A virus defense optimization strategy method based on hybrid strategy
By establishing a multi-party game model in virus defense, considering the target correlation relationship, and using a hybrid strategy to calculate the optimal defense strategy, the problems of rough strategy calculation and fixed investment in the existing technology are solved, and the defense effect is maximized under limited costs and strategy optimization.
Patent Information
- Application Number
- CN202410941030.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-15
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-07-15
AI Technical Summary
The existing technology fails to fully consider the association between goals in virus defense strategies, resulting in rough policy calculations and lack of room for tuning, and the use of pure strategies leads to fixed defense investment and failure to achieve optimization.
Based on hybrid strategies, a multi-party game model is established for single defenders and multi-defenders. The optimal defense strategy is calculated through the initial defense strategy matrix, defense investment matrix, defense utility matrix and loss coefficient matrix, and the correlation relationship between the targets is considered. Indirect influence, the optimal solution is obtained by using mixed integer linear programming.
A model that maximizes defense effectiveness at limited costs is realized, ensuring the overall maximum benefit of the defense party, and continuously tuning the strategy through iterative operations to improve the defense effectiveness.
Smart Images

Figure CN118473836B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of server management, and in particular to a virus defense optimization strategy method based on a hybrid strategy. Background Art
[0002] In the process of information development, the outbreak of various types of computer viruses has posed a major threat to today's network security; once a device in the network is infected with a virus, it will continue to spread and attack by taking advantage of the interrelationships between network nodes in a short period of time, causing more serious losses. Security investment in target nodes can improve defense effectiveness and reduce the risk of virus transmission. Deploying a reasonable defense strategy for target nodes has become an urgent problem to be solved.
[0003] In order to better solve the game equilibrium problem of individual decision-making, game theory has gradually been introduced into information security investment decision-making. However, in the existing technology, when building models and calculating strategies, objects are independent and the mutual influence of related objects is not considered; the common strategy allocation method is uniform distribution or normal distribution, and there is almost no room for strategy regulation.
[0004] In addition, pure strategies require participants to choose only one specific action under each given information situation; this pure strategy simply means that for a certain target, the defender either invests in defense or does not defend; rather than an adjustable strategy with a certain degree of severity. When calculating strategies, the existing technology uses a fixed investment in defense according to the weight of the target; the decision-making method is simple and crude and conforms to common sense, but there is no room for further optimization and adjustment; and the strategy operation is generally one-time, without iterative operations to continuously adjust the strategy, resulting in a rough strategy.
[0005] Therefore, a virus defense optimization strategy method based on a hybrid strategy is needed to solve one or more of the above technical problems. Summary of the invention
[0006] In order to solve one or more of the above technical problems, the present invention provides a virus defense optimization strategy method based on a hybrid strategy, which establishes a multi-party game model common to single defenders and multiple defenders based on the hybrid strategy, and generates a virus defense optimization strategy through the following steps:
[0007] Step 1: Create an initial defense strategy matrix for all defenders ,in, Indicates the initial defense strategy of each target;
[0008] Step 2: Based on the initial strategy matrix For the goal Allocate defense resources and obtain the initial defense investment matrix ;in, , ; Indicates the target Initial allocation of defense resources, Indicates the target Initial defense strategy;
[0009] Step 3: Based on the Initial Defensive Investment Matrix Calculate the initial defense utility matrix ;in, represents the initial defense utility of each target;
[0010] Step 4: Obtain the initial attack utility matrix and maximum attack utility , and set the loss coefficient matrix according to the initial optimal attack target ;
[0011] Step 5: Matrix your initial defensive investments , Initial Defense Utility Matrix U * And the loss coefficient matrix Bring in and calculate the corresponding protection strategy for each target , and obtain the defense strategy matrix ;in,
[0012] ; ;
[0013] Step 6: Go through the Defense Strategy Matrix Update the initial policy matrix , and the new defense utility matrix is obtained by calculation ; and output the defense strategy matrix and the defense utility matrix .
[0014] As a further solution, the multi-party game model includes one or more defenders , a single attacker and several targets ; where each defender Responsible for protecting a group of targets , and there is ( )as well as .
