CN105743710A - Wireless sensor network evolution model building method based on task importance of node - Google Patents

Abstract

The present invention discloses a wireless sensor network evolution model building method based on task importance of a node, which synthetically considers the local-world property of the WSN, residual energy of the node and ability of the network to resist an intentional attack, considers the importance of the tasks carried out by the node, provides a WSN evolution model that approximates a real network circumstance and considers task importance of the node, becomes more accordant with the circumstance in practical application, balances energy consumption of the node, increases lifetime of the network and has strong resistance ability to intentional attacks.

Description

Wireless sensor network evolutionary model construction method based on node tasks importance

Technical field

The present invention relates to wireless sensor network (WSN) and model field, particularly relate to a kind of wireless sensor network evolutionary model construction method based on node tasks importance.

Background technology

As one of base support technology that Internet of Things is important, wireless sensor network has the features such as power supply capacity is limited, multihop routing, self-organizing, dynamic topology.Owing to being subject to the restriction of limited energy budget, WSN is designed, be first also most important challenge is exactly energy efficiency.And the aspect of system, how to build the difficult problem that network topology sane, efficient is always up faced by research worker needs, captures.Since twentieth century end, Complex Networks Theory is progressively introduced into Other subjects field, becomes considerable analytical tool and research means.Barabasi and Albert of the U.S. in 1999 proposes scale-free model, it is found that the uncalibrated visual servo character of complex network, causes the attention of whole world research worker, promote the development that complex network is studied, also it is the structure of WSN sane, efficient, it is provided that a kind of new thinking.

At present achieved with some achievements in the modeling based on the WSN of scale-free model: (PayneJL, EppsteinMJ.EvolutionaryDynamicsonScale-FreeInteractionNe tworksEvolutionaryComputation [J] .IEEETransactions, Volume, 2009,13, Page (s): 895-912.) WSN topology controls from the angle analysis of network evolution.For the feature of WSN self, author is on the basis of scales-free network rising characteristic, it is contemplated that the effectiveness of the node energy consumption of WSN, it is believed that the optimum selecting of newly-increased link is not only relevant with the degree of node, also relevant with the dump energy of node, specific as follows:

$\Π\left(k_i\right)=\frac{f\left(E_i\right)k_i}{\underset{j}{\Σ}f\left(E_j\right)k_j}$

Wherein, Π (k_i) represent new node and there is the probability that node i is connected；k_iRefer to the number of degrees of node；E_iThe present energy of representation node；The relation of function representation residue energy of node and selected connection, is incremented by function.Residue energy of node is more big, and the probability being connected with newly added node is more big.This model balances the energy expenditure of node, and certain length extends network life.But this model not considered, WSN interior joint and link have increasing to have the dynamic behaviour subtracted.null(ZhuHailin,LuoHong,PengHaipeng.ComplexNetworksBasedEnergy-EfficientEvolutionModelforWirelessSensorNet-works[J].Chaos,Solitons&Fractals,2009,41 (4-30): 1828-1835.) WSN evolutionary model EAEM (energy-awareevolutionmodel) and EBEM (energy-balancedevolutionmodel) with uncalibrated visual servo characteristic is proposed based on L-W model，Balance the energy expenditure of node.(LiShudong, LiXiang, YangYixian.Alocal-worldheterogeneousmodelofWSNswithnodea ndlinkdiversity [J] .PhysicalA, StatisticalMechanicsandItsApplications, 2011,390 (6): 1182-1191.) propose a local Cluster Networks model, and demonstrate this network effectiveness in data transmission.Article is thought, the node being newly introduced is when establishing the link with origin node, due to its communication radius, can only be formed with the node in certain distance around and link, namely there is locality.This embodies the characteristic of wireless sensor network to a certain extent more really, but still has certain limitation.Because it does not account for isomerism and the Energy Efficient problem of the energy of sensor network interior joint.And (Luo little Juan, Yu Huiqun, cold spring rosy clouds. based on the Local World Dynamic Evolution Model [J] of wireless sensor network. East China University of Science's journal, 2012,2:216-220.) (symbol is improved literature, Li Wenfeng. based on Wireless sensor network clustering evolutionary model [J] .JournalonCommunications of Local World, 2015.) then further, Local World model basis considers the energy response of node, but still the vulnerability of model has not been improved.And at present the approach of the improvement of uncalibrated visual servo vulnerability is mainly reduced exactly the importance of key node: (Ze-HuiQ, PuW, Chao-MingS, etal.Enhancementofscale-freenetworkattacktolerance [J] .ChinesePhysicsB, 2010,19 (11): 110504.) proposing a kind of SL method, the most of node being about to be joined directly together with key node transfers to set up with its neighbor node to be connected；(Li Jianchun, Wu Xueli, Han Bing, Li Jianyong. a kind of scales-free network [J] that calculated attack is had robustness. Journal of Henan University (natural science edition), 2013,03:324-327.) it is then think that the vulnerability of scales-free network comes from preferentially to connect, by allowing new node be undertaken connecting at random or preferentially connecting by probability selection, with the situation that minimizing preferentially connects, and then reach to improve the purpose of the defensive ability/resistance ability to calculated attack.

