Decision competition method of homogeneous detection engine under software security detection system
By introducing a homogeneous detection engine decision competition method in the software security detection system, and using the decision competition algorithm to select the optimal detection engine, the problem of difficulty in maintaining advancedness and risk in the upgrade process in the existing system during the use cycle is solved, and the problem of the difficulty in maintaining advancedness and upgrading is achieved. Automatic optimization and stable upgrade of the system.
Patent Information
- Application Number
- CN202410808192.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-21
- Publication Date
- 2024-11-22
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
It is difficult for existing software security detection systems to maintain advancedness during their use cycles. There are risks of business cessation, bug instability and subjectivity during the upgrade process, and there is a lack of a competitive environment, resulting in frequent modification and reconstruction of the system, which damages its automation advantages.
Introduce a decision-making competition method of homogeneous detection engines, select the optimal engine among multiple detection engines through the capability detection base module, and use the decision-making competition algorithm to reassign the call probability based on execution performance data, ensuring that the system always uses the most advantageous detection engine and allow new engines to join the competition.
By introducing a competition mechanism, the software security detection system can continue to maintain a local optimal structure during long-term use, avoiding frequent modifications and reconstructions, ensuring stable system performance, and realizing automated upgrades and optimizations within the system.
Smart Images

Figure SMS_1 
Figure SMS_3 
Figure FDA0004905410680000031
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of software security detection, and in particular to a decision competition method of a homogeneous detection engine in a software security detection system. Background Art
[0002] The software security testing system integrates a variety of security testing engines, including white box testing, black box testing, performance testing, etc., which are mainly used for software security evaluation. The current software security testing system has an obvious drawback, that is, once the system is developed and tested and launched, its form and capabilities are fixed, and it is difficult to maintain its advanced nature in a cycle of 3-5 years. During the system's life cycle, due to the degradation of the performance of some functions or some functions are no longer suitable for the development of the organization, it has to face constant modification, and even needs to be overturned and reconstructed. The automation advantage of the system is constantly modified and overturned and reconstructed and flattened.
[0003] The upgrade of traditional software security detection systems is a single, subjective, and controlled mode. Singularity refers to the single upgrade mode. When the system is facing an upgrade, it can only deploy the upgrade software to the system by stopping some or all software services or functions, and then start the software. The risks faced by this method include a period of business suspension, the instability of bugs in the new version of the software, and the continuous artificial "downtime-upgrade" cycle after modification. Subjectivity means that the dimension of the upgrade function is subjective, and the set pros and cons judgment criteria are generally also subjective (including whether the function meets business requirements, the estimated number of users that may increase, etc.), which leads to a significant increase in risk openings (due to the subjectivity of a large number of psychological biases and noise, it is difficult to find the core of the problem for many product planning needs). Control means that there is no competitive environment for the system's software modules. Summary of the invention
[0004] In view of the deficiencies in the prior art, the present invention aims to provide a decision competition method for homogeneous detection engines in a software security detection system.