[0015] As a further solution, the defense strategy matrix It is obtained by mixing the strategy matrices of all defenders, where each row corresponds to the strategy distribution of a defender, using probability Indicates defender To target defense strategy; among them, and .
[0016] As a further solution, the defense utility matrix Indirect effects are also considered, which are determined by the degree of association. To transmit; among them, Indicates the current target Related Target The degree of correlation between them.
[0017] As a further solution, the defender in the multi-party game model To target Defensive investment through Calculate the total defense cost by consuming all defense resources. , Defender The target to be protected requires resources to be consumed , , and the total defense cost Also meets:
[0018] ;
[0019] in, C max Indicates the maximum acceptable consumption of resources. v i,j Indicates the current target To the defender target value.
[0020] As a further solution, the defense utility matrix Through loss coefficient To reflect the attacker's The effectiveness of the attack; where the loss coefficient The target of attack The actual extent of the loss.
[0021] As a further solution, in the defense utility matrix In the target To the defender The utility generated is recorded as , and is calculated by the following formula:
[0022] ;
[0023] in, Indicates the current target own value, Indicates the associated target own value, Indicates the defender's The consumption of resources caused by defensive investments, Indicates the defender's The consumption of resources caused by defensive investments.
[0024] Compared with the related art, the virus defense optimization strategy method based on the hybrid strategy provided by the present invention has the following beneficial effects:
[0025] The present invention studies the virus defense method based on the hybrid strategy, establishes a model for maximizing defense utility under limited cost, transforms the network model with attackers and defenders into the problem of finding the optimal solution in the optimization model, and obtains the optimal solution through a mixed integer linear programming formula, which is the optimal defense strategy of the defender under limited cost. The model fully considers the correlation between the targets, and adopts a hybrid strategy to represent and calculate the optimal defense strategy of each defender, ensuring the maximum benefit of the defender as a whole. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 A schematic diagram of the steps of a virus defense optimization strategy method based on a hybrid strategy provided by the present invention;
[0027] Figure 2 A schematic diagram of the defense strategy provided by the present invention;
[0028] Figure 3 A schematic diagram of parameter values provided by the present invention;
[0029] Figure 4 A schematic diagram of the target value, association relationship and initial strategy distribution provided by the present invention;
[0030] Figure 5 This is a schematic diagram of the total defense utility growth rate provided by the present invention. DETAILED DESCRIPTION
[0031] The present invention will be further described below in conjunction with the accompanying drawings and implementation modes.
[0032] like Figure 1 As shown in FIG. 1 , a virus defense optimization strategy method based on a hybrid strategy is provided. A multi-party game model common to single defenders and multiple defenders is established based on the hybrid strategy, and a virus defense optimization strategy is generated through the following steps:
[0033] Step 1: Create an initial defense strategy matrix for all defenders ,in, Indicates the initial defense strategy of each target;
[0034] For step 1, it is generally established based on the distribution characteristics of the defense object. If it is not clear, it can be determined by uniform distribution or normal distribution by default. For example, if the initial defense strategy conforms to the uniform distribution, the size of the initial defense strategy is ,in k is the total number of targets.
[0035] Step 2: Based on the initial strategy matrix For the goal Allocate defense resources and obtain the initial defense investment matrix ;in, , ; Indicates the target Initial allocation of defense resources, Indicates the target Initial defense strategy;
[0036] Step 3: Based on the Initial Defensive Investment Matrix Calculate the initial defense utility matrix ;in, represents the initial defense utility of each target;
[0037] Step 4: Obtain the initial attack utility matrix and maximum attack utility , and set the loss coefficient matrix according to the initial optimal attack target ;
[0038] For step 4, the size of the attack utility determines the size of the defense utility, so the attack utility matrix and the defense utility matrix [] can be understood as the same matrix, and the corresponding elements of the two are equal in value. Among them, the loss coefficient matrix []The normalized attack utility matrix Get; the greater the target's attack utility, the greater the loss coefficient.
[0039] Step 5: Matrix your initial defensive investments , Initial defense utility matrix U * And the loss coefficient matrix Bring in and calculate the corresponding protection strategy for each target , and obtain the defense strategy matrix ;in,
[0040] ; ;
[0041] Step 6: Go through the Defense Strategy Matrix Update the initial policy matrix , and the new defense utility matrix is obtained by calculation ; and output the defense strategy matrix and the defense utility matrix .
