CN106162720A - A kind of cognition wireless self-organized network nodes stability assessment method based on multiple attribute decision making (MADM) - Google Patents

Abstract

A kind of cognition wireless self-organized network nodes stability assessment method based on multiple attribute decision making (MADM).The present invention is with raising cognition wireless self-organized network nodes availability as target, based on node stability appraisement system, solves cognition wireless self-organized network nodes determination of stability problem.The characteristic collected according to node, adds up and calculates neighbor node stability therein, Criterion Attribute parameter that usable spectrum stability is relevant with annoyance level.By normalized method, these Parameters Transformation are become attribute vector.Assessment affects the influence degree of each index of node stability, calculates each index weight size in node stability appraisement system.Node stability evaluation problem will be converted into Multiple Attribute Decision Problems, in conjunction with priori based on specimen sample, realize the judgement to node current steady sexual state by plus-minus ideal solutions method.Invention introduces node stability evaluation system model, Fuzzy AHP and plus-minus ideal solutions method, make result of calculation be applicable to the fields such as route, sub-clustering, distribution.

Description

A kind of cognition wireless self-organized network nodes stability assessment based on multiple attribute decision making (MADM) Method

Technical field

The present invention relates to the node state evaluation areas at cognition wireless networking, particularly relate to a kind of based on multiple attribute decision making (MADM) Node state appraisal procedure.

Background technology

Cognition wireless self-organizing network is a kind of self-organizing network based on cognitive radio technology.Each cognitive user (Cognitive User, CU) is found by cognitive radios and utilizes frequency spectrum cavity-pocket.CU is detecting possible transmission After chance (frequency spectrum cavity-pocket), by performing dynamic spectrum access, data are passed by cognitive radio (Cognitive Radio) It is passed to recipient.If primary user (Primary User, PU) occurs suddenly, CU has to keep out of the way immediately.Due to cognition wireless The dynamic characteristic of dynamically change and the topology of self-organized network nodes usable spectrum has considerable influence to network availability, therefore Efficiently, the cognition nothing that reliable node stability assessment method is poor to network reliability, link unstable, usable spectrum dynamically changes It is even more important for line self-organizing network.

In recent years, in order to solve the node stability evaluation problem under cognition wireless self-organizing network environment, research worker Launching research from many aspects, acquired achievement mainly has:

(1) the link retention time Forecasting Methodology based on the movement of node

When this method is employed, node needs to obtain self and the position of adjacent node, speed and moving direction.According to joint The position of point and motion track, in conjunction with node motion feature, use the method for geometry to calculate the link pot life of two nodes, Such that it is able to calculate this node to be adjacent the degree of stability of node.The method is used for positioning at any time, moving direction In clear and definite network morphology.

(2) the node stability appraisal procedure based on node signal disturbs

When this method is employed, nodes records and the stream internal interference of neighbors link and inter-stream interference value, and to every pair of chain The interference value on road is weighted adding up.Interference value according to every pair of adjacent link, it was predicted that go out total interference value of present node, from And the degree of stability of node can be calculated.The method is used in the cognition wireless self-organizing network of multi-hop, multithread.

Different according to information needed, (1) can be summarized as needing the method for node location information, (2) then belong to and are not required to The method wanting node location information.In method (1), in the environment of cognition wireless self-organizing network is distributed, the most difficult obtain Obtain position and the speed of destination node, and affecting node stability is not only owing to change in topology causes；Method (2) is current Interference prediction in most of scenes is more rough and ageing very poor.Factors above all constrains cognition wireless from group Knit nodes stability assessment method accuracy in actual applications.

In various cognition wireless self-organizing network environments, affect, by excavating, the factor node factor that node is stable, enter And be analyzed each factor sorting out, improve effectiveness and the accuracy of node stability prediction, improve cognitive nothing further The availability of line self-organizing network.Therefore, conclude and influence degree analysis affecting stable the carrying out of node, to improving cognitive nothing Line availability of ad hoc network has very important significance.

Summary of the invention

The present invention is with raising node cognition wireless availability of ad hoc network as target, to affect the factor of node stability Based on, in conjunction with multi-attribute Decision-making Analysis method, solve cognition wireless self-organized network nodes stability prediction problem.Concrete bag Include:

1. node stability is due to by numerous uncertain factor shadows such as node motion, usable spectrum change and electromagnetic interference Ring, set up node stability indicator evaluation system, analyze each factor influence degree to node stability.

2., based on node stability indicator evaluation system, node stability evaluation problem will be converted into multiple attribute decision making (MADM) and ask Topic, thus realize the judgement of the stability attribute to cognition wireless self-organized network nodes.In order to make result of calculation quick, accurate Really so that result is applicable to the fields such as route, sub-clustering, distribution, and the present invention refer to cross-layer shared mechanism, plus-minus ideal solutions side Method.

