CN106454856A - Spectrum allocation method based on graph coloring and analytic hierarchy process in cognitive radio - Google Patents

Abstract

The invention discloses a spectrum allocation method based on graph coloring and an analytic hierarchy process in cognitive radio. The method comprises the following steps: 1, establishing a recursion order hierarchy structure; 2, constructing paired comparison matrixes; 3, performing single hierarchical sequencing and checking the consistency; 4, performing total hierarchical sequencing, i.e., calculating the weight of relative importance of all factors on each layer relative to the highest layer from top to bottom to obtain a weight sequence of each scheme in the bottom layer relative to a target so as to select a scheme, and checking the consistency too in the total sequencing process, wherein the relative weights of all available spectrum resources about the total target can be normalized into a weight vector by the total hierarchical sequencing, and the spectra corresponding to the maximum value are selected for next allocation in the algorithm; and 5, allocating the selected optimal spectra by using a coloring algorithm in the graph theory. A sequence of spectrum qualities is obtained by analysis and decision on multiple attributes of all current available bands, so that a reasonable spectrum access strategy is provided.

Description

Frequency spectrum distributing method based on Turing pattern formation and analytic hierarchy process (AHP) in cognitive radio

Technical field

The present invention relates to one kind is in cognitive radio system, in the case that frequency spectrum resource is in short supply, make full use of frequency spectrum Resource, maximizes the frequency spectrum allocation algorithm design of the availability of frequency spectrum, belongs to cognitive radio frequency spectrum distribution technique.

Background technology

The spectrum shortage problem of mobile communication business was increasingly severe in the last few years, anxious to bandwidth demand in some frequency ranges While speed increases, the availability of frequency spectrum in some frequency ranges is relatively low, or even also there is the situation that frequency spectrum resource leaves unused.Cognitive nothing The core concept of line electricity is exactly to make Wireless Telecom Equipment have discovery " frequency spectrum cavity-pocket " and the ability that rationally utilizes is not it is considered to be Carry out one of developing direction of wireless communication technology.Cognitive radio is the radio communications system of an intelligence, it by with outer Interacting and carrying out various dimensions spectrum detection of boundary's environment, is sent out by self-adaptative adjustment in the case of primary user not being interfered Penetrate machine parameter and realize the frequency spectrum share with primary user.The appearance of cognitive radio makes " the secondary utilization " of frequency spectrum, and becoming can Can, thus being to realize frequency spectrum in the case of frequency spectrum resource deficiency and dynamically manage, improve the conception of the availability of frequency spectrum having started one Individual brand-new situation.

Frequency spectrum distribution is one of most important technology in cognitive radio networks, finds effective Spectrum Scheme to improve frequency The utilization rate of spectrum resource is it is ensured that cognitive user does not affect primary user use while transmission is in cognitive radio networks research A key problem.At present, several distribution models are proposed after both at home and abroad frequency spectrum distribution being furtherd investigate, mainly Have：Turing pattern formation model, game theoretical model, interference temperature model etc..Frequency spectrum based on Turing pattern formation model distributes early than honeycomb Mobile communication period begins to apply, and has had evolved into the model of relative maturity till now.Cognition in cognitive radio networks User typically accesses the frequency spectrum that primary user is temporarily left unused by the way of waiting for an opportunity, so usable spectrum is by primary user's state, position Determine etc. factor；Also interfere between cognitive user so that they have the characteristic of space-time change simultaneously.Turing pattern formation model is Cognitive user topological structure is abstracted into figure, the essence of the frequency spectrum distribution based on graph theory is exactly that the fixed point to non-directed graph is carried out Color.Multiple frequency spectrum allocation algorithms are gone out based on Turing pattern formation model development, but how these algorithm undue weight have distributed and to have neglected The heterogeneous problem of video spectrum.In cognitive radio networks, because frequency spectrum is heterogeneous, the characteristic of different spectral resource is to differ Sample, such as primary user is to characteristics such as the holding time of frequency spectrum, delay, bandwidth.Cognitive user in wireless network does not have oneself Mandate frequency spectrum can use, simply opportunistic accesses " spectrum interposition " of other authorization channels, these authorized frequency spectrums Due to geographical position, the difference of the aspect such as management strategy, very big difference is existed on spectral characteristic.So, in distribution frequency spectrum During be necessary the problem heterogeneous in view of frequency spectrum.