[0042] It should be noted that the uniqueness of this embodiment lies in the study of virus defense methods based on a hybrid strategy, and the establishment of a model for maximizing defense utility under limited costs. The model fully considers the correlation between targets and uses a hybrid strategy to represent and calculate the optimal defense strategy for each defender, thereby ensuring the maximum benefit of the defender as a whole.
[0043] Before we start to study the problem of virus defense strategy, we need to build a model for the participants in this game to describe their basic information and physical characteristics. This example proposes a network model based on a hybrid strategy that is common to single defenders and multiple defenders:
[0044] There are one or more defenders in the network , a single attacker, and a set of targets (nodes) . Every defender Responsible for protecting a group of targets .
[0045] Now assume ( )as well as (That is, the number of objects protected by each defender , and there is no duplication of protection targets between them).
[0046] The Defense Strategy Matrix It is obtained by mixing the strategy matrices of all defenders, where each row corresponds to the strategy distribution of a defender, using probability Indicates defender To target defense strategy; among them, and On the other hand, assuming that the attacker has limited resources and can only attack one target at a time, assuming that the attacker's target is .
[0047]
[0048] As a further solution, the multi-party game model includes one or more defenders , a single attacker and several targets ; where each defender Responsible for protecting a group of targets , and there is ( )as well as .
[0049] It should be noted that: from this step, the optimization algorithm will be iterated. The stopping condition of the iteration can be a finite number of iterations, or a sufficiently small difference between the defense utility after two iterations.
[0050] In a specific embodiment, Figure 2 As shown, its corresponding defense strategy for:
[0051]
[0052] In the strategy matrix, the mth row represents the mth defender, and the nth column represents the defender protecting the nth target. Figure 2 The corresponding strategy matrix shows that the first defender is only responsible for protecting targets 3 and 7, and the defense strategies are 0.5 and 0.5 respectively; similarly, the second row represents that the second defender protects targets 1, 4, and 8, and the defense strategies are 0.333333, 0.333333, and 0.333333 respectively; the third row represents that the third defender protects targets 2, 5, and 6, and the defense strategies are 0.333333, 0.333333, and 0.333333 respectively.
[0053] Relevance :The existence of association affects the effectiveness of both attack and defense. The virus attack starts with an initial target. After the attacker causes direct damage to the target, it will also cause indirect damage to other targets due to the association. The defender allocates certain defense resources to the target, improves its defense effectiveness, and indirectly increases the defense effectiveness of other targets associated with the target. This indirect impact is determined by the degree of association. To deliver, Indicates the current target Related Target The degree of correlation between them.
[0054] Defensive Investment : The defender's defensive investment in the target is calculated by Calculate. The total defense resources consumed by all defenders is ; Among them, the defender The target to be protected requires resources to be consumed , ; and, the total defense cost It should also meet: ; This is because if the target Defense investment costs Target value If the target is still large, then the defender's defense strategy for that target is meaningless.
[0055] Loss coefficient : Once the attack occurs, use the parameter To reflect the attacker's The effectiveness of the attack, so Target When there is no attack in the system and the target of attack is not hour, The value of is 1; otherwise The value of will be defined according to the actual damage level of the attacked target ( A value of 0 indicates that the target is completely lost after being attacked).
[0056] Utility: Defender vs. Target Investing in defense consumes resources ( ). If, now the attacker has Launch an attack, and the defender's protection resource allocation for the target is also at this position (attack and defense), then in this game, the target To the defender The utility generated is recorded as , the utility of the attack is recorded as We use attack (defense) utility to describe the capabilities of both attackers and defenders, and stipulate that the utility depends only on the attack (defense) strategy towards the target.
