CN105471844A - Cloud service dynamic combination method based on trust synthesis - Google Patents

Abstract

A cloud service dynamic combination method based on trust synthesis belongs to the field of trustworthy cloud computing security. A cloud service is decomposed into a set of basic trust and experience trust by defining the trust attribute of the cloud service: (1) basic trust: the nature of trust is regarded as the objective attribute of an object, and the evaluation problem of basic trust is modeled into a judgment problem of the cloud service subject decomposition attribute; and (2) experience trust: trust is defined as object interaction to produce subjective measurement of trust, and the evaluation problem of experience trust is modeled into a judgment problem of the interaction behavior between cloud service subjects. The method comprises the steps of defining the trust attribute of a cloud service, establishing an evaluation mechanism of cloud service basic trust, establishing an evaluation mechanism of cloud service experience trust, establishing a cloud service combined trust propagation model, and calculating the cloud service combined trust value. Simulation experiment results show that trustworthy cloud service can be effectively organized and provided in a continuously-changing cloud environment by use of the proposed method.

Description

A kind of cloud service dynamic composition method based on trust combination

Technical field

The present invention relates to a kind of novel cloud service dynamic composition method, belong to credible cloud computing security fields.

Background technology

In recent years, with cloud computing be the novel computation schema of representative present magnanimity, dynamically, the feature such as autonomous, collaborative, evolution, in order to adapt to the change of service mode and various application demand, cloud computing needs effectively manage with method that is more flexible, that adapt to and provide cloud service.

It is convenient and simultaneously efficient to bring at this novel computation schema of cloud computing, and what can not ignore is also bring huge challenge in data security and privacy.All kinds of services in cloud service combination come from different cloud computing providers, and as the significant consideration in services selection, the accuracy that service quality (QoS, quilityofservice) is evaluated is its important prerequisite selected and basis.But under cloud computing environment, simple QoS cannot guarantee the quality of serving, because the authenticity of QoS data is often difficult to ensure: on the one hand, ISP for reaching certain object, may issue false QoS data, with user cheating; On the other hand, service consumer also likely for certain object, thus provides the QoS data of spurious feedback, or is subject to the impact of oneself factor, and the QoS data of feedback does not often meet with truth.Secondly, may there is the service entities of malice, swindle or falseness in each participant of Services Composition; In addition, to the trust evaluation of cloud service main body interbehavior, just and workable evaluation mechanism is lacked.

The concept that the present invention uses for reference trust computing proposes a kind of cloud service dynamic composition method, the method is by the trust attribute of definition cloud service, be decomposed into the set that basis is trusted and experience is trusted: 1) basis is trusted, the essence of trust is considered as the objective attribute of object, the evaluation problem that basis is trusted is modeled as the decision problem of cloud service main body being decomposed to attribute; 2) experience is trusted, and be defined as trusting the subjective measurement that object interaction produces trust, the evaluation problem that experience is trusted is modeled as the decision problem to interbehavior between cloud service main body.The simulation experiment result shows, the method proposed can effectively be organized and provide believable cloud service under the cloud environment of lasting change.

Summary of the invention

Step (1) definition cloud service trust attribute

It seems from the angle of service compute, the core of the credibility of trust object (TU, TrustUnit) is access authorization and the control of identity-based, namely requires that the main body determined one's identity carries out the operation of conjunction rule to resource.In fact this comprise the trust problem of two aspects: 1) the basic trust attribute of trust object.2) the experience trust attribute of trust object.

Step (2) sets up the evaluation mechanism of trusting on cloud service basis

The definition that step (2.1) basis is trusted

The essence of trust is considered as the objective attribute of object, namely trusts and depend on object and exist, produce along with the generation of object, wither away along with the extinction of object.Can not depart from object itself and consider trust problem, trust attribute and object identity are bound, and the state residing for object itself determined.In order to ensure that operation that trust object TU sends is all reflections strictly according to the facts of its intention, main body must be guaranteed and naturally extend the credibility of attribute, main body and the credibility that naturally extends thereof are defined as basic trust.