[0005] In order to achieve the above object, the present invention adopts the following technical solution:
[0006] A decision competition method for homogeneous detection engines in a software security detection system, the specific process is as follows:
[0007] S1. After the upper layer module (user end) triggers a detection engine call, the capability detection base module selects a detection engine from the competing detection engines according to the calling algorithm and uses the detection engine to perform the software security detection this time. The specific process of calling the algorithm is as follows:
[0008] In the initial state, P i(first) represents the calling probability of each detection engine at the initial moment of the software security detection system, P i (first) = 1 / n, i∈n, where n is the number of detection engines participating in the competition, i, n∈N + (N + is a positive integer); at the moment of entering the first call cycle, the call probability of each detection engine is P i (second), P i (second) and P i (first) equal;
[0009] Execution Space Sp i is the range in which the detection engine i is selected, which is expressed as a data interval, Sp i The larger the interval of , the higher the probability that detection engine i is selected;
[0010] Execution Space SP i =[1,P i *100], probability of obtaining total space SpALL = SP i ∪[Random((1-P i )*100)]; Set the random value function V(p) = Random(SpALL,1); Random(x) represents the random number function, Random(SpALL,1) means randomly selecting a number in SpALL; P i is the calling probability i of the detection engine; the execution space here is represented as a set of closed intervals of data. When the data selected by the random value function hits the value of the execution space corresponding to a detection engine, it means that it is selected and will be called for execution; the execution space of each detection engine is independent;
[0011] S2. The monitoring module actively obtains the execution performance data of the detection engine for this software security detection and submits it to the capability detection base module;
[0012] The execution performance data of the detection engine includes running time and subjective running feedback; setting ef i is the execution performance of the detection engine i, t i is the running time of detection engine i, v i is the running rate of detection engine i, fb i is the subjective running feedback of detection engine i, that is, the user's evaluation value of detection engine i after it runs; before entering the first call cycle, all execution performance data of each detection engine are zero;
[0013] S3. After the software security test is completed, the capability detection base module analyzes the detection engine advantage according to the latest execution performance data of each detection engine, and redistributes the call probability of each detection engine participating in the competition through the decision competition algorithm. At this point, a call cycle ends; after multiple call cycles, the call probability of each detection engine will converge to a fixed range. Due to the accumulation of selection advantages, a certain detection engine will be continuously executed. At this time, the capability detection base module completes the selection process of the advantageous detection engine, and the software security detection system will always call the detection engine with the greater advantage;
[0014] The specific process of the decision competition algorithm is:
[0015] After the called detection engine completes a software security test, the capability detection base module receives the latest execution performance data of the detection engine collected by the monitoring module, and compares the execution performance data of each competing detection engine, assigning different weights according to the quality of the execution performance data of each detection engine, and then weighting the calling probability of each detection engine according to the assigned weight to obtain the new calling probability of each detection engine.
[0016] Furthermore, the capability detection base module allows new detection engines to be added.
[0017] Furthermore, the detection engines participating in the competition are homogeneous detection engines, which are called ecological niches. The following process is used to determine whether the detection engines are homogeneous:
[0018] (1) Assume that a detection engine class has a feature set of ∧(λ1, λ2, λ3…λn); Assume that two detection engines A and B have feature sets A*(A1, A2, A3…An) and B*(B1, B2, B3…Bn) respectively;
[0019] (2) Let S(A*, ∧) represent the matching degree of detection engine A for class ∧, with cardinality n;
[0020] (3) Assume that the matching function between the set A* and ∧ is as follows:
[0021]
[0022] A*∩∧ represents the same features of the sets A* and ∧, A*-∧ represents the features that exist in the A* set but not in the ∧ set, and ∧-A* represents the features that exist in the ∧ set but not in the A* set;
[0023] (4) When S(A*, ∧)>50%, it is determined that the set A belongs to the members of the ∧ class;
[0024] (5) Similarly, when S(B*, ∧)>50%, it is determined that the B* set belongs to the members of the ∧ class, and A and B belong to the same detection engine.
[0025] The present invention also provides a software security detection system for implementing the above method, comprising an upper layer module, a capability detection base module and a monitoring module;
[0026] The upper layer module is used to initiate the call of the detection engine;
[0027] The capability detection base module is used to select a detection engine to be called according to the calling algorithm after the upper-level module initiates the calling of the detection engine, and use the detection engine to perform the current software security detection; and in a calling cycle, after the called detection engine completes the software security detection, it receives the latest execution performance data of the detection engine collected by the monitoring module, performs detection engine advantage analysis according to each detection engine, and redistributes the calling probability of each detection engine participating in the competition through the decision competition algorithm;
[0028] The monitoring module is used to actively obtain the performance data of the called detection engine during the software security detection process and pass it to the capability detection base module.
[0029] The beneficial effect of the present invention is that the present invention introduces the concept of equal competition position into the architecture of the software security detection system, that is, when the system needs to call a certain capability, there are multiple homogeneous detection engines in the system that can realize the capability, and multiple detection engines compete on the equal competition position. Excellent functions are continuously participating in competition, adjusting weights, and being selected, and are finally selected after multiple cycles of competition within the system, and functions with poor performance are gradually eliminated. Therefore, although the software security detection system has been used for a long time, due to the internal competition mechanism, it does not need to improve its capabilities through patchwork, and can continuously maintain the local optimal accumulation, thereby ensuring the optimal structure of the system. DETAILED DESCRIPTION
[0030] The present invention will be further described below. It should be noted that this embodiment is based on the technical solution and provides a detailed implementation method and a specific operation process, but the protection scope of the present invention is not limited to this embodiment.