Content of the invention

In order to overcome the list colouring algorithm based on Turing pattern formation model in existing cognitive radio networks to distribute in frequency spectrum The deficiency that aspect exists, invention introduces the analytic hierarchy process (AHP) in multiobjective decision-making is it is proposed that be based on Turing pattern formation and level The cognitive radio spectrum allocation method of analytic approach.This algorithm is by being analyzed to multiple attributes of currently all available frequency bands Decision-making, obtains the sequence of Frequency spectrum quality, provides rational frequency spectrum access strategy.

The technical solution adopted for the present invention to solve the technical problems is：

Frequency spectrum distributing method based on Turing pattern formation and analytic hierarchy process (AHP) in a kind of cognitive radio, described frequency spectrum distribution side Method comprises the following steps：

First, set up recursive hierarchy structure

First according to the relation between each element in decision system, they are decomposed into different parts, each group Part is become to be referred to as element, then according to groups elements are formed the level of complementary correlation, the part to next layer for the last layer by attribute Or whole elements play dominating role, material is thus formed dominance relation successively；It is divided into three below level：

A) destination layer, that is, top：The predeterminated target of finger problem；

B) rule layer, i.e. intermediate layer：Refer to the criterion of impact realization of goal；

C) decision-making level, i.e. lowermost layer：Refer to promote the measure of realization of goal；

The decision rule that selection bandwidth, time delay, shake, this four attributes of packet loss select as frequency spectrum；

2nd, construct pairwise comparison matrix

Weight after establishing recursive hierarchy structure, to each element a certain criterion in secondary with regard to last layer of same level The property wanted is compared two-by-two, and is quantified, and just constitutes pairwise comparison matrix；

3rd, Mode of Level Simple Sequence and its consistency check

Mode of Level Simple Sequence refers to the sequence power for certain factor relative importance of last layer time for the corresponding factor of same level Weight, is obtained by the corresponding characteristic vector of Maximum characteristic root normalizing judgment matrix；And uniformity inspection is carried out to judgment matrix Test；

4th, total hierarchial sorting and its consistency check

Total hierarchial sorting refers to calculate the power to top relative importance for all factor of each layer from top to down Weight, obtains the weight sequencing to target for each scheme in the bottom, thus carrying out the selection of scheme, this total sequencer procedure is also carried out one The inspection of cause property；

All usable spectrum resources can be normalized into one with regard to the relative weighting of general objective through total hierarchial sorting Weight vectors, the frequency spectrum that algorithm is therefrom chosen corresponding to maximum carries out ensuing distribution；

5th, using the colouring algorithm in graph theory, the optimal spectrum selected is allocated.

Further, in described step 2, if n factor C of a certain layer will be compared₁,C₂,…,C_nTo last layer factor O Impact, take two factors C every time_iAnd C_jCompare it to the importance of target factor O it is simply that calculating corresponding weight W₁, W₂,…,W_n, by 1-9 scaling law to weight corresponding assignment, be defined as a_ij, whole comparative results pairwise comparison matrix A table Show：

A=(a_ij)_n×n(1)

Pairwise comparison matrix has following property：

When above formula is all set up to pairwise comparison matrix all elements, then this matrix is called consistency matrix；

Here set destination layer A to rule layer B₁,B₂,B₃,B₄Pairwise comparison matrix be：

Wireless terminal real-time monitoring available frequency band, according to active wireless network state obtain each frequency range respectively to bandwidth, Time delay, shake, the parameter value of four attributes of packet loss, determine corresponding weight, thus assigning to the element in pairwise comparison matrix Value, draws rule layer B₁,B₂,B₃,B₄With respect to P₁,P₂,…,P_nPairwise comparison matrix, with B₁To P₁,P₂,…,P_nThe one-tenth of effect As a example comparator matrix：

Further, in described step 3, calculate characteristic vector and adopt and method, process is as follows：

A) each column vector of A is normalized：

B) rightSue for peace by row

C) above-mentioned matrix-vector is normalized,Obtain W=(W₁,W₂,…,W_n)^TFor weight vector；

D) calculate AW；

E) calculateThis is the approximation of eigenvalue of maximum.

Further, in described step 3, the process of consistency check is as follows：

The first step:Calculate coincident indicator CI：

Wherein, n is the exponent number of pairwise comparison matrix；

Second step:Table look-up the corresponding random index RI of determination, it is judged that matrix different rank obtains mean random Coincident indicator RI；

3rd step:Calculate consistency ration CR and judged：

As CR ＜ 0.1 it is believed that the uniformity of judgment matrix is acceptable, during CR ＞ 0.1 it is believed that judgment matrix not Meet coherence request, need this judgment matrix is revised again.