Step (2.2) basis trust authorization and refute

Agent object is in basic trusted status T → { C ₁, C ₂, C ₃..., conclusion C ₁, C ₂, C ₃... for main body is in the lower attribute that should have of basic trust state, these attributes can not be exhaustive.Carry out plausible reasonning based on probability calculation relation between attribute, basis is trusted and carries out authorization and refute.

Step (3) sets up the evaluation mechanism that cloud service experience is trusted

The definition that step (3.1) experience is trusted

Although trusting be the objective attribute of object, such attribute may beyond expression of words, measure and assess.Other Subjective and Objective can only by the subjective measurement of trusting it with the mutual generation of object.Based on human society, mutual and coordination mechanism, gives the definition that experience is trusted, and experience trusts the subjective measurement showing as and interbehavior between agent object is produced to trust.

The authorization that step (3.2) experience is trusted

Can cause occurring deviation to the estimation of destination object degree of belief when lacking interactive history or interaction times deficiency when between object.Utilize Chebyshe inequality and Bernstein inequality, draw sample size, estimate that the relation between confidential interval and confidence level is authorized experience trust.

Step (4) sets up cloud service combination belief propagation model

The formal specification that step (4.1) combines cloud service

After the trust of single cloud service is estimated, need to study based on the cloud service dynamic combined algorithm of trusting, first provide the formal specification of related notion in cloud service combination.

Step (4.2) analyzes the route of transmission of trusting in cloud service combination

Trust in cloud service combination and have two kinds of routes of transmission: directly transmit and recommend to transmit.Direct transmission i.e. two main bodys have direct interaction experiences, do not need to recommend to form trusting relationship through third party.Indirect transfer i.e. two main bodys do not have direct interaction experiences, need to recommend to form trusting relationship through third party.

Step (5) calculates cloud service combination trust value

First clear and definite cloud service combination can be carried out trusting the precondition calculated, and the basis trust proposed based on this method and experience are trusted the credible cloud service filtered out and could be participated in calculating.

Step (5.1) calculates the trust value existing between cloud service and can reach in card letter node situation.

Step (5.2) calculates the trust value not existing between cloud service and can reach in card letter node situation.

Accompanying drawing explanation

Fig. 1: cloud service combination schematic diagram.

Fig. 2: Services Composition belief propagation path profile.

Embodiment

Cloud service main body natural quality comprises identity, the computing capability of main body, and it naturally extends and comprises its possible operation platform, operating environment.

Cloud service attribute comprises: the title of cloud service and explanation; The title of geographical group belonging to cloud service, or the position of cloud service; For following the tracks of the label of cloud service; Create or revise the date and time of cloud service; The relevant information of disposing in cloud service.

Step (1) definition cloud service trust attribute

In conjunction with the physiology evidence of trust and the basis of sociology interpersonal relationships model, use for reference the concept of trust computing simultaneously, by trust object (TU, TrustUnit) (BT is trusted based on abstract, BasicTrust) and experience trust two tuples of (ET, ExperienceTrust):

TU＝<BT，ET>(1)

Generally, if the behavior of a trust object TU is always consistent with expection, then can be described as credible (trustworthy).

Step (2) sets up the evaluation mechanism of trusting on cloud service basis

The definition that step (2.1) basis is trusted

Main body natural quality comprises identity, the computing capability of main body, and it naturally extends and comprises possible operation platform, operating environment.Due to body attribute can not exhaustive, the set that abstract definition is a series of attributes that it contains is trusted on the basis of cloud service main body TU agent object is that basic believable T state or basis are incredible state.

When agent object meets community set { C ₁, C ₂, C ₃..., namely

T→{C ₁，C ₂，C ₃，……}.(2)

It is claimed to be that basis is believable, or T state.When agent object all can not meet community set, namely

conclusion cannot meet, (3)

It is claimed to be that basis is believable, or state.