Compared with prior art, the innovation of the present invention is: based on node stability appraisement system, it is adaptable to Different cognition wireless self-organizing network forms, and there is adaptive ability.It is embodied in:

1., based on node stability appraisement system, in the environment of various cognition wireless self-organizing networks, each node is all Can be with the attribute information of influence factor in distributed recording appraisement system, required information is compared with other method and is easily obtained ?.

2., during the use of influence factor's attribute information, priori based on node, in different network environments Middle adjustment network environment relevant parameter, makes the present invention have adaptive ability.

Accompanying drawing explanation

Fig. 1 node stability appraisal procedure flow process

Fig. 2 node stability indicator evaluation system schematic diagram

Fig. 3 plus-minus ideal solutions schematic diagram

Detailed description of the invention

Refering to Fig. 1, node is accurate by cross-layer cooperation interaction mechanism collector node stability indicator appraisement system (refering to Fig. 2) The then factor attribute information in layer, generates fuzzy evaluation decision matrix by these parameters；By decision index system property value normalization, Fuzzy evaluation decision matrix is changed into standard and obscures decision matrix, the standard belonged to each feature of egress current state Value.Then the weight of each characteristic attribute is allocated, determines each factor of influence influence degree to node steady statue. Based on sampled data as canonical matrix, node stability evaluation problem is converted to Multiple Attribute Decision Problems, finally by just Minus ideal result method, it is judged that the steady statue of node.Invention introduces node stability evaluation system model, plus-minus ideal solutions Method and Fuzzy AHP, make result of calculation be applicable to the fields such as route, sub-clustering, distribution, mainly comprise in actual enforcement Three phases is implemented.

First, based on node stability appraisement system, set up attribute vector.

By to cognition wireless self-organizing network environment to node stability factorial analysis, a node can be constructed steady Qualitative evaluation system, as shown in Figure 2.It can be seen that the factor affecting node stability is summarized as 3 aspects, 6 Individual index.3 aspects refer to: neighbor node stability, usable spectrum stability and annoyance level.6 indexs refer to: one jumps neighbour Occupying node number, with the link stability of a hop neighbor node, with the number of available channels of a hop neighbor node, available channel changes Activity statistics, with the channel quality of a hop neighbor node and the channel quantity that is interfered.In order to build reflection node stability State, needs acquisition index initial data, and wherein the index of reflection usable spectrum stability and annoyance level need to pass through cross-layer cooperation Mode obtain, then achievement data is calculated and normalized, finally obtains attribute matrix.

Achievement data collection refers to, with calculating, 6 marker data information that acquisition node is current, and some of which index needs Directly gathering initial data to obtain, some index needs calculating further to initial data.Specifically, index 1 passes through net Network layers Topology Management mechanism directly obtains, and index 3 is obtained by MAC layer related protocol, and index 4 is link down frequency, it is possible to Obtained by MAC layer related protocol.Index 2,5 utilizes the achievement of forefathers to be calculated, and index 6 is currently may be used with each neighbor node The channel number of certain threshold values it is higher than with channel bit error rate.It addition, index 2,3,5 calculate with the desired value of neighbor node after, first Carry out adding up and then average, finally obtain index 2,3,5 final desired values.

Index 2 with the computing formula of the link stability of a hop neighbor node is:

$L_{ij}=\min (\frac{T_{ij}}{E{\left(T\right)}^{,}},1)---\left(1\right)$

Wherein T_ijRepresenting that node i and node j estimate the time kept, E (T) shows the phase of the time of node i and node j holding Prestige value.

Index 5 with the computing formula of the channel quality of each neighbor node is:

${\mathrm{Thr}}_{(i,j)}={\Σ}_{b\∈B_i\∩B_j}{\mathrm{\α}}_b\×\frac{{\mathrm{Thr}}_{(i,j),b}}{{\mathrm{maxThr}}_{(i,j),b}}---\left(2\right)$

Work as p_{Loss, b}=0, T_{F, b}=1 and B_iWhen being the applicable frequency range of node i, maxThr_{(i, j), b}=Thr_{(i, j), b}.Weights a_b (a_b≤ 1) different spectral characteristics (interference level of successive bands, channel error rate, path loss) is reflected.

Owing to some desired value is the bigger the better (i.e. desired value is the biggest, and node is the most stable), and some desired value be the least more Good, so needing desired value is normalized.For the index (index 1,2,3 and 5) being the bigger the better, normalization formula For:

$x^{*}=\frac{x-\min }{\max -\min }---\left(3\right)$

The biggest more poor for desired value (i.e. desired value is the biggest, and node is the most unstable), i.e. index 4 and 6, normalization formula For:

$x^{*}=1-\frac{x-\min }{\max -\min }---\left(4\right)$

After normalization, finally obtain the attribute vector of one 6 dimension

Second, determine that decision index system attribute weight distributes.