In described step 4, m factor A of A layer₁,A₂,…,A_m, a is ordered as to general objective Z₁,a₂,…,a_m, B layer n Factor is to factor A in the A of upper strata_jMode of Level Simple Sequence be b_1j,b_2j,…,b_nj(j=1,2 ..., m), the total hierarchial sorting of B layer, I.e. i-th factor of B layer to the weights of general objective is：

In described step 4, the consistency check of total hierarchial sorting：If B is layer B₁,B₂,…,B_nTo factor A in A layer_j(j= 1,2 ..., Mode of Level Simple Sequence coincident indicator m) is CI_j, random index is RI_j, then the uniformity of total hierarchial sorting Ratio is：

As CR ＜ 0.1, level consistency check is passed through, and accepts the result of final sequence；Otherwise will be readjusted those The element value of the high pairwise comparison matrix of Consistency Ratio.

The technology design of the present invention is：The algorithm of this patent is exactly by the analytic hierarchy process (AHP) in Multiobjective Decision Making Method (Analytic Hierarchy Process, AHP) applies in frequency spectrum decision-making, initialization system after updating nodal information, According to the bandwidth of each frequency spectrum recording, time delay, shake, this four attributes of packet loss numerical value, set up using analytic hierarchy process (AHP) Recursive hierarchy structure, arranges corresponding pairwise comparison matrix, finally draws the sequence of frequency spectrum utilization benefit, then utilizes in graph theory Colouring algorithm the optimal spectrum selected is allocated.So both effectively increased the utilization benefit of cognitive user, and to the greatest extent might be used The overall efficiency of whole cognition wireless network can be maximized, thus define one complete from frequency spectrum decision-making to the frame of distribution Structure.

Beneficial effects of the present invention are mainly manifested in：Effectively increase the utilization benefit of cognitive user, but as maximum as possible Change the overall efficiency of whole cognition wireless network, thus define one complete from frequency spectrum decision-making to the framework of distribution.Fully Make use of rare frequency spectrum resource, improve resource utilization.

Brief description

Fig. 1 is the optimal spectrum decision model figure based on AHP algorithm.

Fig. 2 is hierarchy Model schematic diagram.

Fig. 3 is the greedy frequency spectrum allocation algorithm flow chart of binding hierarchy analytic approach.

Fig. 4 is the fair spectrum allocation algorithm flow chart of binding hierarchy analytic approach.

Specific embodiment

The invention will be further described below in conjunction with the accompanying drawings.

Reference Fig. 1～Fig. 4, the frequency spectrum distributing method based on Turing pattern formation and analytic hierarchy process (AHP) in a kind of cognitive radio, Initialization system after updating nodal information, according to the bandwidth of each frequency spectrum recording, time delay, shake, this four genus of packet loss Property numerical value, set up recursive hierarchy structure using analytic hierarchy process (AHP), corresponding pairwise comparison matrix be set, finally show that frequency spectrum makes With the sequence of benefit, then using the colouring algorithm in graph theory, the optimal spectrum selected is allocated；Comprise the following steps：

First, set up recursive hierarchy structure.First according to the relation between each element in decision system, they are decomposed into not Same part, each part is referred to as element, then according to groups elements are formed the level of complementary correlation by attribute, on One layer of some or all of element to next layer plays dominating role, material is thus formed dominance relation successively.Typically can divide For three below level.

A) destination layer (top)：The predeterminated target of finger problem；

B) rule layer (intermediate layer)：Refer to the criterion of impact realization of goal；

C) decision-making level's (lowermost layer)：Refer to promote the measure of realization of goal；

In frequency spectrum selects, cognitive user always wants to be switched to high bandwidth, low time delay, low jitter, low packet loss ratio Deng frequency spectrum on.Therefore, the judgement that selection bandwidth, time delay, shake, this four attributes of packet loss select as frequency spectrum herein is accurate Then.Hierarchy Model is as shown in Figure 1.More interpretational criterias can be introduced in more complex environment so as to closer to reality.

2nd, construct pairwise comparison matrix.After establishing recursive hierarchy structure, between upper and lower level, the membership of element is just true Fixed.Each element of same level is compared two-by-two with regard to the importance of a certain criterion in last layer time, and by its Quantify, just constitute pairwise comparison matrix.