[0031] Example 1
[0032] This embodiment provides a decision competition method for homogeneous detection engines in a software security detection system, and the specific process is as follows:
[0033] S1. After the upper layer module (user end) triggers a detection engine call, the capability detection base module selects a detection engine from the competing detection engines according to the calling algorithm and uses the detection engine to perform the software security detection this time. The specific process of calling the algorithm is as follows:
[0034] In the initial state, P i (first) represents the calling probability of each detection engine at the initial moment of the software security detection system, P i (first) = 1 / n, i∈n, where n is the number of detection engines participating in the competition, i, n∈N + (N + is a positive integer); at the moment of entering the first call cycle, the call probability of each detection engine is P i (second), P i (second) and P i (first) equal;
[0035] Execution Space Sp i is the range in which the detection engine i is selected, which is expressed as a data interval, Sp i The larger the interval of , the higher the probability that detection engine i is selected;
[0036] Execution Space SP i =[1,P i *100], probability of obtaining total space SpALL = SP i ∪ Set the random value function V(p) = Random(SpALL,1); Random here refers to the random function in the computer, the first parameter is the value range, the second parameter is the value space, Random(SpALL,1) means randomly selecting a number in SpALL; P i is the calling probability i of the detection engine. The execution space here is represented as a set of closed intervals of data. When the data selected by the random value function hits the value of the execution space corresponding to a detection engine, it means that it is selected and will be called for execution. For example, the initial probability of A is 50%, so its execution space is [1,50], and the total probability acquisition space is [1,100]; when the calling probability of A increases to 55%, its execution space is [1,55], and the total probability acquisition space is [1,2,...,55,61,68,...] (Note: after 55, the subsequent obfuscation values, the space is 45 numbers, the numbers are random), and so on. The execution space of each detection engine is independent.
[0037] In the first few calling cycles, the probability of incremental execution of the detection engine is not high, so each detection engine is likely to be called at least once.
[0038] S2. The monitoring module actively obtains the execution performance data of the detection engine for this software security detection and submits it to the capability detection base module;
[0039] The execution performance data of the detection engine includes running time and subjective running feedback; setting ef i is the execution performance of the detection engine i, t i is the running time of detection engine i, v i is the running rate of detection engine i, fb i is the subjective operation feedback of detection engine i (i.e., the user's evaluation value of detection engine i after it runs); before entering the first call cycle, all execution performance data of each detection engine are zero.
[0040] S3. After the software security test is completed, the capability detection base module will analyze the detection engine advantages according to the latest execution performance data of each detection engine, and redistribute the calling probability of each detection engine participating in the competition through the decision competition algorithm. At this point, a calling cycle ends. After multiple calling cycles, the calling probability of each detection engine will converge to a fixed range. Due to the accumulation of selection advantages, a certain detection engine will be continuously executed. At this time, the capability detection base module completes the selection process of the advantageous detection engine, and the software security detection system will always call the detection engine with more advantages. The specific process of the decision competition algorithm is as follows:
[0041] After the called detection engine completes a software security test, the capability detection base module receives the latest execution performance data of the detection engine collected by the monitoring module, and compares the execution performance data of each competing detection engine, assigning different weights according to the quality of the execution performance data of each detection engine, and then weighting the calling probability of each detection engine according to the assigned weight to obtain the new calling probability of each detection engine.
[0042] For example, when detection engine A and detection engine B participate in the competition, if the execution performance of detection engine B is better, t A >t B 、v A <v B or fb A <fb B When B >ef A , then the calling probability P of detection engine B B =P B *110%, the calling probability of detection engine A remains unchanged. At this time, the calling probabilities of detection engine A and detection engine B are different. The calling probability of detection engine B is improved due to the change of its ef value.