The distribution of decision index system attribute weight refers to according to Criterion Attribute node stability be affected size, determines this index Weights in node stability appraisement system.Specifically, it is simply that determine that accompanying drawing 2 destination layer is to 6 in rule layer model The weights of index.Total method is, first has to obtain the weights distribution of every layer of index, then obtains at the weights according to upper strata The weight of lower floor's (i.e. 6 indexs).

The weights of every layer of index divide and refer to carry out every layer of index of rule layer weight distribution and calculate, point following steps meter Calculate:

1) fuzzy matrix is calculated.When the rule layer all index of kth layer is compared two-by-two, use the language of Triangular Fuzzy Number Speech transformational rule is changed, and Triangular Fuzzy Number is with m_ijFor the closed interval of intermediate value, and m_ijIt is 1 in analytic hierarchy process (AHP)～9 marks Angle value, as shown in table 1 below.Obtain fuzzy number a that index all to k layer compares two-by-two_ij=(l_ij, m_ij, h_ij), it is assumed that pass judgment on specially The number of family is t, then the final Triangular Fuzzy Number of index j is by kth layer index iFalse If kth layer has h index, then obtain a h_k×h_kFuzzy matrix A=(a_ij)h_k×h_k.It should be noted that this matrix There is reciprocity, i.e.

1～9 scale synopsis of table 1 relative significance of attribute contrast

2) the fuzzy mearue value of parameter.The fuzzy survey matrix obtained according to upper step, i-th index is to its of same layer k The comprehensive fuzzy mearue value of its all factorComputing formula as follows:

$S_i^k={\Σ}_{j=1}^n{a}_{ij}^k\·{\left({\Σ}_{i=1}^{n_k}{\Σ}_{j=1}^{n_k}{a}_{ij}^k\right)}^{-1}---\left(5\right)$

3) calculating the weight distribution of kth layer i-th index, computing formula is as follows

WhereinCalculation be

$V(S_i^k\≥S_j^k)=\left\{\begin{array}{cc}1& m_1\≥m_2\\ \frac{l_2-h_1}{(m_1-h_1)-(m_2-l_2)}& m_1\≤m_2,h_1\≥l_2,i,j=1,2,\mathrm{...},n_k,i\≠j\\ 0& otherwise\end{array}\right.---\left(6\right)$

Then, the weight of kth layer index is assigned as:

$W^k=\{w_1^k,\mathrm{...},w_i^k,\mathrm{...},w_{h_k}^k\}---\left(7\right)$

4) the weight distribution of calculation criterion layer bottom index.Owing to this programme rule layer is two-layer, ground floor has three Index, the second layer has six indexs, so according to above-mentioned front 3 steps, can respectively obtain index weights vector

$W^1=(w_1^1,w_2^1,w_3^1)---\left(8\right)$

With

$W=(w_1^2,w_2^2,w_3^2,w_4^2,w_5^2,w_6^2),---\left(9\right)$

Then, Criterion Attribute weight is assigned as:

$W=(\frac{w_1^2}{w_1^2+w_2^2}*w_1^1,\frac{w_2^2}{w_1^2+w_2^2}*w_2^2,\frac{w_3^2}{w_3^2+w_4^2}*w_2^1,\frac{w_4^2}{w_3^2+w_4^2}*w_2^1,\frac{w_5^2}{w_5^2+w_6^2}{w}_3^1,\frac{w_6^2}{w_5^2+w_6^2}{w}_3^1)---\left(10\right)$

3rd, node stability state decision-making judgement calculates.

After node stability appraisement system index being carried out weight distribution according to expert estimation, it is necessary to carry out node steady Qualitative state calculates.Node stability state computation is divided into 2 steps.First sample according to the attribute vector built, Egress is unstable, secondary stable and attribute vector during steady statue:Then these sample status attributes to Amount is saved in Sample Storehouse as knowledge.Secondly, the attribute vector currently built according to nodeCalculateWith each sample shape The matching similarity of state attribute vector, finds out sample state corresponding to matching similarity peak as the current state of node. The computational methods flow process of matching similarity is following (saying accompanying drawing 3 refering to explanation):

1) decision space is built.

If node stability appraisement system index number n, sample status attribute vector m, then build a n-dimensional space S_n, each decision-making state i is mapped as S_nA bit in space, characterizes sample state i at S_nIn locus, and these Point is referred to as the spatial point of positive ideal solutionWhen existing equal with the distance of the positive ideal solution of each state in order to avoid evaluation state The situation that state matching similarity is good and bad cannot be judged, to each positive desired node, set n-dimensional space S_nMinus ideal result.This In positive and negative ideal space point be virtual, each Criterion Attribute of negative ideal space point is corresponding positive preferable solution space The worst state of point.

2) digital simulation similarity.