If comparing n factor C of a certain layer₁,C₂,…,C_nImpact to last layer factor O, takes two factors every time C_iAnd C_jCompare it to the importance of target factor O it is simply that calculating corresponding weight W₁,W₂,…,W_n.By the 1-9 shown in table 1 Scaling law corresponding assignment to weight, is defined as a_ij.

Table 1

All comparative result pairwise comparison matrix A represents.

A=(a_ij)_n×n(1)

Pairwise comparison matrix has following property：

When above formula is all set up to pairwise comparison matrix all elements, then this matrix is called consistency matrix.

Here set destination layer A to rule layer B₁,B₂,B₃,B₄Pairwise comparison matrix be：

Wireless terminal real-time monitoring available frequency band, according to active wireless network state obtain each frequency range respectively to bandwidth, Time delay, shake, the parameter value of four attributes of packet loss, determine corresponding weight, thus assigning to the element in pairwise comparison matrix Value, draws rule layer B₁,B₂,B₃,B₄With respect to P₁,P₂,…,P_nPairwise comparison matrix.With B₁To P₁,P₂,…,P_nThe one-tenth of effect As a example comparator matrix：

3rd, Mode of Level Simple Sequence and its consistency check.Mode of Level Simple Sequence refers to the corresponding factor of same level for upper one The weight order of level factor relative importance.Can be by normalizing the Maximum characteristic root λ of judgment matrix A_maxCorresponding spy Levy vectorial (weight vector) W to obtain.So being substantially to calculate weight vector.

Calculating weight vector has eigenvalue method and method, root method, power method etc., adopts herein and method.As follows with the step of method：

F) each column vector of A is normalized：

G) rightSue for peace by row

H) above-mentioned matrix-vector is normalized,Obtain W=(W₁,W₂,…,W_n)^TFor weight vector.

I) calculate AW.

J) calculateThis is the approximation of eigenvalue of maximum.

According to above method, taking matrix A as a example, try to achieve weight vector W=(0.488,0.190,0.089,0.233)^T, it is right The Maximum characteristic root λ answering_max=4.025.In the same manner, B can be calculated₁,B₂,B₃,B₄Maximum characteristic root and its weight vector.

One correct judgment matrix importance ranking has certain logical laws, if for example A is more important than B, B is again than C Important, then logically, A should be obvious more important than C, if the A result more important than C occurs when comparing two-by-two, this judgement square Battle array violates conformance criteria, is logically irrational.Therefore require judgment matrix satisfaction one generally in practice Cause property, need to carry out consistency check.Only pass through inspection, could illustrate that judgment matrix is logically rational, could continue Result is analyzed.

The step of consistency check is as follows：

The first step:Calculate coincident indicator CI (consistency index).

Wherein, n is the exponent number of pairwise comparison matrix.

Second step:Table look-up and determine corresponding random index RI (random index).

It is judged that matrix different rank tables look-up 2, obtain Aver-age Random Consistency Index RI.

Table 2

3rd step:Calculate consistency ration CR (consistency ratio) and judged.

As CR ＜ 0.1 it is believed that the uniformity of judgment matrix is acceptable, during CR ＞ 0.1 it is believed that judgment matrix not Meet coherence request, need this judgment matrix is revised again.

Taking matrix A as a example, carry out consistency check.

Try to achieve CI=0.0083, table look-up and can obtain RI=0.90.Then CR=0.0092 ＜ 0.1.Therefore by inspection.

4th, total hierarchial sorting and its consistency check.Total hierarchial sorting refers to calculate each layer of institute from top to down There is the weight to top (general objective) relative importance for the factor, so just can obtain the weight to target for each scheme in the bottom Sequence, thus carry out the selection of scheme.This process is also required to carry out consistency check.As shown in Fig. 2 m factor A of A layer₁, A₂,…,A_m, a is ordered as to general objective Z₁,a₂,…,a_m.N factor of B layer is to factor A in the A of upper strata_jMode of Level Simple Sequence be b_1j,b_2j,…,b_nj(j=1,2 ..., m), the total hierarchial sorting of B layer, that is, i-th factor of B layer to the weights of general objective be：The total hierarchial sorting table of B layer is as shown in table 3.

Table 3

The consistency check of total hierarchial sorting：If B is layer B₁,B₂,…,B_nTo factor A in upper strata (A layer)_j(j=1,2 ..., m) Mode of Level Simple Sequence coincident indicator be CI_j, random index is RI_j, then the Consistency Ratio of total hierarchial sorting be

As CR ＜ 0.1, level consistency check is passed through, and can accept the result finally sorting.Otherwise to readjust The element value of the high pairwise comparison matrix of those Consistency Ratios.

Arrive this, all usable spectrum resources can be normalized into regard to the relative weighting of general objective through total hierarchial sorting One weight vectors, the frequency spectrum that algorithm is therefrom chosen corresponding to maximum carries out ensuing distribution, improves frequency spectrum effect further Benefit.

5th, distributed greedy algorithm and distributed fairness algorithm and the frequency spectrum Decision Evaluation based on analytic hierarchy process (AHP) are combined Method, proposed by the present invention based on the modified frequency spectrum allocation algorithm of Turing pattern formation and analytic hierarchy process (AHP) realize flow process such as Fig. 3 and Shown in Fig. 4.Initialization system after updating nodal information, according to the bandwidth of each frequency spectrum recording, time delay, shake, packet loss The numerical value of this four attributes, sets up recursive hierarchy structure using analytic hierarchy process (AHP), arranges corresponding pairwise comparison matrix, final Go out the sequence of frequency spectrum utilization benefit, then using the colouring algorithm in graph theory, the optimal spectrum selected is allocated.

Simulation result shows, the frequency spectrum allocation algorithm based on Turing pattern formation and analytic hierarchy process (AHP) improves compared with original algorithm Network trap, makes better use of the frequency spectrum resource of preciousness.

Claims (6)

1. in a kind of cognitive radio the frequency spectrum distributing method based on Turing pattern formation and analytic hierarchy process (AHP) it is characterised in that：Described Frequency spectrum distributing method comprises the following steps：

First, set up recursive hierarchy structure

First according to the relation between each element in decision system, they are decomposed into different parts, each composition portion Point referred to as element, then according to groups elements are formed the level of complementary correlation by attribute, last layer to the part of next layer or All element plays dominating role, material is thus formed dominance relation successively；It is divided into three below level：

A) destination layer, that is, top：The predeterminated target of finger problem；

B) rule layer, i.e. intermediate layer：Refer to the criterion of impact realization of goal；

C) decision-making level, i.e. lowermost layer：Refer to promote the measure of realization of goal；

The decision rule that selection bandwidth, time delay, shake, this four attributes of packet loss select as frequency spectrum；

2nd, construct pairwise comparison matrix

Importance after establishing recursive hierarchy structure, to each element a certain criterion in secondary with regard to last layer of same level Compared two-by-two, and quantified, just constituted pairwise comparison matrix；

3rd, Mode of Level Simple Sequence and its consistency check

Mode of Level Simple Sequence refers to the weight order for certain factor relative importance of last layer time for the corresponding factor of same level, leads to The corresponding characteristic vector of Maximum characteristic root crossing normalization judgment matrix obtains；And consistency check is carried out to judgment matrix；

4th, total hierarchial sorting and its consistency check

Total hierarchial sorting refers to calculate the weight to top relative importance for all factor of each layer from top to down, obtains The weight sequencing to target for each scheme in the bottom, thus carrying out the selection of scheme, this total sequencer procedure is also carried out uniformity Inspection；

All usable spectrum resources can be normalized into a weight with regard to the relative weighting of general objective through total hierarchial sorting Vector, the frequency spectrum that algorithm is therefrom chosen corresponding to maximum carries out ensuing distribution；

5th, using the colouring algorithm in graph theory, the optimal spectrum selected is allocated.

2. the frequency spectrum distributing method based on Turing pattern formation and analytic hierarchy process (AHP) in cognitive radio as claimed in claim 1, its It is characterised by：In described step 2, if n factor C of a certain layer will be compared₁,C₂,…,C_nImpact to last layer factor O, Take two factors C every time_iAnd C_jCompare it to the importance of target factor O it is simply that calculating corresponding weight W₁,W₂,…,W_n, lead to Cross 1-9 scaling law to weight corresponding assignment, be defined as a_ij, whole comparative result pairwise comparison matrix A represent：

A=(a_ij)_n×n(1)

Pairwise comparison matrix has following property：

$\left\{\begin{array}{c}{a}_{ij}>0\\ a_{ji}=1/a_{ij}\\ a_{ii}=1\end{array}\right.---\left(2\right)$

