CN104639638B - Based on user's updating belief method that dynamic game opinion is classified with cloud service - Google Patents
Based on user's updating belief method that dynamic game opinion is classified with cloud service Download PDFInfo
- Publication number
- CN104639638B CN104639638B CN201510066543.8A CN201510066543A CN104639638B CN 104639638 B CN104639638 B CN 104639638B CN 201510066543 A CN201510066543 A CN 201510066543A CN 104639638 B CN104639638 B CN 104639638B
- Authority
- CN
- China
- Prior art keywords
- user
- cloud service
- service provider
- service request
- request
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 239000011159 matrix material Substances 0.000 claims abstract description 22
- 230000004044 response Effects 0.000 claims abstract description 10
- 238000012937 correction Methods 0.000 claims description 53
- 230000002159 abnormal effect Effects 0.000 claims description 48
- 230000006399 behavior Effects 0.000 claims description 17
- 230000000977 initiatory effect Effects 0.000 claims description 4
- 230000008901 benefit Effects 0.000 description 8
- 230000008569 process Effects 0.000 description 7
- 230000008859 change Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 230000007123 defense Effects 0.000 description 5
- 230000035945 sensitivity Effects 0.000 description 5
- 238000001514 detection method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 238000007670 refining Methods 0.000 description 3
- 230000007423 decrease Effects 0.000 description 2
- 238000010295 mobile communication Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Storage Device Security (AREA)
Abstract
The present invention relates to a kind of user's updating belief method of classifying with cloud service based on dynamic game opinion, comprise the following steps: (1) sets up the payoff matrix of cloud service provider and trusted users, cloud service provider and insincere user respectively, set up the incomplete information dynamic game model of cloud service provider and trusted users, cloud service provider and insincere user based on this; (2) cloud service is divided into sensitive service and non-sensitive service, carries out updating belief respectively, with monitor user ' behavior; (3) in conjunction with the current beliefs value of the historical behavior sequence calculating user of user, to provide different response strategies.The method is conducive to cloud service is averaged out in continuity and fail safe two.
Description
Technical Field
The invention relates to the technical field of cloud services, in particular to a user belief correction method based on a dynamic game theory and cloud service classification.
Background
In recent years, cloud computing has led to a new business revolution as a new industrial model, and has caused a booming of researchers around the world. However, the existing cloud system architecture enables a cloud service system to have a huge user amount, especially the diversity of user behaviors, which causes the system to face more threats which are difficult to prevent; the high availability and fault tolerance of the system is also an important aspect of security, since the use of too harsh security policies may cause service interruptions for users. Therefore, the cloud service provider must improve its services from several aspects: availability management, access control management and access monitoring; there is an inconsistent diversity in complex cloud environments, and the key to cloud computing success is the balance between business interests and potential risks.
Due to service diversity in a cloud computing environment, the cloud computing environment has different importance and sensitivity, in some existing researches, credibility analysis is carried out on user (including an attacker) behaviors, and then safety protection of a system is achieved through a safety strategy, and the methods generally have the problems of poor adaptivity, poor fault tolerance and the like, namely, the cloud computing environment with low level can be too harsh, so that the continuity of cloud services is poor, and the cloud computing environment with high sensitivity has insufficient safety defense and the like. For example, in the application document [1] (document [1 ]: chenui, field, yangyang, cloud computing environment, based on a dynamic game theory, user behavior model and analysis [ J ] electronics report 2011(08):1818 1823.), when the game times are increased, the method for correcting the beliefs is found to be too severe, namely the belief value is reduced too slowly, namely the belief value cannot be reduced by long-time normal requests of the user in the later period, so as to recover the belief value, thereby improving the service access authority, and experiments also find that the belief value is basically unchanged even if the normal request times (namely the game times) of the user reach 90 times. The problems that this mechanism causes: 1) the system can reject the normal request of the user for a long time due to the early misoperation of the user; 2) for non-sensitive services, the method also greatly influences the service continuity and availability of the system, and the cloud service provider benefits are reduced.
Disclosure of Invention
The invention aims to provide a user belief correction method based on a dynamic game theory and cloud service classification, which is favorable for balancing the cloud service in the aspects of continuity and safety.
In order to achieve the purpose, the technical scheme of the invention is as follows: a user belief correction method based on a dynamic game theory and cloud service classification is characterized by comprising the following steps:
(1) respectively establishing payment matrixes of a cloud service provider and a trusted user and payment matrixes of the cloud service provider and a non-trusted user, and establishing an incomplete information dynamic game model of the cloud service provider and the trusted user and the incomplete information dynamic game model of the cloud service provider and the non-trusted user based on the payment matrixes;
(2) the cloud service is divided into sensitive service and non-sensitive service, and belief correction is respectively carried out to monitor user behaviors;
(3) and calculating the current belief value of the user by combining the historical behavior sequence of the user to give different reaction strategies.