[0057] Assuming that the defense utility depends only on the attacked target and the defender's security strategy for the target (defense cost), the attacker can only choose one target with the largest attack utility to attack each time. The value of itself is recorded as , Defender To target Protecting consumes defensive investment ,Target If destroyed or infected, it will attack other targets The probability of infection or the magnitude of the impact is determined by the degree of association. "OK. If the attacker now Launch an attack, then the utility value of the defender in this game is recorded as The defensive utility is expressed as (1.1):
[0058] (1.1)
[0059] like Figure 3As shown in the figure, there are two defenders, each protecting a target. The value of each target is 1, the correlation between them is 0.5, and the defense investment is 0.5. Assuming that the attacker chooses to attack the first target, the loss coefficient of the attacked target is 0.5. From formula (1.1), we can get that the defense utility of target 1 and target 2 is 1.75 and 2.25 respectively.
[0060] In this attack and defense model, the goal The relationship between the attack utility and the defense utility generated satisfies:
[0061] (1.2)
[0062] The attacker always chooses a target that maximizes the attack effect; therefore, the maximum attack utility is obtained by equation (1.3), and the optimal attack target is obtained at the same time. :
[0063] (1.3)
[0064] In a specific algorithm, the algorithm simulates the game between the virus and a single defender; the defender only needs to consider the target to be protected and the association relationship within the target group:
[0065] Since there is one defender protecting the target matrix T, the available defense investment for the defender is , giving an initial defense strategy matrix that satisfies uniform distribution , the initial defense strategy size is ; and according to the initial strategy matrix , from (1.4) to the target Allocate defense resources:
[0066] (1.4)
[0067] This gives us an initial defense investment matrix .
[0068] Get the initial defense investment matrix for all targets Then, the initial defense utility matrix is calculated by formula (1.1) ; According to equations (1.2) and (1.3), the initial attack utility matrix is calculated , Maximum attack utility , initial best attack target ( may not be unique); based on the target most likely to be attacked by the attacker , set the loss coefficient matrix .
[0069] Initial Defense Investment Matrix , and the initial defense utility matrix , loss coefficient matrix Substitute into (1.5) and (1.6) to calculate the corresponding new protection strategy for each target :
[0070] (1.5)
[0071] (1.6)
[0072] Get a new defense strategy matrix Then, substitute into (1.4) to get the new defensive investment matrix , and the new defense utility matrix is calculated by formula (1.1) , and the result , Output.
[0073] To intuitively illustrate how the above algorithm implements defense strategy deployment in a network environment, we give the following example: Now assume that there are 8 target computers (or computer groups) in a company, and we define the data value in each target computer as the target value , the correlation is Indicates that the initial defense strategy is Representation; According to the network relationship between computers, the following topology is simplified Figure 4 , and the degree of connection between any two computers is expressed as In order to prevent viruses from attacking the company's internal equipment, the company invested a total of , used to cultivate employee security skills, upgrade system patches, update firewall configurations, strictly manage ports, and back up important data in multiple locations. Now it is required to limit the total defense cost. Next, an algorithm is used to give a reasonable allocation strategy to maximize the utility of defense.
[0074] For the convenience of calculation, we take the value of each target The value is 1, which means the correlation between targets Unified as 0.5, given the total defense cost .
[0075] The initial defense strategy matrix under a single defender:
[0076]
[0077] Total defense investment available to defenders , substituting into (1.4) we can get the initial defense investment matrix:
[0078]
[0079] The defense utility matrix of all targets obtained according to formula (1.1)
[0080]
[0081] After converting equations (1.2) and (1.3) into attack utility, we get: , at this time, the attacker already knows that in the current defense strategy, attacking the fourth computer will gain the greatest attack utility; according to the attacker's most likely target is the fourth computer, the loss coefficient matrix is set:
[0082]
[0083] Substituting the above known conditions into equations (1.5) and (1.6), we can obtain the iterative defense strategy matrix:
[0084]
[0085] After obtaining the defense strategy matrix, we To update the defensive investment matrix:
[0086]
[0087] Finally, the defense utility matrix is updated according to formula (1.1):
[0088]
[0089] To verify the effectiveness of this solution, given the same total defense investment C, we compared the calculation results of this solution with the results obtained without considering the association relationship. When the defender makes a decision without considering the association relationship between the targets, he often makes a decision based on the value of the target. To arrange the corresponding defensive investment; its basic relationship satisfies:
[0090] (1.7)
[0091] According to formula (1.7), we can obtain the matrix of defensive investment c without considering the correlation relationship:
[0092] (1.8)
[0093] The defender does not consider the correlation when making decisions, but when calculating the defense utility, it is necessary to calculate it in actual situations (considering the correlation), so we substitute the defense investment c into (1.1) to obtain the defense utility matrix .