[0043] After several calling cycles (for example, after 15 calling cycles), the weighted calling of the better detection engine B will gradually slow down (at this time the probability of detection engine B being called exceeds 80%), and eventually the software security detection system will enter a steady state, that is, the software security detection system will always call detection engine B.
[0044] It should be noted that 80% is an estimated value. It is expected that the software security detection system will enter a stable state when the probability of calling the competitive detection engine is greater than or equal to 80%.
[0045] At this point, detection engine B has completed the entire competition process and is selected by the environment, while detection engine A, which has insufficient execution efficiency, is eliminated. That is, the probability of detection engine B being called enters a stable period, while other detection engines cannot continue to get the opportunity to be called due to their low calling probability.
[0046] The scenario where multiple detection engines participate in the competition is similar. After the initial process call is completed, the detection engine set Ω = {A, B, C...}, when the call probability of the dominant engine A changes to the original 110%, it is not much different from the call probability of other engines in the set Ω' = {B, C...}. With the superposition of the system operation cycle, the probability advantage of the dominant engine being called is also accumulating. After accumulating to a certain period, the call image of the dominant engine will develop exponentially until it enters the call plateau (the call plateau here means that after a certain call period, the system tends to balance, and the dominant engine will be selected with a high probability). The main idea of the algorithm is advantage accumulation. In the case of no accumulated advantage, the call probability of the detection engines in the set Ω is relatively small.
[0047] In this embodiment, the capability detection base module allows new detection engines to join. When more detection engines compete, the software security detection system can achieve the best results. After the competition process converges, the engine result data generated by the competition will be uploaded to the decision module for sorting out the rules of the abstract competition process. These rules will provide users with a better perspective on competition strategies and provide data basis for future strategy reorganization.
[0048] It should be noted that the detection engines participating in the competition are homogeneous detection engines, which are called ecological niches. The following process is used to determine whether the detection engines are homogeneous:
[0049] (1) Assume that a detection engine class has a feature set of ∧(λ1, λ2, λ3...λn). Assume that two detection engines A and B have feature sets A*(A1, A2, A3...An) and B*(B1, B2, B3...Bn) respectively.
[0050] (2) Let S(A*,∧) represent the matching degree of detection engine A for class ∧, with cardinality n.
[0051] (3) Assume that the matching function between the set A* and ∧ is as follows:
[0052]
[0053] A*∩∧ represents the same features of the sets A* and ∧, A*-∧ represents the features that exist in the A* set but not in the ∧ set, and ∧-A* represents the features that exist in the ∧ set but not in the A* set;
[0054] (4) When S(A*,∧)>50%, it is determined that the set A is a member of the ∧ class.
[0055] (5) Similarly, when S(B*, ∧)>50%, it is determined that the B* set belongs to the members of the ∧ class, and A and B belong to the same detection engine.
[0056] Example 2
[0057] This embodiment provides a software security detection system for implementing the method described in Embodiment 1, including an upper layer module, a capability detection base module and a monitoring module;
[0058] The upper layer module is used to initiate the call of the detection engine;
[0059] The capability detection base module is used to select a detection engine to be called according to the calling algorithm after the upper-level module initiates the calling of the detection engine, and use the detection engine to perform the current software security detection; and in a calling cycle, after the called detection engine completes the software security detection, it receives the latest execution performance data of the detection engine collected by the monitoring module, performs detection engine advantage analysis according to each detection engine, and redistributes the calling probability of each detection engine participating in the competition through the decision competition algorithm;
[0060] The monitoring module is used to actively obtain the performance data of the called detection engine during the software security detection process and pass it to the capability detection base module.
[0061] For those skilled in the art, various corresponding changes and modifications can be made according to the above technical solutions and concepts, and all of these changes and modifications should be included in the protection scope of the claims of the present invention.