Further, the payment matrix of the cloud service provider and the trusted user is as follows: the trusted user sends a normal service request, and matrix payment values corresponding to the service request accepted, the service request not responded and the service request rejected by the cloud service provider are respectively as follows: ((1-α)(U n -C n ),(1-α)(S n -Q n ))、(-αq i C n ,-αq i (S n +Q n ))、(-α(1-q i )(C n +P a ),-α(1-q i )(S n +Q n ))；
The payment matrix of the cloud service provider and the untrusted user is as follows: the method comprises the following steps that an untrusted user sends an abnormal service request, and matrix payment values corresponding to the service request acceptance, service request non-response and service request rejection of a cloud service provider are respectively as follows: (β(U a -C a ),-β(U a +Q n ))、(-(1-β)q i (C a ),(1-β)q i (U i -Q n ))、(-(1-β)(1-q i )(C a +P a ),(1-β)(1-q i )(U a -Q n ) ); the untrusted user sends out a normal service request, and matrix payment values corresponding to the service request accepted, the service request not responded and the service request rejected by the cloud service provider are respectively as follows: ((1-α)(U n -C n ),(1-α)(S n -Q n ))、(-αq i C n ,-αq i (S n +Q n ))、(-α(1-q i )(C n +P a ),-α(1-q i )(S n +Q n ))；
Wherein each matrix payment value (A, B) comprises two terms, the former term A represents the profit of the user, and the latter term represents the profit of the cloud service provider;αin order to realize the false alarm rate of the system,U n the benefit obtained when the user issues a normal service request to be allowed access,C n the overhead of initiating a normal service request for the user,S n the income obtained when the cloud service provider allows the normal service request, and the utility when refusing the normal service request is-S n ，Q n Receives a request overhead for the cloud service provider,q i the probability of not responding is adopted when the cloud service provider receives an abnormal service request,P a sending an abnormal service request for a user is penalized by detection,U a the utility gained when sending out an abnormal service request for an untrusted user, which is not blocked, is also lost revenue for the cloud service provider,C a the cost of sending an unusual service request for an untrusted user,U i for the benefit of the cloud service provider in adopting the non-response strategy for the abnormal service request,βthe system failure rate.
Further, the belief correction method of the sensitive service comprises the following steps: assuming game by adopting refined Bayesian equilibrium correction modelAt the nth stage, the belief of the cloud service provider to the user is BnThe untrusted user sends an abnormal service request with a probability P, and the trusted user sends a normal service request with a probability 1; for a cloud service provider:
probability of sending normal service request by untrusted user；
Probability of different service requests sent by untrusted users；
When the cloud service provider receives the normal service request, if Bn>0.5，Bn+1Comprises the following steps:
(1)
otherwise Bn+1= 0.5; wherein, Bn+1Beliefs of the cloud service provider to the user for the (n + 1) th stage of the game;
when the cloud service provider receives the abnormal service request, Bn+1Comprises the following steps:
(2)
in the above formulas (1) and (2), p is:
(3)
the belief correction method of the non-sensitive service comprises the following steps: based on a refined Bayesian equilibrium correction model, supposing that the nth stage of the game is that the belief of the cloud service provider to the user is BnIs not limited toThe method comprises the steps that a credible user sends an abnormal service request with a probability P, and the credible user sends a normal service request with a probability 1; for a cloud service provider:
probability of sending normal service request by untrusted user；
Probability of different service requests sent by untrusted users；
When the cloud service provider receives the normal service request, if Bn>0.5, then Bn+1Comprises the following steps:
(4)
otherwise Bn+1=0.5；
When the cloud service provider receives the abnormal service request, Bn+1Comprises the following steps:
(5)。
the invention has the advantages that a dynamic game theory and cloud service classification-based user belief correction method is provided, the method firstly constructs an incomplete information dynamic game model between a cloud service provider and a user, analyzes the behavior sequence of the user at each stage, then divides the cloud service into sensitive service and non-sensitive service according to the importance of the cloud service accessed by the user, monitors the user behavior by adopting an improved belief correction method, and finally calculates the current belief value of the user by combining the historical behavior sequence of the user to give different reaction strategies, so that the continuity of the cloud service is ensured to the maximum extent, and meanwhile, the safety of the cloud service can be improved. The method solves the problems that the credibility of user behaviors is usually only considered and a uniform security strategy is adopted in the prior art, namely the cloud service continuity is poor due to the fact that the cloud service of low level is possibly too harsh, and the security defense is insufficient for the cloud service of high sensitivity.