Step (2.2) basis trust authorization and refute

(1) (authorization is trusted on basis) establishes cloud service agent object TU to be that basis is believable, that is: T → { C ₁, C ₂, C ₃..., obtain under the believable condition in cloud service agent object basis, attribute C _ithe probability set up is

P(C _i|T)＝1.(4)

Attribute C _ithe probability itself occurred is P (C _i), P (T|C _i) represent confirmation C _ithe reliability of rear T.So, each attribute C of cloud service main body _iconfirmation, the conjecture that all cloud service main body can be made to be in T state becomes more reliable, and P (T|C _i) probability increase degree and attribute C _iprobability P (the C that itself occurs _i) and the properties C of state _ithe probability occurred be inversely proportional to.

(2) (basis trust is refuted) establishes cloud service agent object TU to be that basis is incredible, or is in state, can under cloud service main body be in basic incredible condition by formula (4), attribute C _iinvalid probability is

$P\left({\stackrel{\&OverBar;}{C}}_{\mathrm{\&iota;}}\right|\stackrel{\&OverBar;}{T})=1-P\left({\stackrel{\&OverBar;}{C}}_{\mathrm{\&iota;}}\right|T)=1-(1-P(C_i|T\left)\right)=1---\left(5\right)$

To each attribute C _irefute, the conjecture that all cloud service main body can be made to be in T state becomes more unreliable, i.e. cloud service main body is in state possibility increase, and probability increase degree and to conclusion C _irefute probability and the probability occurred under state be inversely proportional to.

Step (3) sets up the evaluation mechanism that cloud service experience is trusted

The definition that step (3.1) experience is trusted

Experience trusts the subjective measurement showing as and interbehavior between cloud service agent object TU is produced to trust, if there is m object it provides the probability of success of a certain service to be namely (P is met _i, 1-P _i) 0-1 distribution.By P _ibe defined as main body TU _idegree of belief, because this trust is relevant with internodal interactive history or experience, therefore also referred to as experience degree of belief.TU _imay with other objects TU _jthere occurs n time alternately, i, j ∈ m, wherein number of success is u, and the frequency of failure is v, then object TU _ito TU _jtrust be estimated as

Trust(TU _i|TU _j)＝u/n(6)

The authorization that step (3.2) experience is trusted

For the calculating of degree of belief, in the ideal case, formula (6) provides a certain service probability of success P for cloud service main body _iunbiased esti-mator.But when between object, lacking interactive history, interaction times, directly using formula (6) degree of belief to destination object to estimate by being not enough to.Utilize Chebyshe inequality and Bernstein inequality, introduce sample size, estimate that the relation between confidential interval and confidence level is authorized experience trust.

If two main body TU in cloud service combination _i, TU _j, describe its each interaction results with 0-1 distribution, its result stochastic variable ξ represents, so the expectation average of interaction results is E (ξ)=P, and variance is σ ²(ξ)=σ=P (1-P).If cloud service main body TU _i, TU _jmutual total degree be n, number of success is u, and so u/n is the unbiased esti-mator of P.If intersubjective mutual total degree is sample size n sample range, sets simultaneously and estimate that confidential interval is as allowing the deviation range occurred when ε is calculating experience degree of belief.As sample size n sample range>=p ²(1-p ²)/ε ², start as just by the confidence level that the degree of belief P of u/n to main body estimates, illustrate as sample size n sample range>=p ²(1-p ²)/ε ²time, the main body experience degree of belief that formula (6) calculates can be utilized; Simultaneously when after, approach 1 by the confidence level that the degree of belief P of u/n to main body estimates with index speed, the increase along with sample size is described, the experience degree of belief utilizing formula (6) to calculate is just more reliable.

Step (4) sets up cloud service combination belief propagation model

The formal specification that step (4.1) combines cloud service

If a cloud service is combined as P, as shown in Figure 1, this cloud service combination is made up of multiple cloud service provider, P={w ₁... w _n.And each cloud service provider can provide multiclass to have the cloud service of difference in functionality, w _i={ S _i1... S _in.Define two tuples (S, L), wherein S represents involved cloud service set, and L represents the set of cloud service combinatorial path, such as L (S ₁₁→ S ₂₃) represent S ₁₁and S ₂₃the service combination path formed, and there is direct interaction.