Claims
1. A decision competition method for homogeneous detection engines in a software security detection system, characterized in that: The specific process is: S1. After the upper layer module (user end) triggers a detection engine call, the capability detection base module selects a detection engine from the competing detection engines according to the calling algorithm and uses the detection engine to perform the software security detection this time. The specific process of calling the algorithm is as follows: In the initial state, P i (first) represents the calling probability of each detection engine at the initial moment of the software security detection system, P i (first) = 1 / n, i∈n, where n is the number of detection engines participating in the competition, i, n∈N + (N + is a positive integer); at the moment of entering the first call cycle, the call probability of each detection engine is P i (second), P i (second) and P i (first) equal; Execution Space Sp i is the range in which the detection engine i is selected, which is expressed as a data interval, Sp i The larger the interval of , the higher the probability that detection engine i is selected; Execution Space SP i =[1,P i *100], probability of obtaining total space SpALL = SP i ∪[Random((1-P i )*100)]; Set the random value function V(p) = Random(SpALL,1); Random(x) represents the random number function, Random(SpALL,1) means randomly selecting a number in SpALL; P i is the calling probability i of the detection engine; the execution space here is represented as a set of closed intervals of data. When the data selected by the random value function hits the value of the execution space corresponding to a detection engine, it means that it is selected and will be called for execution; the execution space of each detection engine is independent; S2. The monitoring module actively obtains the execution performance data of the detection engine for this software security detection and submits it to the capability detection base module; The execution performance data of the detection engine includes running time and subjective running feedback; setting ef i To detect the execution performance of engine i, t i is the running time of detection engine i, v i is the running rate of detection engine i, fb i is the subjective running feedback of detection engine i, that is, the user's evaluation value of detection engine i after it runs; Before entering the first call cycle, each execution performance data of each detection engine is zero; S3. After the software security test is completed, the capability detection base module analyzes the detection engine advantage according to the latest execution performance data of each detection engine, and redistributes the call probability of each detection engine participating in the competition through the decision competition algorithm. At this point, a call cycle ends; after multiple call cycles, the call probability of each detection engine will converge to a fixed range. Due to the accumulation of selection advantages, a certain detection engine will be continuously executed. At this time, the capability detection base module completes the selection process of the advantageous detection engine, and the software security detection system will always call the detection engine with the greater advantage; The specific process of the decision competition algorithm is: After the called detection engine completes a software security test, the capability detection base module receives the latest execution performance data of the detection engine collected by the monitoring module, and compares the execution performance data of each competing detection engine, assigning different weights according to the quality of the execution performance data of each detection engine, and then weighting the calling probability of each detection engine according to the assigned weight to obtain the new calling probability of each detection engine.
2. The method according to claim 1, characterized in that: The Capability Detection Base module allows new detection engines to be added.
3. The method according to claim 1, characterized in that: The detection engines participating in the competition are homogeneous detection engines, which are called ecological niches. The following process is used to determine whether the detection engines are homogeneous: (1) Assume that a detection engine class has a feature set of ∧(λ1, λ2, λ3…λn); Assume that two detection engines A and B have feature sets A*(A1, A2, A3…An) and B*(B1, B2, B3…Bn) respectively; (2) Let S(A*, ∧) represent the matching degree of detection engine A for class ∧, with cardinality n; (3) Assume that the matching function between the set A* and ∧ is as follows: A*∩∧ represents the same features of the sets A* and ∧, A*-∧ represents the features that exist in the A* set but not in the ∧ set, and ∧-A* represents the features that exist in the ∧ set but not in the A* set; (4) When S(A*, ∧)>50%, it is determined that the set A belongs to the members of the ∧ class; (5) Similarly, when S(B*, ∧)>50%, it is determined that the B* set belongs to the members of the ∧ class, and A and B belong to the same detection engine.
4. A software security detection system for implementing the method described in any one of claims 1 to 3, characterized in that: It includes an upper module, a capability detection base module and a monitoring module; The upper layer module is used to initiate the call of the detection engine; The capability detection base module is used to select a detection engine to be called according to the calling algorithm after the upper layer module initiates the calling of the detection engine, and use the detection engine to perform the current software security detection; And it is used to receive the latest execution performance data of the detection engine collected by the monitoring module after the called detection engine completes the software security detection in a calling cycle, perform detection engine advantage analysis according to each detection engine, and redistribute the calling probability of each detection engine participating in the competition through the decision competition algorithm; The monitoring module is used to actively obtain the performance data of the called detection engine during the software security detection process and pass it to the capability detection base module.