Drawings
FIG. 1 is a flow chart of an implementation of an embodiment of the present invention.
Fig. 2 is a representation of a cloud service provider and a trusted user gaming extension in an embodiment of the present invention.
Fig. 3 is a representation of gaming extension of a cloud service provider and a non-creditable user according to an embodiment of the invention.
Fig. 4 is a diagram of belief change of the algorithm in document [1] in the case of the game of the request sequences a1 and B1 in the embodiment of the present invention (after 90 games above).
Fig. 5 is a diagram of belief change of the algorithm of document [1] in the case of gaming the request sequences a1 and B1 in the embodiment of the present invention (after 30 miscellaneous requests are completed, 60 normal service requests are continuously sent).
Fig. 6 is a diagram of the belief correction of "sensitive services" in the case of gaming the request sequences a1 and B1 in the embodiment of the present invention.
Fig. 7 is a diagram of the belief correction of "sensitive services" in the case of gaming the request sequences a2 and B2 in the embodiment of the present invention.
Fig. 8 is a diagram of the belief correction of "insensitive service" when the request sequences a1 and B1 are played in the embodiment of the present invention.
Fig. 9 is a diagram of the belief correction of "insensitive service" when the request sequences a2 and B2 are played in the embodiment of the present invention.
Detailed Description
The invention discloses a user belief correction method based on a dynamic game theory and cloud service classification, which comprises the following steps as shown in figure 1:
(1) payment matrixes of the cloud service provider and the trusted users and payment matrixes of the cloud service provider and the non-trusted users are respectively established, and incomplete information dynamic game models of the cloud service provider and the trusted users and the cloud service provider and the non-trusted users are established based on the payment matrixes.
The payment matrix of the cloud service provider and the trusted user is as follows: the trusted user sends a normal service request, and matrix payment values corresponding to the service request accepted, the service request not responded and the service request rejected by the cloud service provider are respectively as follows: ((1-α)(U n -C n ),(1-α)(S n -Q n ))、(-αq i C n ,-αq i (S n +Q n ))、(-α(1-q i )(C n +P a ),-α(1-q i )(S n +Q n ))；
The payment matrix of the cloud service provider and the untrusted user is as follows: the method comprises the following steps that an untrusted user sends an abnormal service request, and matrix payment values corresponding to the service request acceptance, service request non-response and service request rejection of a cloud service provider are respectively as follows: (β(U a -C a ),-β(U a +Q n ))、(-(1-β)q i (C a ),(1-β)q i (U i -Q n ))、(-(1-β)(1-q i )(C a +P a ),(1-β)(1-q i )(U a -Q n ) ); the untrusted user sends out a normal service request, and the cloud service provider receives the service requestMatrix payment values corresponding to the non-response service request and the rejection service request are respectively as follows: ((1-α)(U n -C n ),(1-α)(S n -Q n ))、(-αq i C n ,-αq i (S n +Q n ))、(-α(1-q i )(C n +P a ),-α(1-q i )(S n +Q n ))；
Wherein each matrix payment value (A, B) comprises two terms, the former term A represents the profit of the user, and the latter term represents the profit of the cloud service provider;αin order to realize the false alarm rate of the system,U n the benefit obtained when the user issues a normal service request to be allowed access,C n the overhead of initiating a normal service request for the user,S n the income obtained when the cloud service provider allows the normal service request, and the utility when refusing the normal service request is-S n ，Q n Receives a request overhead for the cloud service provider,q i the probability of not responding is adopted when the cloud service provider receives an abnormal service request,P a sending an abnormal service request for a user is penalized by detection,U a the utility gained when sending out an abnormal service request for an untrusted user, which is not blocked, is also lost revenue for the cloud service provider,C a the cost of sending an unusual service request for an untrusted user,U i for the benefit of the cloud service provider in adopting the non-response strategy for the abnormal service request,βthe system failure rate.
(2) The cloud service is divided into sensitive service and non-sensitive service, and belief correction is respectively carried out to monitor user behaviors.