Constantly be applied to every field along with WSN, the energy consumption of WSN, network robustness etc. require more and more higher.Structure excellent performance and the WSN model meeting reality have been focus directions to instruct the practical application of WSN.

Summary of the invention

The technical problem to be solved is for defect involved in background technology, it is provided that a kind of wireless sensor network evolutionary model construction method based on node tasks.

The present invention solves above-mentioned technical problem by the following technical solutions:

A kind of wireless sensor network evolutionary model construction method based on node tasks importance, comprises step in detail below:

Step 1), generate and have m₀The initial network of individual node, m₀For natural number；

Step 2), increase new node and give new node primary power and task significance；

Step 3), set up the linking relationship of new node and current network；

Step 3.1), select the Local World of M node composition new node closest with new node in current network；

Step 3.2), according to below equation calculate new node with in its Local World each node link probability:

$\Π\left(k_i\right)=(1-q)\frac{M}{N\left(t\right)}\frac{{\mathrm{\α}}_{i}f\left(E_i\right)k_i}{\underset{j=1,2,\mathrm{...},M}{\Σ}{\mathrm{\α}}_{j}f\left(E_j\right)k_j}+\frac{q}{N\left(t\right)}$

Wherein_,Π(k_i) referring to the probability that node i and new node link, q is coefficient value set in advance, 0≤q≤1, and N (t) is the node total number of t network, and M is the number of the Local World interior joint of new node, E_iFor the present energy of node i, α_iFor the task significance of node i, k_iFor the degree of node i, f (E_i) for present energy and the Π (k of node i_i) between function；

Step 3.3), set up the linking relationship of new node and its Local World interior joint according to new node with the probability that links of each node in its Local World；

Step 4), calculate, according to below equation, the probability that in current network, each node is deleted；

${\Π}^{*}\left(k_i\right)=\frac{1}{N\left(t\right)}$

Wherein, Π^*(k_i) for current network interior joint i be deleted probability；

Step 5), node each in current network is deleted according to its probability being deleted, and deletes the all-links being deleted between node and network；

Step 6), wait the time step T preset；

Step 7), repeat step 2) to step 6), until the node total number in current network is equal to the nodes threshold value preset.

The present invention adopts above technical scheme compared with prior art, has following technical effect that

1. more agree with the situation in practical application；

2. balance the energy expenditure of node, improve the life-span of network；

3. pair malicious attack has stronger defensive ability/resistance ability.

Accompanying drawing explanation

Fig. 1 is the evolution flow chart of the model that the present invention proposes；

Fig. 2 is the node degree distribution situation of change with Local World scale M；

Fig. 3 is the average path length situation of change with Local World scale M；

Fig. 4 is the average path length under calculated attack and network connectivty；

When Fig. 5 is M=m, the degree distribution curve of different probability of erasure.

Detailed description of the invention

Below in conjunction with accompanying drawing, technical scheme is described in further detail:

The invention discloses a kind of wireless sensor network evolutionary model construction method based on node tasks importance, comprise step in detail below:

Step 1), generate and have m₀The initial network of individual node, m₀For natural number；

Step 2), increase new node and give new node primary power and task significance；

Step 3), set up the linking relationship of new node and current network；

Step 3.1), select the Local World of M node composition new node closest with new node in current network；

Step 3.2), according to below equation calculate new node with in its Local World each node link probability:

$\Π\left(k_i\right)=(1-q)\frac{M}{N\left(t\right)}\frac{{\mathrm{\α}}_{i}f\left(E_i\right)k_i}{\underset{j=1,2,\mathrm{...},M}{\Σ}{\mathrm{\α}}_{j}f\left(E_j\right)k_j}+\frac{q}{N\left(t\right)}$

