gateway node; Interval parameter θ ₁=0.1, θ ₂=0.3;

Carrying out between multi-region after disturbance, according to the particle global optimum position p' after formula (10) calculation perturbation _gcorresponding adaptive value functional value f (p' _g), if f is (p' _g) <f (p _g), Ze Geng new particle global optimum position p _g=p' _g, otherwise, keep particle global optimum position p _gconstant;

Step 5, repetition step 2 are to step 4 until iterations arrives maximum iteration time P, particle global optimum position p _grepresentative gateway location is the optimum deployed position coordinate of gateway in wireless sensor network.

2. the gateway deployment method under a kind of information physical emerging system according to claim 1, is characterized in that: in step one, inertia weight ω=0.729, Studying factors c ₁=c ₂=1.49, maximum iteration time P=500.

3. the gateway deployment method under a kind of information physical emerging system according to claim 2, it is characterized in that: in step 3, the importance based on gateway covering radius index exceedes gateway load equilibrium level, and scale coefficient value is λ=0.1.