The belief correction method of the sensitive service comprises the following steps: using refining BayesThe balance correction model assumes that the Nth stage of the game, and the belief of the cloud service provider to the user is BnThe untrusted user sends an abnormal service request with a probability P, and the trusted user sends a normal service request with a probability 1; for a cloud service provider:
probability of sending normal service request by untrusted user；
Probability of different service requests sent by untrusted users；
When the cloud service provider receives the normal service request, if Bn>0.5，Bn+1Comprises the following steps:
(1)
otherwise Bn+1= 0.5; wherein, Bn+1Beliefs of the cloud service provider to the user for the (n + 1) th stage of the game;
when the cloud service provider receives the abnormal service request, Bn+1Comprises the following steps:
(2)
in the above formulas (1) and (2), p is:
(3)
the belief correction method of the non-sensitive service comprises the following steps: based on a refined Bayes equilibrium correction model, assuming the nth stage of the game, cloudThe belief of the service provider to the user is BnThe untrusted user sends an abnormal service request with a probability P, and the trusted user sends a normal service request with a probability 1; for a cloud service provider:
probability of sending normal service request by untrusted user；
Probability of different service requests sent by untrusted users；
When the cloud service provider receives the normal service request, if Bn>0.5, then Bn+1Comprises the following steps:
(4)
otherwise Bn+1=0.5；
When the cloud service provider receives the abnormal service request, Bn+1Comprises the following steps:
(5)。
(3) and calculating the current belief value of the user by combining the historical behavior sequence of the user to give different reaction strategies.
The present invention and related art to which the present invention relates are further described below.
Game model
The security situation of the cloud service is abstracted into a continuous multiple incomplete information dynamic game process between a cloud service provider and a user, and the belief value of the user is calculated dynamically through a belief correction method in the game process so as to give a strategy. The following gives the set of policies and the payout matrix for both parties to the game and ultimately the game model. In order to make the process of the game have research significance, the model has the following premises:
(1) both sides of the game are rational, and the adopted strategies are rational: the service provider pursues the self income and can not provide the service with negative income; the untrusted user will not launch an unprofitable attack.
(2) Both sides of the game aim to pursue the maximum profit: the two parties of the game adopt the strategy in the game process according to the maximum effectiveness obtained by adopting the strategy.
Game payment matrix
Game model GM (Game model) is composed of three parts to form a game man setP s Policy collectionS s Policy utilityU s ；
Define 1 office people set: P s (Personset):P s ={P c (the cloud service provider) of the mobile communication terminal,P u (user) };P u ={P 0(the trusted user) of the user,P 1(untrusted users) };
define 2 policy set:policy collectionS s ={S c (the cloud service provider) of the mobile communication terminal,S u (user) }:S c ={A(the service request is accepted),I(not responding to the service request),D(deny the service request);S p0={Q 0(normal service request) };S p1={{Q 0(Normal service request), a retaining pocketQ 1(exception service request) };
define 3 payment matrices:the payment or utility of different participants forms a matrix, and the payment value of the matrix in the invention has two items: the former item represents the user's profit, the latterOne item represents the revenue of a cloud service provider.
According to the above 3 definitions, we can derive two game matrixes between the cloud service provider and the trusted users, and between the cloud service provider and the threat source users, as shown in tables 1 and 2 below:
table 1 payment matrix between cloud service provider and trusted user
Table 2 payment matrix between cloud service provider and untrusted users
Wherein,αin order to realize the false alarm rate of the system,U n the benefit obtained when the user issues a normal service request to be allowed access,C n the overhead of initiating a normal service request for the user,S n the income obtained when the cloud service provider allows the normal service request, and the utility when refusing the normal service request is-S n ，Q n Receives a request overhead for the cloud service provider,q i the probability of not responding is adopted when the cloud service provider receives an abnormal service request,P a sending an abnormal service request for a user is penalized by detection,U a the utility gained when sending out an abnormal service request for an untrusted user, which is not blocked, is also lost revenue for the cloud service provider,C a the cost of sending an unusual service request for an untrusted user,U i for the benefit of the cloud service provider in adopting the non-response strategy for the abnormal service request,βthe system failure rate.
Representation of gaming process
The extended form representation of the game process is as follows: in the game methodology of the present invention, a specific game extension form representation can be generated from an initial state of a user sending a request, and the extension form representation of the present invention represents cloud service providers and trusted users, and cloud service providers and untrusted users in extension forms as shown in fig. 2 and fig. 3.
Belief correction based on cloud service classification
In the fields of artificial intelligence and databases, the belief correction refers to a process of updating original information and beliefs by adopting new information. In the game of incomplete information, a probability inference (used for calculating the probability of the user being untrustworthy in the invention) is carried out on the participators of the incomplete information, the probability of the inference is the belief B (belief), the calculated value is the belief value, and the system gives a corresponding security policy according to the value domain of the belief value.
The invention provides an improved cloud service belief correction method according to different importance degrees of cloud services.