Wherein, Π (k_i) referring to the probability that node i and new node link, q is coefficient value set in advance, 0≤q≤1, and N (t) is the node total number of t network, and M is the number of the Local World interior joint of new node, E_iFor the present energy of node i, α_iFor the task significance of node i, k_iFor the degree of node i, f (E_i) for present energy and the Π (k of node i_i) between function；

Step 3.3), set up the linking relationship of new node and its Local World interior joint according to new node with the probability that links of each node in its Local World；

Step 4), calculate, according to below equation, the probability that in current network, each node is deleted；

${\Π}^{*}\left(k_i\right)=\frac{1}{N\left(t\right)}$

Wherein, Π^*(k_i) for current network interior joint i be deleted probability；

Step 5), node each in current network is deleted according to its probability being deleted, and deletes the all-links being deleted between node and network；

Step 6), wait the time step T preset；

Step 7), repeat step 2) to step 6), until the node total number in current network is equal to the nodes threshold value preset.

Below with mean field theory, this model is analyzed, first the rate of change of solution node degree:

$\frac{\∂k_i}{\∂t}=m\Π\left(k_i\right)-p\frac{k_i}{N\left(t\right)}=m*\frac{(1-q)M}{N\left(t\right)}*\frac{{\mathrm{\α}}_{i}f\left(E_i\right)k_i}{\underset{j=1,2,\mathrm{...},M}{\Σ}{\mathrm{\α}}_{j}f\left(E_j\right)k_j}-p\frac{k_i}{N\left(t\right)}$

Wherein, 0≤p≤1, m≤M≤m₀+t.And each time step T is represented as p=1, and while increasing a new node, delete again Geju City node, the node total number of network can't change.This situation and the actual of WSN are not inconsistent, so this special circumstances are not performed an analysis.

The scale M Taxonomic discussion in the local area world below:

(1) M=m

Now, the probability that the node being newly introduced node all with Local World establishes the link is all the same, preferentially connects principle and loses meaning.At moment t:

N (t)=m₀+t(1-p)

So,

$\begin{array}{c}\frac{\∂k_i}{\∂t}=m*\frac{(1-q)M}{N\left(t\right)}*\frac{1}{M}-p\frac{k_i}{N\left(t\right)}\\ =m*\frac{(1-q)M}{m_0+t(1-p)}*\frac{1}{M}-p*\frac{k_i}{m_0+t(1-p)}\\ \≈\frac{m(1-q)-{\mathrm{pk}}_i}{t(1-p)}\end{array}$

Solve partial differential equation, can obtainIt can be seen that the degree of this network is distributed as exponential form, and p is more big, and degree distribution curve declines more fast.

(2) m < M≤m₀+t

In this case, new node chooses m node in Local World according to preferentially connecting principle, and establishes the link.Can obtain:

$\frac{\&part;k_i}{\&part;t}=m*\frac{(1-q)M}{N\left(t\right)}*\frac{{\mathrm{\&alpha;}}_{i}f\left(E_i\right)k_i}{M\stackrel{\&OverBar;}{\mathrm{\&alpha;}E}<k\left(t\right)>}-p\frac{k_i}{N\left(t\right)}$

Wherein,The expectation of task significance is performed for node；It it is the expectation of network node energy；<k (t)>is the average degree of the node of moment t, and k is the degree of node.If total limit number of network is b (t), then have:

$\left\{\begin{array}{c}\frac{db\left(t\right)}{dt}=m-p<k\left(t\right)>\\ b\left(t\right)=\frac{<k\left(t\right)>*N\left(t\right)}{2}\end{array}\right.$

Solution:

$b\left(t\right)=\frac{m(1-p)}{(1+p)}*t$

So can obtain:

$\frac{\&part;k_i}{\&part;t}\frac{{\mathrm{\&alpha;}}_{i}f\left(E_i\right)k_i(1+p)(1-q)}{2\stackrel{\&OverBar;}{\mathrm{\&alpha;}E}t(1-p)}-\frac{{\mathrm{pk}}_i}{(1-p)t}$

By initial condition k_i(t_i)=m, can obtain:

$k_i\left(t\right)=m*{\left(\frac{t}{t_i}\right)}^{\mathrm{\&beta;}}$

Wherein,

$\mathrm{\&beta;}=\frac{{\mathrm{\&alpha;}}_{i}f\left(E_i\right)(1+p)(1-p)}{2\stackrel{\&OverBar;}{\mathrm{\&alpha;}E}(1-p)}-\frac{p}{(1-p)}$