Digital simulation similarity, method is to measure current attribute vector to be assessed respectively to tie up at n with each positive Negative ideal point Distance in space, weighs the matching similarity of current attribute vector to be assessed and each sample state with this.

Current attribute vector l to be assessed_tPositive preferable solution space point with sample state iDistance computing formula as follows:

$d_{li}^{+}=\sqrt{{\Σ}_{j=1}^m{(l_{tj}-x_{ij}^{+})}^2},j=1,2,\mathrm{...}m.---\left(11\right)$

WhereinRepresent positive preferable solution space pointJth attribute at S_nIn component value.

Current attribute vector l to be assessed_tMinus ideal result spatial point with sample state iDistance computing formula as follows:

$d_{li}^{-}=\sqrt{{\Σ}_{j=1}^m{(l_{tj}-x_{ij}^{-})}^2},j=1,2,\mathrm{...}m---\left(12\right)$

The most current attribute vector to be assessed with the matching similarity of sample state i is:

$C_{li}=\frac{d_{li}^{-}}{d_{li}^{-}+d_{li}^{+}}---\left(13\right)$

3) present node stability status is calculated.

Due to node stability appraisement system index number 6, sample status attribute vector 3, according to step one and two, Current attribute vector l to be assessed can be calculated_tMatching similarity C with each sample state i_li(i=1,2,3), then work as prosthomere Point stability status is:

J={i | C_li is max}(14)。

Claims (1)

1. a cognition wireless self-organized network nodes stability assessment method based on multiple attribute decision making (MADM), comprises node stability Appraisement system builds, index weights based on Fuzzy AHP calculates and node stability based on plus-minus ideal solutions method Judgement calculates, it is characterised in that:

1) ginseng that each node is relevant with each index in periodically collecting current local cognition wireless self-organizing network environment Number, Internet to these parameter acquiring and analysis, builds node stability indicator evaluation system by cross-layer interaction mechanism, analyzes Each factor influence degree to node stability；Node stability indicator evaluation system is divided into destination layer, rule layer and decision-making Solution layer, wherein destination layer is node stability evaluation；Rule layer includes 2 layers, and ground floor includes neighbor node stability, available Frequency spectrum stability and annoyance level；The second layer includes 6 concrete evaluation indexes, respectively one hop neighbor node number, jumps with one The link stability of neighbor node, with the number of available channels of a hop neighbor node, the available channel change activity system of present node Meter, with the link-quality of a hop neighbor node, the channel statistical that present node is interfered；Decision-making party pattern layer includes three certainly Plan, is respectively stable, secondary stable and unstable three classes；

2) according to node stability appraisement system, the weight of following steps calculation criterion each index of layer is taked to distribute:

A) use Triangular Fuzzy Number, the rule layer all index of kth layer is compared two-by-two, obtains fuzzy matrix A；

B) according to fuzzy matrix A, the i-th index comprehensive fuzzy mearue value to other all factors of same layer k is calculated

C) calculating the weight distribution of kth layer i-th index, computing formula is as follows:

And j=1,2 ..., h_k, whereinCalculation be

$V\left(S_i^k\≥S_j^k\right)=\left\{\begin{array}{cc}1& m_1\≥m_2\\ \frac{l_2-h_1}{\left(m_1-h_1\right)-\left(m_2-l_2\right)}& \begin{array}{ccc}{m}_1\≤m_2,& h_1\≥l_2,& i,j=1,2,...,n_k,i\≠j\end{array}\\ 0& otherwise\end{array}\right.$

D) the weight distribution of calculation criterion layer bottom index；

According to above-mentioned calculation procedure, obtain the weight of 6 indexs of the rule layer second layer:

$W=(\frac{w_1^2}{w_1^2+w_2^2}*w_1^1,\frac{w_2^2}{w_1^2+w_2^2}*w_2^2,\frac{w_3^2}{w_3^2+w_4^2}*w_2^1,\frac{w_4^2}{w_3^2+w_4^2}*w_2^1,\frac{w_5^2}{w_5^2+w_6^2}{w}_3^1,\frac{w_6^2}{w_5^2+w_6^2}{w}_3^1)$

WhereinThe weight vectors of three indexs of rule layer ground floor；

3) according to the weight allocation vector of node stability appraisement system and each index of rule layer, following steps are used to calculate joint The stability status that point is current:

A) build decision space, each decision-making state i is mapped as S_nA bit in space, characterizes sample state i at S_nIn sky Between position；

B) current attribute vector to be assessed and each positive Negative ideal point distance in n-dimensional space are measured, digital simulation similarity, Computing formula is as follows:

$C_{li}=\frac{d_{li}^{-}}{d_{li}^{-}+d_{li}^{+}}$

WhereinWithComputation attribute vector l respectively_tDistance with the plus-minus ideal solutions spatial point of sample state i；

C) according to matching similarity, present node stability status is calculated.