Sensitive service belief correction
The sensitive service belief correction method adopts a refined Bayesian equilibrium correction model when analyzing the user type and receiving a user request, and assumes the nth stage of the game, at the moment, the belief of a cloud service provider to the user is BnIf the probability that the untrusted user sends an abnormal request is p, then:
probability P1 of untrusted user sending normal service requestQ0=(1-α)(1-p)+pβ；
Probability P1 of different service request sent by untrusted userQ1=α(1-p)+p(1-β)；
Probability P0 for trusted user to send normal service requestQ0=(1-α)；
Probability P0 of an abnormal service request sent by a trusted userQ1=α；
According to Bayes' theorem, the formula for improving belief correction is as follows:
(1) when the cloud service provider receives the normal service request, if (B)n>0.5), then:
(1)
otherwise Bn+1=0.5；
(2) When the cloud service provider receives the abnormal service request, the following steps are provided:
(2)
in the formulae (1) and (2)
(3)
The formula (3) is demonstrated in the following formula (6) to formula (11).
Under the model, in the initial stage of the game between the user and the cloud service provider, the prior probability of the cloud service provider to the user is 0.5 (not lower than 0.5), and the trust level can be divided for the belief value: (0.5, 0.75) represents that the user is credible, (0.75, 0.85) represents that the user is more credible but limited in authority, (0.85, 0.95) represents that the user is suspicious, the cloud service provider checks, warns and (0.95, 1) represents that the user is not credible and directly cancels the access and use authority of the user and penalizes.
Non-sensitive service belief correction
Compared with the method for improving the adaptivity of the cloud service provider due to the fact that the safety importance degree and the sensitivity degree are low, the method for correcting the non-sensitive service beliefs is based on a refined Bayesian equilibrium correction model, but the belief value is reduced faster than that of a sensitive service correction method. The method ensures that the user can better use the service provided by the cloud service provider, and improves the income of the cloud service provider. The correction method comprises the following steps: when a normal request is received, the probability that the trusted user sends the normal request is calculated by 1 through the Bayesian rule, and the specific steps are as follows:
(1) when the cloud service provider receives the normal service request, if (B)n>0.5), then:
(4)
otherwise Bn+1=0.5；
(2) When the cloud service provider receives the abnormal service request, the following steps are provided:
(5)
wherein p is the same as formula (3).
Refined Bayesian Nash equilibrium analysis
After the game starts, the beliefs of the player are corrected according to the observed behaviors of other participants, and the strategic selection of the player is made according to the changing beliefs. The existence of the refined Bayesian equilibrium of the model is proved, the rationality and the correctness of the model can be obtained, and the theoretical basis and the formula calculation of the belief correction are provided during the belief correction.
The refining Bayesian equilibrium in the invention needs to meet the following conditions:
the first condition is as follows: the persons in the stations participating in the game act independently of one another.
And a second condition: the strategy selection (combination) of the people in the game in the strategy gathering bureau of the people in the given bureau is optimal to the strategy selection of the people in other bureaus.
And (3) carrying out a third condition: and in each continuous game, the subsequent game reaches the new refined Bayesian balance.
Refined bayesian nash equalization of the gaming stages of the present invention is discussed below.
In the nth stage of game starting, the belief of the cloud service provider to the user is BnAnd (4) showing. In the game, the actions of the two game parties are not influenced mutually, and the strategies adopted by the user and the cloud service provider are obviously independent, so that the condition is met. Setting the probability of sending an abnormal service request by a user as p, and setting the probability of sending a normal service request as 1-p; similarly, when the cloud service provider receives an abnormal service request, the defense strategy is adopted with the probability q: i (not responding to the service request), D (rejecting the service request), and adopting a (accepting the service request) with a probability of 1-q when the cloud service provider receives a normal service request.
When the cloud service provider adopts the service request rejection strategy, the expected function of the profit is as follows:
(6)
the three parts of the expectation are respectively the sum of the effectiveness that the normal service request sent by the false-reported credible user is rejected, the normal service request sent by the false-reported credible user and the abnormal service request sent by the credible user is not missed and rejected.
When the cloud service provider adopts the strategy of not responding to the service request, the expected function of the income is as follows:
(7)
the three parts of the expectation are the sum of the effectiveness of normal service requests sent by non-responding credible users, normal service requests sent by non-responding credible users and abnormal service requests sent by non-responding credible users which are not reported in a missing mode.
When the cloud service provider adopts a service request receiving strategy, the expected function of the income is as follows:
(8)
the three parts of the expectation are respectively the sum of the effectiveness of receiving normal service requests sent by the credible users which are not misreported, receiving normal service requests sent by the credible users which are not misreported and receiving abnormal service requests sent by the credible users which are not misreported.
In conclusion, the mathematical expectation of the cloud service provider is
(9)
The first order partial derivative for q is found for E according to equation (9):
(10)
order to(ii) a The probability formula (3) of the first-order optimization extreme point, which is solved when the user can not be credited to send the abnormal service request
(11)
E1+ E2-E3=0 indicates that the utility obtained by the cloud service provider electing to accept the service request or to take a defense policy is the same at this time.