Step (4.2) analyzes the route of transmission of trusting in cloud service combination

Fig. 2 illustrates the route of transmission of trusting in cloud service combination.Direct transmission as in Fig. 2, cloud service S ₁₂and S ₂₁between have direct interaction experiences, trust just can directly transmit.Indirect transfer as in Fig. 2, S ₁₂to S ₆₃between there is no direct interaction experiences, need to pass through S ₃₅or S ₅₂indirect transfer forms forward travel path S ₁₂→ S ₃₅→ S ₆₃or S ₁₂→ S ₅₂→ S ₆₃transmit and trust.

But, trust recommendation pipeline in, trust not always forward transmit.As in Fig. 2, S ₁₂and S ₄₆between there is not the Trust transitivity path of forward.Using for reference the related notion in belief propagation, providing as given a definition.

Definition 1: can reach card letter node two does not have the main body of direct interaction experiences can carry out trust evaluation by third-party recommendation, and forms forward belief propagation path, i.e. A → B, B → C, A → B → C, so claims B for reaching card letter node.

As shown in Figure 2, S ₁₂and S ₄₆between do not exist can reach card letter node, but there is common interactive object S in them ₂₁and S ₃₅.By comparing S ₁₂→ S ₂₁and S ₄₆→ S ₂₁trust tightness q (namely to same interaction agent S ₂₁the similarity of evaluate trust), form reverse travel path S ₁₂→ S ₂₁← S ₄₆transmit and trust.

Step (5) calculates cloud service combination trust value

Set the trust value ∈ [0,1] between two cloud services, function T:W × W → [0,1]; A, B ∈ W, represent the trust value (comprise directly trust and indirectly trust) of A to B.

Define tlv triple <b, d, a u> and represent that a cloud service is to the trusting degree of another cloud service, and meet

b+d+u＝1(7)

Wherein b, d, u assign weight represent respectively trust, distrust and uncertain.

Step (5.1) calculates the trust value existing between cloud service and can reach in card letter node situation.

Definition 2 connects operator if A to the trust of B is b to the trust of C is $T_C^B$ $=(b_C^B,d_C^B,u_C^B),$ Then A to the trust of C is $T_C^{AB}=(b_C^{AB},d_C^{AB},u_C^{AB}).$

Wherein, $b_C^{AB}=b_B^A{b}_C^B,d_C^{AB}=b_B^A{d}_C^B,u_C^{AB}=1-b_B^A{b}_C^B-b_B^A{d}_C^B.$ By using symbol assigned operation, definition $T_C^{AB}=T_B^A\&CircleTimes;T_C^B.$

Define 3 Aggregation Operators if A to the trust of C is b to the trust of C is $T_C^B$ $=(b_C^B,d_C^B,u_C^B),$ Then A, B to the trust of C are $T_C^{AB}=(b_C^{AB},d_C^{AB},u_C^{AB}).$

Wherein, $b_C^{AB}=\frac{b_C^A{u}_C^B+b_C^B{u}_C^A}{K},d_C^{AB}=\frac{d_C^A{u}_C^B+d_C^B{u}_C^A}{K},u_C^{AB}=1-\frac{b_C^A{u}_C^B+b_C^B{u}_C^A}{K}-\frac{d_C^A{u}_C^B+d_C^B{u}_C^A}{K},$ $K=$ $u_C^A+u_C^B-u_C^A{u}_C^B,$ K≠0。By using symbol assigned operation, definition $T_C^{AB}=T_C^A\&CirclePlus;T_C^B.$

If A to C has two trust recommendation forward travel path A → B → C, A → D → C.Setting the trust value that this two paths obtains respectively is for path A → B → C, if A to the trust value of B is b to the trust value of C is utilize trust operator to try to achieve trust, obtain

$T{1}_C^A=T_B^A\&CircleTimes;T_C^B---\left(8\right)$

In like manner, try to achieve

$T{2}_C^A=T_D^A\&CircleTimes;T_C^D---\left(9\right)$

So, last overall trust value

$T_C^A=T{1}_C^A\&CirclePlus;T{2}_C^A---\left(10\right)$

Step (5.2) calculates the trust value not existing between cloud service and can reach in card letter node situation.