[0094] Next, we compare the results of whether the defender considers the correlation relationship under different investments. The results are shown in Table 1.1:
[0095] Table 1.1 Comparison of total defense utility with and without considering correlation under different defense investments
[0096]
[0097] Combined with Table 1.1 and Figure 5 It is not difficult to find that decisions that take correlation into account often produce higher defense utility than decisions that do not take correlation into account, and as the defense investment increases (provided that the total defense investment does not exceed the total target value), the growth rate of defense utility also increases; this shows that the more the defender values the security of the target and the more defense investment is given, the more the impact of correlation should be considered. This algorithm can help a single defender fully consider the impact of correlation when defending against virus attacks, arrange a reasonable allocation method for it, and ensure that the defense utility is better than the utility without considering correlation.
[0098] The above are only embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A virus defense optimization strategy method based on a hybrid strategy, characterized in that: Based on the hybrid strategy, a multi-party game model common to single defenders and multiple defenders is established, and the optimal virus defense strategy is generated through the following steps: Step 1: Create an initial defense strategy matrix for all defenders ,in, Indicates the initial defense strategy of each target; Step 2: Based on the initial strategy matrix For the goal Allocate defense resources and obtain the initial defense investment matrix ;in, , ; Indicates the target Initial allocation of defense resources, Indicates the target Initial defense strategy, C is the total defense cost; Step 3: Based on the Initial Defensive Investment Matrix Calculate the initial defense utility matrix ;in, represents the initial defense utility of each target; Step 4: Obtain the initial attack utility matrix and maximum attack utility , and set the loss coefficient matrix according to the initial optimal attack target ; Step 5: Matrix your initial defensive investments , Initial defense utility matrix U * And the loss coefficient matrix Bring in and calculate the corresponding protection strategy for each target , and obtain the defense strategy matrix ;in, ; ; Step 6: Go through the Defense Strategy Matrix Update the initial policy matrix , and the new defense utility matrix is obtained by calculation ; and output the defense strategy matrix and the defense utility matrix ; Among them, the defense utility matrix Indirect effects are also considered, which are determined by the degree of association. To transmit; among them, Indicates the current target Related Target The degree of correlation between The Defense Utility Matrix Through loss coefficient To reflect the attacker's The effectiveness of the attack; where the loss coefficient The target of attack The actual extent of the loss is related to In the defense utility matrix In the target To the defender The utility generated is recorded as , and is calculated by the following formula: ; in, Indicates the current target own value, Indicates the associated target own value, Indicates the defender's The consumption of resources caused by defensive investments, Indicates the defender's The consumption of resources caused by defensive investments.
2. The virus defense optimization strategy method based on a hybrid strategy according to claim 1, characterized in that: The multi-party game model includes one or more defenders , a single attacker and several targets ; where each defender Responsible for protecting a group of targets , and there is ( )as well as .
3. The virus defense optimization strategy method based on a hybrid strategy according to claim 2 is characterized in that: The Defense Strategy Matrix It is obtained by mixing the strategy matrices of all defenders, where each row corresponds to the strategy distribution of a defender, using probability Indicates defender To target defense strategy; among them, and .
4. The virus defense optimization strategy method based on a hybrid strategy according to claim 3 is characterized in that: In the multi-party game model, the defender To target Defensive investment through Calculate the total defense cost by consuming all defense resources. , Defender The target to be protected requires resources to be consumed , , and the total defense cost Also meets: ; in, C max Indicates the maximum acceptable consumption of resources. v i,j Indicates the current target To the defender target value.
Citation Information
Patent Citations
Network security efficiency evaluation method based on attack and defense association matrix
CN109547242A
Security assessment method and device based on fuzzy matrix and Nash equilibrium, and medium
CN113935039A
Network security policy AI autonomous defense method and system based on security ontology modeling
CN114401113A
Active defense method, device and system for multiple attackers
CN116684152A
Destination addressing for transactions associated with a distributed ledger
US20230036694A1