The same can be obtained by calculating the expectation of the untrusted user and deriving:
probability q formula (4) of defense strategy adopted by cloud service provider
(12)
And refining Bayesian equilibrium is achieved in the game stage, and if the strategy combinations (p, q) of the two parties are changed, the respective effects are reduced, and the condition two is met. The strategy combination (p, q) is the optimal combination, and the model constructs a multi-stage repeated game, so the balance can be achieved in the subsequent game behaviors, namely the condition three is met, and the Bayesian Nash balance exists.
Results and analysis of the experiments
Here, the experimental data of document [1] and the experimental data of 3 times of repeated expansion thereof are used, and then the performance of the algorithm of the present invention is analyzed by comparison of the results with that of document [1 ]. In the user behavior sequence of the experiment, a normal request of the cloud terminal user is represented by 0, and an abnormal request is represented by 1.
The request sequence A1 sent by user A is
[0,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0];
The request sequence B1 sent by user B is
[0,0,0,0,0,1,1,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0]
The belief change situation after the above 90 games is performed by adopting the belief modification model in the document [1], and the experimental result is shown in fig. 4.
It can be seen from fig. 4 that the belief correction method can very effectively discriminate an untrusted user and a trusted user and timely adopt a policy protection system when the game frequency of the algorithm in the document [1] is not large; but after the game times are multiple, when the beliefs of the users rise, the beliefs are difficult to fall again, which is not fair for the users. For example, the user sends 3 consecutive exception requests, which in practice is very low in probability, meaning that the system has the possibility of 3 consecutive false positives.
To further explore the possibility of belief-corrected changes in the literature, we next explore the change in user belief when the user continues to send 60 normal service requests after completing 30 miscellaneous requests, as shown in fig. 5.
The request sequence A2 sent by user A is
[0,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0];
The request sequence B2 sent by user B is
[0,0,0,0,0,1,1,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0];
As can be seen from FIG. 5, after the user A sends a large number of normal requests, the belief change trend is clear, the belief value slowly decreases, the actual effect is relatively met, and the change amount of the belief value is small. However, the present invention considers that the user a and the user B should be judged to be the trusted users even after 60 times or more of normal requests. However, the belief value of the user B is decreased too slowly or even almost unchanged, which is unfair for the user B and seriously damages the service continuity of the cloud service, and thus the continuous denial of service brought by the unfairness can lead to the reduction of the revenue of the service provider. Therefore, on the basis of improving the belief correction method, reasonable reduction of the belief value in a proper range is considered, and in the invention, cloud services are divided into sensitive service belief correction and non-sensitive service belief correction according to importance and sensitivity degree, and finally, an experimental result of the belief correction is given.
Sensitive service belief correction
Aiming at sensitive services in cloud computing, a sensitive service belief correction algorithm is adopted to carry out game on a request sequence A1 and a request sequence B1, and the experimental result is shown in FIG. 6; the game is played for the request sequence a2 and the request sequence B2, and the experimental results are shown in fig. 7.
As can be seen in fig. 6, the user B sometimes sends a normal request, but belongs to a spoofed server behavior, and there are many abnormal requests sent continuously, and the belief value increases, which indicates that the probability of the user whose user attribute is not trusted increases; and the abnormal requests sent by the user A are less, and the system can report by mistake or operate by mistake. As can be seen from fig. 7, after the user a and the user B send a large number of normal requests, the belief values both slowly decrease to the credible belief range, and the punishment time for the user B sending abnormal requests for a large number of times in the previous period is also reasonable.
Non-sensitive service belief correction
Aiming at non-sensitive services in cloud computing, a non-sensitive service belief correction algorithm is adopted to carry out game on a request sequence A1 and a request sequence B1, and the experimental result is shown in FIG. 8; the game is played for the request sequence A2 and the request sequence B2, and the experimental result is shown in FIG. 9.
Comparing fig. 6 and fig. 8, it can be seen that the difference between the non-sensitive service belief correction method and the sensitive service belief correction method is that when the system receives a normal request, the amplitude of the belief correction to the user is reduced faster within a reasonable range, which is beneficial to the service continuity of the user (reducing the punishment factor of the system to the user), and gives better adaptivity to the user to reduce the belief value of the user, and meanwhile, for some untrusted users attempting to attack the service provider system, the system can still punish the untrusted user based on the belief, thereby ensuring the reasonable security and the sufficient profit of the system.
The above are preferred embodiments of the present invention, and all changes made according to the technical scheme of the present invention that produce functional effects do not exceed the scope of the technical scheme of the present invention belong to the protection scope of the present invention.