Due to

$P\left(k_i\right(k)<k)=P(t_i>{\left(\frac{m}{k}\right)}^{1/\mathrm{\&beta;}}t)$

The interpolation of new node is the method according to constant duration simultaneously, n₀For initial time, then have:

$P\left(t_i\right)=\frac{1}{n_0+t}$

Before institute, formula can turn to:

$P\left(k_i\right(k)<k)=1-{\left(\frac{m}{k}\right)}^{1/\mathrm{\&beta;}}*\frac{t}{n_0+t}$

In sum, the degree distribution function of whole network is:

$P\left(k\right)=\frac{\&part;P\left(k_i\right(t)<k)}{\&part;k}=\frac{2m^{1/\mathrm{\&beta;}}}{n_0+t}{k}^{-(1+1/\mathrm{\&beta;})}$

Wherein, degree distribution exponent is:

γ=1+1/ β

Visible, P (k) and Local World scale M, energy function f (E) etc. are relevant, and network node scale onrelevant, have obvious uncalibrated visual servo characteristic.

In order to know more about the technology contents of the present invention, specifically introduce the embodiment of this model below in conjunction with emulation.In emulation experiment, it is assumed that node is randomly dispersed in two dimensional surface.The primary power of node, all it is normal distribution N (0,1).

Fig. 1 gives the flow chart of evolution algorithmic.Such as Fig. 1, firstly generate m₀The initial network of individual node.Then, the final scale of setting network, the end condition namely developed.Then being exactly the introduction of new node, the primary power E and node tasks importance α of new node are known.Local World before new node establishes the link it needs to be determined that belonging to it.Local World in model is limited in order to simulate WSN application time point radius, can only establish the link this phenomenon with the node within the scope of communication radius.As for the determination of Local World, as shown in Figure 2 and Figure 3:

Internet pricing when Fig. 2 gives m=4, p=0.2 is distributed the situation of change with Local World scale M.Now, network does not only have the introduction of new node, there is also the situation that node failure is dead.From Figure 2 it can be seen that this distribution curve entirety presents head of the horse shape.And in k >=m part, curve entirety exponentially curve when M is less；M is gradually increased, and degree distribution curve " is stretched and tighten " one-tenth power type curve gradually.And as M=2000, whole evolved network presents obvious uncalibrated visual servo characteristic.

Fig. 3 gives the network average path length (APL) situation of change with Local World scale M.As seen from Figure 3, the average path length of network reduces along with the increase of M, and when M increases to 40, average path length reduces about 40%.But when M continues to increase, now the average path length of network tends towards stability.

Thus illustrating, in the practical application of WSN, we, by increasing the mode of the transmitting power of node, increase Local World scale, thus improving the data transmission efficiency of network.Specifically much as transmitting power, Local World scale determine.And Local World scale should be much, then to analyze with the situation of change of M with specific reference to APL.In the emulation provided, M desirable 40.

Local World is once it is determined that it is necessary to preferentially connect.By evolution algorithmic it can be seen that affected by q with the connection probability of Nodus Nelumbinis Rhizomatis point.During q=0, preferentially connect in Local World；It it is then random connection during q=1.The determination of concrete q value, as shown in Figure 4:

Fig. 4 is when M=40, take q=0, q=0.2, q=0.4 respectively, q=0.6, q=0.8, q=1 emulate, and with (Luo little Juan, Yu Huiqun, cold spring rosy clouds. based on the Local World Dynamic Evolution Model [J] of wireless sensor network. East China University of Science's journal, 2012,2:216-220.) model that proposes contrasted.Supplementing, the knot removal in emulation is the mode taking the manually node that deletion degree is maximum.During q=0, network is similar with BA scales-free network, and during q=1, network then deteriorates to common random network.When knot removal ratio is 0, q from 0 toward 1, WSN is by scales-free network to random network transition, and average path length also increases therewith.Then, analyzing from the angle of WSN vulnerability: during q=0, as long as 1% node that degree of leaving out is maximum, network is not just in connection；During q=0.2, delete the maximum node of degree of 6%, just there will be isolated module；Q >=0.6 time, even if leaving out 10%, network still possesses connectedness.And sieve et al. propose model when in the face of calculated attack performance with q=0 time this paper model similar, at 1% node that degree of leaving out is maximum, network just loses connectedness.