If A to D has two trust recommendation reverse travel path A → B ← C, A → D ← C.Setting the trust value that this two paths obtains respectively is for path A → B ← C, if A to the trust value of B is c to the trust value of B is utilize and trust algorithm, calculate the trust tightness q1 of A → B and C → B, obtain wherein

$q1=\sqrt{{({\mathrm{\&alpha;}}_B^2-\frac{(u_B^C+1)}{(u_B^C+v_B^C+2)})}^2+\left(\frac{(u_B^C+1)(v_B^C+1)}{{(u_B^C+v_B^C+2)}^2(u_B^C+v_B^C+3)}\right)}---\left(11\right)$

${\mathrm{\&alpha;}}_B^A=Trust\left(A\right|B)=u_B^A/n_B^A---\left(12\right)$

In like manner, obtain

$T{2}_C^A=<q2,1-q2>---\left(13\right)$

If A and C has n common interactive object, then last overall trust value

$T_C^A=<{\mathrm{\&Sigma;}}_{n=1}^N{q}_n,{\mathrm{\&Sigma;}}_{n=1}^N(1-q_n)>---\left(14\right)$

Claims (2)

1. based on a cloud service dynamic composition method for trust combination, it is characterized in that: the implementation process of the method is as follows,

Step (1) definition cloud service trust attribute

It seems from the angle of service compute, the core of the credibility of trust object (TU, TrustUnit) is access authorization and the control of identity-based, namely requires that the main body determined one's identity carries out the operation of conjunction rule to resource; In fact this comprise the trust problem of two aspects: 1) the basic trust attribute of trust object; 2) the experience trust attribute of trust object;

Step (2) sets up the evaluation mechanism of trusting on cloud service basis

The definition that step (2.1) basis is trusted

The essence of trust is considered as the objective attribute of object, namely trusts and depend on object and exist, produce along with the generation of object, wither away along with the extinction of object; Can not depart from object itself and consider trust problem, trust attribute and object identity are bound, and the state residing for object itself determined; In order to ensure that operation that trust object TU sends is all reflections strictly according to the facts of its intention, main body must be guaranteed and naturally extend the credibility of attribute, main body and the credibility that naturally extends thereof are defined as basic trust;

Step (2.2) basis trust authorization and refute

Agent object is in basic trusted status T → { C ₁, C ₂, C ₃..., conclusion C ₁, C ₂, C ₃... for main body is in the lower attribute that should have of basic trust state, these attributes can not be exhaustive; Carry out plausible reasonning based on probability calculation relation between attribute, basis is trusted and carries out authorization and refute;

Step (3) sets up the evaluation mechanism that cloud service experience is trusted

The definition that step (3.1) experience is trusted

Although trusting be the objective attribute of object, such attribute may beyond expression of words, measure and assess; Other Subjective and Objective can only by the subjective measurement of trusting it with the mutual generation of object; Based on human society, mutual and coordination mechanism, gives the definition that experience is trusted, and experience trusts the subjective measurement showing as and interbehavior between agent object is produced to trust;

The authorization that step (3.2) experience is trusted

Can cause occurring deviation to the estimation of destination object degree of belief when lacking interactive history or interaction times deficiency when between object; Utilize Chebyshe inequality and Bernstein inequality, draw sample size, estimate that the relation between confidential interval and confidence level is authorized experience trust;

Step (4) sets up cloud service combination belief propagation model

The formal specification that step (4.1) combines cloud service

After the trust of single cloud service is estimated, need to study based on the cloud service dynamic combined algorithm of trusting, first provide the formal specification of related notion in cloud service combination;

Step (4.2) analyzes the route of transmission of trusting in cloud service combination

Trust in cloud service combination and have two kinds of routes of transmission: directly transmit and recommend to transmit; Direct transmission i.e. two main bodys have direct interaction experiences, do not need to recommend to form trusting relationship through third party; Indirect transfer i.e. two main bodys do not have direct interaction experiences, need to recommend to form trusting relationship through third party;

Step (5) calculates cloud service combination trust value

First clear and definite cloud service combination can be carried out trusting the precondition calculated, and the basis trust proposed based on this method and experience are trusted the credible cloud service filtered out and could be participated in calculating;

Step (5.1) calculates the trust value existing between cloud service and can reach in card letter node situation;

Step (5.2) calculates the trust value not existing between cloud service and can reach in card letter node situation.

2. a kind of cloud service dynamic composition method based on trust combination according to claim 1, is characterized in that: the implementation process of the method is as follows,

Cloud service main body natural quality comprises identity, the computing capability of main body, and it naturally extends and comprises its possible operation platform, operating environment;

Cloud service attribute comprises: the title of cloud service and explanation; The title of geographical group belonging to cloud service, or the position of cloud service; For following the tracks of the label of cloud service; Create or revise the date and time of cloud service; The relevant information of disposing in cloud service;

Step (1) definition cloud service trust attribute

In conjunction with the physiology evidence of trust and the basis of sociology interpersonal relationships model, use for reference the concept of trust computing simultaneously, by trust object (TU, TrustUnit) (BT is trusted based on abstract, BasicTrust) and experience trust two tuples of (ET, ExperienceTrust):

TU＝<BT，ET>(1)

Generally, if the behavior of a trust object TU is always consistent with expection, then can be described as credible (trustworthy);

Step (2) sets up the evaluation mechanism of trusting on cloud service basis

The definition that step (2.1) basis is trusted

Main body natural quality comprises identity, the computing capability of main body, and it naturally extends and comprises possible operation platform, operating environment; Due to body attribute can not exhaustive, the set that abstract definition is a series of attributes that it contains is trusted on the basis of cloud service main body TU agent object is that basic believable T state or basis are incredible state;

When agent object meets community set { C ₁, C ₂, C ₃..., namely

T→{C ₁，C ₂，C ₃，……}.(2)

It is claimed to be that basis is believable, or T state; When agent object all can not meet community set, namely

conclusion cannot meet, (3)

It is claimed to be that basis is believable, or state;

Step (2.2) basis trust authorization and refute

(1) (authorization is trusted on basis) establishes cloud service agent object TU to be that basis is believable, that is: T → { C ₁, C ₂, C ₃..., obtain under the believable condition in cloud service agent object basis, attribute C _ithe probability set up is

P(C _i|T)＝1.(4)

Attribute C _ithe probability itself occurred is P (C _i), P (T|C _i) represent confirmation C _ithe reliability of rear T; So, each attribute C of cloud service main body _iconfirmation, the conjecture that all cloud service main body can be made to be in T state becomes more reliable, and P (T|C _i) probability increase degree and attribute C _iprobability P (the C that itself occurs _i) and the properties C of state _ithe probability occurred be inversely proportional to;

(2) (basis trust is refuted) establishes cloud service agent object TU to be that basis is incredible, or is in state, can under cloud service main body be in basic incredible condition by formula (4), attribute C _iinvalid probability is

$P\left({\stackrel{\&OverBar;}{C}}_{\mathrm{\&iota;}}\right|\stackrel{\&OverBar;}{T})=1-P\left({\stackrel{\&OverBar;}{C}}_{\mathrm{\&iota;}}\right|T)=1-(1-P(C_i|T\left)\right)=1---\left(5\right)$

To each attribute C _irefute, the conjecture that all cloud service main body can be made to be in T state becomes more unreliable, i.e. cloud service main body is in state possibility increase, and probability increase degree and to conclusion C _irefute probability and the probability occurred under state be inversely proportional to;

Step (3) sets up the evaluation mechanism that cloud service experience is trusted

The definition that step (3.1) experience is trusted

Experience trusts the subjective measurement showing as and interbehavior between cloud service agent object TU is produced to trust, if there is m object it provides the probability of success of a certain service to be namely (P is met _i, 1-P _i) 0-1 distribution; By P _ibe defined as main body TU _idegree of belief, because this trust is relevant with internodal interactive history or experience, therefore also referred to as experience degree of belief; TU _imay with other objects TU _jthere occurs n time alternately, i, j ∈ m, wherein number of success is u, and the frequency of failure is v, then object TU _ito TU _jtrust be estimated as

Trust(TU _i|TU _j)＝u/n(6)

The authorization that step (3.2) experience is trusted

For the calculating of degree of belief, in the ideal case, formula (6) provides a certain service probability of success P for cloud service main body _iunbiased esti-mator; But when between object, lacking interactive history, interaction times, directly using formula (6) degree of belief to destination object to estimate by being not enough to; Utilize Chebyshe inequality and Bernstein inequality, introduce sample size, estimate that the relation between confidential interval and confidence level is authorized experience trust;

If two main body TU in cloud service combination _i, TU _j, describe its each interaction results with 0-1 distribution, its result stochastic variable ξ represents, so the expectation average of interaction results is E (ξ)=P, and variance is σ ²(ξ)=σ=P (1-P); If cloud service main body TU _i, TU _jmutual total degree be n, number of success is u, and so u/n is the unbiased esti-mator of P; If intersubjective mutual total degree is sample size n sample range, sets simultaneously and estimate that confidential interval is as allowing the deviation range occurred when ε is calculating experience degree of belief; As sample size n sample range>=p ²(1-p ²)/ε ², start as just by the confidence level that the degree of belief P of u/n to main body estimates, illustrate as sample size n sample range>=p ²(1-p ²)/ε ²time, the main body experience degree of belief that formula (6) calculates can be utilized; Simultaneously when after, approach 1 by the confidence level that the degree of belief P of u/n to main body estimates with index speed, the increase along with sample size is described, the experience degree of belief utilizing formula (6) to calculate is just more reliable;

Step (4) sets up cloud service combination belief propagation model

The formal specification that step (4.1) combines cloud service

If a cloud service is combined as P, this cloud service combination is made up of multiple cloud service provider, P={w ₁... w _n; And each cloud service provider can provide multiclass to have the cloud service of difference in functionality, w _i={ S _i1... S _in; Define two tuples (S, L), wherein S represents involved cloud service set, and L represents the set of cloud service combinatorial path, such as L (S ₁₁→ S ₂₃) represent S ₁₁and S ₂₃the service combination path formed, and there is direct interaction;

Step (4.2) analyzes the route of transmission of trusting in cloud service combination

Cloud service S ₁₂and S ₂₁between have direct interaction experiences, trust just can directly transmit; In indirect transfer, S ₁₂to S ₆₃between there is no direct interaction experiences, need to pass through S ₃₅or S ₅₂indirect transfer forms forward travel path S ₁₂→ S ₃₅→ S ₆₃or S ₁₂→ S ₅₂→ S ₆₃transmit and trust;

But, trust recommendation pipeline in, trust not always forward transmit; S ₁₂and S ₄₆between there is not the Trust transitivity path of forward; Using for reference the related notion in belief propagation, providing as given a definition;

Definition 1: can reach card letter node two does not have the main body of direct interaction experiences can carry out trust evaluation by third-party recommendation, and forms forward belief propagation path, i.e. A → B, B → C, A → B → C, so claims B for reaching card letter node;

S ₁₂and S ₄₆between do not exist can reach card letter node, but there is common interactive object S in them ₂₁and S ₃₅; By comparing S ₁₂→ S ₂₁and S ₄₆→ S ₂₁trust tightness q (namely to same interaction agent S ₂₁the similarity of evaluate trust), form reverse travel path S ₁₂→ S ₂₁← S ₄₆transmit and trust;

Step (5) calculates cloud service combination trust value

Set the trust value ∈ [0,1] between two cloud services, function T:W × W → [0,1]; A, B ∈ W, represent the trust value (comprise directly trust and indirectly trust) of A to B;

Define tlv triple <b, d, a u> and represent that a cloud service is to the trusting degree of another cloud service, and meet

b+d+u＝1(7)

Wherein b, d, u assign weight represent respectively trust, distrust and uncertain;

Step (5.1) calculates the trust value existing between cloud service and can reach in card letter node situation;

Definition 2 connects operator if A to the trust of B is b to the trust of C is $T_C^B$ $=(b_C^B,d_C^B,u_C^B),$ Then A to the trust of C is $T_C^{AB}=(b_C^{AB},d_C^{AB},u_C^{AB});$

Wherein, $b_C^{AB}=b_B^A{b}_C^B,d_C^{AB}=b_B^A{d}_C^B,$ $u_C^{AB}=1-b_B^A{b}_C^B-b_B^A{d}_C^B;$ By using symbol assigned operation, definition $T_C^{AB}=T_B^A\&CircleTimes;T_C^B;$

Define 3 Aggregation Operators if A to the trust of C is b to the trust of C is $T_C^B$ $=(b_C^B,d_C^B,u_C^B),$ Then A, B to the trust of C are $T_C^{AB}=(b_C^{AB},d_C^{AB},u_C^{AB});$

Wherein, $b_C^{AB}=\frac{b_C^A{u}_C^B+b_C^B{u}_C^A}{K},$ $d_C^{AB}=\frac{d_C^A{u}_C^B+d_C^B{u}_C^A}{K},$ $u_C^{AB}=1-\frac{b_C^A{u}_C^B+b_C^B{u}_C^A}{K}-\frac{d_C^A{u}_C^B+d_C^B{u}_C^A}{K},$ $K=$ $u_C^A+u_C^B-u_C^A{u}_C^B,$ K ≠ 0; By using symbol assigned operation, definition

If A to C has two trust recommendation forward travel path A → B → C, A → D → C; Setting the trust value that this two paths obtains respectively is for path A → B → C, if A to the trust value of B is b to the trust value of C is utilize trust operator to try to achieve trust, obtain

$T{1}_C^A=T_B^A\&CircleTimes;T_C^B---\left(8\right)$

In like manner, try to achieve

$T{2}_C^A=T_D^A\&CircleTimes;T_C^D---\left(9\right)$

So, last overall trust value

$T_C^A=T{1}_C^A\&CirclePlus;T{2}_C^A---\left(10\right)$

Step (5.2) calculates the trust value not existing between cloud service and can reach in card letter node situation;

If A to D has two trust recommendation reverse travel path A → B ← C, A → D ← C; Setting the trust value that this two paths obtains respectively is for path A → B ← C, if A to the trust value of B is c to the trust value of B is utilize and trust algorithm, calculate the trust tightness q1 of A → B and C → B, obtain $T{1}_C^A=<q1,1-q1>,$ Wherein

$q1=\sqrt{{({\mathrm{\&alpha;}}_B^A-\frac{(u_B^C+1)}{(u_B^C+v_B^C+2)})}^2+\left(\frac{(u_B^C+1)(v_B^C+1)}{{(u_B^C+v_B^C+2)}^2(u_B^C+v_B^C+3)}\right)}---\left(11\right)$

${\mathrm{\&alpha;}}_B^A=Trust\left(A\right|B)=u_B^A/n_B^A---\left(12\right)$

In like manner, obtain

$T{2}_C^A=<q2,1-q2>---\left(13\right)$

If A and C has n common interactive object, then last overall trust value

$T_C^A=<{\mathrm{\&Sigma;}}_{n=1}^N{q}_n,{\mathrm{\&Sigma;}}_{n=1}^N(1-q_n)>---\left(14\right).$