Claims (1)
1. A user belief correction method based on a dynamic game theory and cloud service classification is characterized by comprising the following steps:
(1) respectively establishing payment matrixes of a cloud service provider and a trusted user and payment matrixes of the cloud service provider and a non-trusted user, and establishing an incomplete information dynamic game model of the cloud service provider and the trusted user and the incomplete information dynamic game model of the cloud service provider and the non-trusted user based on the payment matrixes;
(2) the cloud service is divided into sensitive service and non-sensitive service, and belief correction is respectively carried out to monitor user behaviors;
(3) calculating the current belief value of the user by combining the historical behavior sequence of the user to give out different reaction strategies;
the payment matrixes of the cloud service provider and the trusted user are that the trusted user sends a normal service request, and matrix payment values corresponding to the cloud service provider for receiving the service request, not responding the service request and rejecting the service request are respectively ((1- α) (U)n-Cn),(1-α)(Sn-Qn))、(-αqiCn,-αqi(Sn+Qn))、(-α(1-qi)(Cn+Pa),-α(1-qi)(Sn+Qn))；
The payment matrixes of the cloud service provider and the untrusted user are that the untrusted user sends out an abnormal service request, and matrix payment values corresponding to the cloud service provider accepting the service request, not responding to the service request and rejecting the service request are respectively (β) (U)a-Ca),-β(Ua+Qn))、(-(1-β)qi(Ca),(1-β)qi(Ui-Qn))、(-(1-β)(1-qi)(Ca+Pa),(1-β)(1-qi)(Ua-Qn) The matrix payment values corresponding to the service request, the non-response service request and the service rejection request of the cloud service provider are respectively ((1- α) (U)n-Cn),(1-α)(Sn-Qn))、(-αqiCn,-αqi(Sn+Qn))、(-α(1-qi)(Cn+Pa),-α(1-qi)(Sn+Qn))；
Wherein each matrix payment value (A, B) comprises two items, the former item A represents the income of a user, the latter item represents the income of a cloud service provider, α represents the false alarm rate of the system, UnBenefits obtained when a normal service request is made for a user and access is allowed, CnOverhead of initiating normal service requests for users, SnThe income obtained when the cloud service provider allows the normal service request is obtained, and the utility when the normal service request is rejected is-Sn，QnReceived for cloud service providerOverhead of one request, qiProbability of non-response for cloud service provider when receiving an abnormal service request, PaSending abnormal service requests for users with a detected penalty, UaThe utility gained when sending out an abnormal service request for an untrusted user is not blocked, and is also lost revenue for the cloud service provider, CaCost of sending an abnormal service request for an untrusted user, UiThe yield when the cloud service provider takes the non-response strategy for the abnormal service request is shown as β, which is the system failure rate;
the belief correction method of the sensitive service comprises the following steps: adopting a refined Bayesian equilibrium correction model, and assuming the nth stage of the game, the belief of the cloud service provider to the user is BnThe untrusted user sends an abnormal service request with a probability P, and the trusted user sends a normal service request with a probability 1; for a cloud service provider:
probability P1 of untrusted user sending normal service requestQ0＝(1-α)(1-p)+pβ；
Probability of abnormal service request P1 sent by untrusted userQ1＝α(1-p)+p(1-β)；
When the cloud service provider receives the normal service request, if Bn>0.5，Bn+1Comprises the following steps:
otherwise Bn+10.5; wherein, Bn+1Beliefs of the cloud service provider to the user for the (n + 1) th stage of the game;
when the cloud service provider receives the abnormal service request, Bn+1Comprises the following steps:
wherein p is:
the belief correction method of the non-sensitive service comprises the following steps: based on a refined Bayesian equilibrium correction model, supposing that the nth stage of the game is that the belief of the cloud service provider to the user is BnThe untrusted user sends an abnormal service request with a probability P, and the trusted user sends a normal service request with a probability 1; for a cloud service provider:
probability P1 of untrusted user sending normal service requestQ0＝(1-α)(1-p)+pβ；
Probability of abnormal service request P1 sent by untrusted userQ1＝α(1-p)+p(1-β)；
When the cloud service provider receives the normal service request, if Bn>0.5, then Bn+1Comprises the following steps:
otherwise Bn+1＝0.5；
When the cloud service provider receives the abnormal service request, Bn+1Comprises the following steps:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510066543.8A CN104639638B (en) | 2015-02-10 | 2015-02-10 | Based on user's updating belief method that dynamic game opinion is classified with cloud service |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510066543.8A CN104639638B (en) | 2015-02-10 | 2015-02-10 | Based on user's updating belief method that dynamic game opinion is classified with cloud service |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104639638A CN104639638A (en) | 2015-05-20 |