When above formula is all set up to pairwise comparison matrix all elements, then this matrix is called consistency matrix；

Here set destination layer A to rule layer B₁,B₂,B₃,B₄Pairwise comparison matrix be：

$A=\left(\begin{array}{cccc}{a}_{11}& a_{12}& a_{13}& a_{14}\\ a_{21}& a_{22}& a_{23}& a_{24}\\ a_{31}& a_{32}& a_{33}& a_{34}\\ a_{41}& a_{42}& a_{43}& a_{44}\end{array}\right)=\left(\begin{array}{cccc}1& 3& 5& 3\\ 1/3& 1& 2& 1\\ 1/5& 1/2& 1& 1/2\\ 1/3& 1& 2& 1\end{array}\right)---\left(3\right)$

Wireless terminal real-time monitoring available frequency band, according to active wireless network state obtain each frequency range respectively to bandwidth, time delay, Shake, the parameter value of four attributes of packet loss, determine corresponding weight, thus to the element assignment in pairwise comparison matrix, obtaining Go out rule layer B₁,B₂,B₃,B₄With respect to P₁,P₂,…,P_nPairwise comparison matrix, with B₁To P₁,P₂,…,P_nIn contrast with effect Compared with as a example matrix：

$B_1=\left(\begin{array}{cccc}{b}_{11}& b_{12}& ...& b_{1n}\\ b_{21}& b_{22}& ...& b_{2n}\\ ...& ...& b_{ij}& ...\\ b_{n1}& b_{n2}& ...& b_{nn}\end{array}\right)---\left(4\right).$

3. the frequency spectrum distributing method based on Turing pattern formation and analytic hierarchy process (AHP) in cognitive radio as claimed in claim 1 or 2, It is characterized in that：In described step 3, calculate characteristic vector and adopt and method, process is as follows：

A) each column vector of A is normalized：

B) rightSue for peace by row

C) above-mentioned matrix-vector is normalized,Obtain W=(W₁,W₂,…,W_n)^TFor weight vector；

D) calculate AW；

E) calculateThis is the approximation of eigenvalue of maximum.

4. the frequency spectrum distributing method based on Turing pattern formation and analytic hierarchy process (AHP) in cognitive radio as claimed in claim 3, its It is characterised by：In described step 3, the process of consistency check is as follows：

The first step:Calculate coincident indicator CI：

$CI=\frac{{\mathrm{\λ}}_{max}-n}{n-1}---\left(5\right)$

Wherein, n is the exponent number of pairwise comparison matrix；

Second step:Table look-up the corresponding random index RI of determination, it is judged that matrix different rank to obtain mean random consistent Property index RI；

3rd step:Calculate consistency ration CR and judged：

$CR=\frac{CI}{RI}---\left(6\right)$

As CR ＜ 0.1 it is believed that the uniformity of judgment matrix is acceptable, it is believed that judgment matrix does not meet during CR ＞ 0.1 Coherence request, needs this judgment matrix is revised again.

5. the frequency spectrum distributing method based on Turing pattern formation and analytic hierarchy process (AHP) in cognitive radio as claimed in claim 1 or 2, It is characterized in that：In described step 4, m factor A of A layer₁,A₂,…,A_m, a is ordered as to general objective Z₁,a₂,…,a_m, B layer n Individual factor is to factor A in the A of upper strata_jMode of Level Simple Sequence be b_1j,b_2j,…,b_nj(j=1,2 ..., m), the level of B layer is always arranged Sequence, that is, i-th factor of B layer to the weights of general objective be：

6. the frequency spectrum distributing method based on Turing pattern formation and analytic hierarchy process (AHP) in cognitive radio as claimed in claim 5, its It is characterised by：In described step 4, the consistency check of total hierarchial sorting：If B is layer B₁,B₂,…,B_nTo factor A in A layer_j(j= 1,2 ..., Mode of Level Simple Sequence coincident indicator m) is CI_j, random index is RI_j, then the uniformity of total hierarchial sorting Ratio is：

$CR=\frac{\underset{j=1}{\overset{m}{\Σ}}{a}_j{\mathrm{CI}}_j}{\underset{j=1}{\overset{m}{\Σ}}{a}_j{\mathrm{RI}}_j}---\left(7\right)$

As CR ＜ 0.1, level consistency check is passed through, and accepts the result of final sequence；Otherwise to be readjusted those consistent The element value of the high pairwise comparison matrix of sex rate.