Obviously, at q by the process of 0 to 1, the APL of network is becoming big, but the defensive ability/resistance ability of the many calculated attacks of network is strengthening, and reducing APL thus and strengthening defensive ability/resistance ability is competing to a certain extent.Can only carry out accepting or rejecting, balancing between the two according to practical situation, and then it is higher and calculated attack has the WSN model of stronger defensive ability/resistance ability to obtain energy consumption.For this simulation example, desirable q=0.6.

Q value is determined, preferentially connects.After connection, the situation dead in order to reproduce the inefficacy of WSN application interior joint, with the probability random erasure node of p=0.2, link.In Figure 5, WSN degree distribution curve when giving M=m, q=0 is with the situation of change of p.Now, the node being newly introduced establishes the link with nodes all in Local World with identical probability, preferentially connects principle and loses meaning already.By mathematical derivation before it can be seen that degree distribution curve entirety this moment should present the feature of exponential.Observing analogous diagram 5, as p=0, degree distribution is exponential really on the whole.And during p ≠ 0, although degree distribution curve still meets exponential characteristic on the whole, but occur in that the minority degree node less than m.This is because now p is not equal to 0, part of nodes is dead, and link failure, naturalness distribution curve can be affected.It addition, can also be apparent from from Fig. 5, along with the increase of p, degree distribution exponent is also becoming big, and spends distribution curve and decline faster, and this is also consistent with mathematical analysis.

Introduce and after knot removal at new node, update network topology.Now, complete one and take turns evolution.Repeat above step, until network reaches scale of presetting, then introduction of developing, export result.

The present invention is from the angle of WSN dynamic evolution, it is proposed that a kind of wireless sensor network evolutionary model construction method based on node tasks importance.This model has considered following several characteristic: (1) is based on the preferentially connection of Local World；(2) based on the preferentially connection of Energy-aware；(3) the preferentially connection of undertaking based on node of task；(4) node, link have increasing to have to subtract；(5) defensive ability/resistance ability to malicious attack.This model had both agreed with the practical situations of WSN, again in efficiency and performance in the resisting of calculated attack is good, the practical application of WSN was had certain directive significance.

Those skilled in the art of the present technique it is understood that unless otherwise defined, all terms used herein (include technical term and scientific terminology) and have with the those of ordinary skill in art of the present invention be commonly understood by identical meaning.Should also be understood that in such as general dictionary, those terms of definition should be understood that have the meaning consistent with the meaning in the context of prior art, and unless defined as here, will not explain by idealization or excessively formal implication.

Above-described detailed description of the invention; the purpose of the present invention, technical scheme and beneficial effect have been further described; it is it should be understood that; the foregoing is only the specific embodiment of the present invention; it is not limited to the present invention; all within the spirit and principles in the present invention, any amendment of making, equivalent replacement, improvement etc., should be included within protection scope of the present invention.

Claims (1)

1. based on the wireless sensor network evolutionary model construction method of node tasks importance, it is characterised in that comprise step in detail below:

Step 1), generate and have m₀The initial network of individual node, m₀For natural number；

Step 2), increase new node and give new node primary power and task significance；

Step 3), set up the linking relationship of new node and current network；

Step 3.1), select the Local World of M node composition new node closest with new node in current network；

Step 3.2), according to below equation calculate new node with in its Local World each node link probability:

$\&Pi;\left(k_i\right)=(1-q)\frac{M}{N\left(t\right)}\frac{{\mathrm{\&alpha;}}_{i}f\left(E_i\right)k_i}{\underset{j=1,2,...,M}{\&Sigma;}{\mathrm{\&alpha;}}_{j}f\left(E_j\right)k_j}+\frac{q}{N\left(t\right)}$

Wherein_,Π(k_i) referring to the probability that node i and new node link, q is coefficient value set in advance, 0≤q≤1, and N (t) is the node total number of t network, and M is the number of the Local World interior joint of new node, E_iFor the present energy of node i, α_iFor the task significance of node i, k_iFor the degree of node i, f (E_i) for present energy and the Π (k of node i_i) between function；

Step 3.3), set up the linking relationship of new node and its Local World interior joint according to new node with the probability that links of each node in its Local World；

Step 4), calculate, according to below equation, the probability that in current network, each node is deleted；

${\&Pi;}^{*}\left(k_i\right)=\frac{1}{N\left(t\right)}$

Wherein, Π^*(k_i) for current network interior joint i be deleted probability；

Step 5), node each in current network is deleted according to its probability being deleted, and deletes the all-links being deleted between node and network；

Step 6), wait the time step T preset；

Step 7), repeat step 2) to step 6), until the node total number in current network is equal to the nodes threshold value preset.