| CN104639638B true CN104639638B (en) | 2016-03-09 |
Family
ID=53217939
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510066543.8A Active CN104639638B (en) | 2015-02-10 | 2015-02-10 | Based on user's updating belief method that dynamic game opinion is classified with cloud service |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104639638B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106100893A (en) * | 2016-07-05 | 2016-11-09 | 合肥工业大学 | Service opening and the security control appraisal procedure of cloud service environment based on dynamic game |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108295475B (en) * | 2017-07-18 | 2021-09-28 | 腾讯科技（深圳）有限公司 | Data processing method and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102231743A (en) * | 2011-06-30 | 2011-11-02 | 江苏南大苏富特科技股份有限公司 | Attack-graph-based intrusion response mode |
| CN102281549A (en) * | 2011-08-22 | 2011-12-14 | 北京航空航天大学 | Mobile Ad-hoc NETwork (MANET) resource admission control method based on game theory |
| CN102333307A (en) * | 2011-09-28 | 2012-01-25 | 北京航空航天大学 | Wireless sensor network (WSN) trust evaluation method based on subjective belief |
| CN104079576A (en) * | 2014-07-02 | 2014-10-01 | 天津大学 | Dynamic cooperation alliance structure forming method based on Bayes alliance game |
-
2015
- 2015-02-10 CN CN201510066543.8A patent/CN104639638B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102231743A (en) * | 2011-06-30 | 2011-11-02 | 江苏南大苏富特科技股份有限公司 | Attack-graph-based intrusion response mode |
| CN102281549A (en) * | 2011-08-22 | 2011-12-14 | 北京航空航天大学 | Mobile Ad-hoc NETwork (MANET) resource admission control method based on game theory |
| CN102333307A (en) * | 2011-09-28 | 2012-01-25 | 北京航空航天大学 | Wireless sensor network (WSN) trust evaluation method based on subjective belief |
| CN104079576A (en) * | 2014-07-02 | 2014-10-01 | 天津大学 | Dynamic cooperation alliance structure forming method based on Bayes alliance game |
Non-Patent Citations (1)
| Title |
|---|
| 云计算环境下基于动态博弈论的用户行为模型与分析;陈亚睿 等;《电子学报》;20110831;第39卷(第8期);第2-6部分 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106100893A (en) * | 2016-07-05 | 2016-11-09 | 合肥工业大学 | Service opening and the security control appraisal procedure of cloud service environment based on dynamic game |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104639638A (en) | 2015-05-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kumar et al. | The Impact of Adversarial Attacks on Federated Learning: A Survey | |
| AU2019210493A1 (en) | Anomaly detection to identify coordinated group attacks in computer networks | |
| RU2510982C2 (en) | User evaluation system and method for message filtering | |
| Nicolas et al. | Blockchain system defensive overview for double-spend and selfish mining attacks: A systematic approach | |
| CN102510337A (en) | Quantitative risk and income self-adaptive dynamic multiple-factor authentication method | |
| CN101425926A (en) | Network order regulating method based on credit evaluation | |
| Herrick | The social side of ‘cyber power’? Social media and cyber operations | |
| CN115907029A (en) | Defense method and system for federal learning virus attack | |
| Jiang et al. | Sybil attacks and defense on differential privacy based federated learning | |
| Mishra et al. | Detection of DDoS vulnerability in cloud computing using the perplexed bayes classifier | |
| CN104639638B (en) | Based on user's updating belief method that dynamic game opinion is classified with cloud service | |
| Belouch et al. | Comparison of ensemble learning methods applied to network intrusion detection | |
| CN102231743A (en) | Attack-graph-based intrusion response mode | |
| CN115296830B (en) | Network collaborative attack modeling and hazard quantitative analysis method based on game theory | |
| Yang et al. | DeMAC: Towards detecting model poisoning attacks in federated learning system | |
| Zhou et al. | Detecting suspicious transactions in a virtual-currency-enabled online social network | |
| Yuan et al. | A Modified Hybrid Method Based on PSO, GA, and K‐Means for Network Anomaly Detection | |
| Chen et al. | Dynamic threshold strategy optimization for security protection in Internet of Things: An adversarial deep learning‐based game‐theoretical approach | |
| Guan et al. | A Bayesian Improved Defense Model for Deceptive Attack in Honeypot-Enabled Networks | |
| Nandal et al. | Cyber security against ddos malware spoofing attacks using machine learning with genetic algorithm | |
| CN113315763A (en) | Network security defense method based on heterogeneous group evolution game | |
| CN106100893B (en) | The service opening and security control appraisal procedure of cloud service environment based on dynamic game | |
| CN118432950B (en) | Big data-based computer network security defense system | |
| Xiao et al. | FedRMA: A Robust Federated Learning Resistant to Multiple Poisoning Attacks | |
| Zhang et al. | A network security situation analysis framework based on information fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |