Methods and systems for detecting micro-variable traffic patterns and tracing attackers.

By deploying traffic micro-variation honeypots and virtual simulation port clusters, attackers can be collected and traced in real time, solving the problem of not being able to actively detect and locate attackers in existing technologies, and realizing automated defense and autonomous iteration of defense capabilities.

CN123122901APending Publication Date: 2026-11-17HUANENG INFORMATION TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611270022.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-20
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing network defense technologies are unable to proactively detect and capture new types of unknown scanning traffic, such as slow-probing attacks, and cannot identify attackers or form a complete defense loop.

Method used

Deploy lightweight honeypots with micro-variation of traffic, build virtual simulation port clusters, monitor and redirect probe traffic in real time, collect multi-dimensional behavioral data, trace the attackers through source tracing formulas, synchronize with the boundary firewall for interception, and periodically optimize honeypot parameters.

Benefits of technology

It enables proactive trapping of unknown probe traffic and precise tracing of attackers, automatic interception of attacks, reduced operation and maintenance costs, and enhanced defense capabilities and autonomous iteration capabilities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123122901A_ABST
    Figure CN123122901A_ABST
Patent Text Reader

Abstract

This application provides a method and system for detecting and tracing attackers using traffic micro-variation detection behavior, belonging to the field of network deception defense and attack tracing technology. The method includes the following steps: S1: Deploying lightweight honeypots with traffic micro-variation around business assets; S2: Real-time monitoring of network-wide detection and scanning traffic; S3: Generating a raw dataset of four-dimensional detection behavior; S4: Processing the raw dataset of four-dimensional detection behavior output by S3 using a traffic trajectory correlation tracing calculation formula; S5: Synchronizing the tracing data obtained in S4 to the boundary firewall and web application firewall; S6: Iteratively optimizing the traffic micro-variation offset interval defined in S2 based on the dataset. This application combines fixed simulation ports with fake services without real business logic to construct lightweight honeypots. The core business fields of the packets are not modified throughout the process, allowing the honeypot traffic to closely resemble real business characteristics. The deployment process consumes few resources and meets internal network security audit standards.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of network deception defense and attack tracing technology, and more specifically, to a method and system for detecting and tracing attackers through traffic micro-variation detection behavior. Background Technology

[0002] Currently, there are two main types of defense methods in the field of cybersecurity, both of which have significant technical shortcomings. The first type is vulnerability scanning honeypot interference technology that integrates deception defense. The core objective of this technology is only to hide the behavior of the scanning initiator itself. It can only cause simple interference to the scanning process and does not have the ability to actively capture external probe traffic, locate attackers, coordinate with devices to intercept attacks, or autonomously optimize simulation features. The second type is traditional network anti-probe technology, which adopts a passive defense logic. It only relies on a preset static feature library to identify and intercept known probe behavior. It cannot actively perceive or capture network-wide asset mapping, slow-probing new unknown scan traffic, or fully collect multi-dimensional behavioral information of attackers. It also cannot reconstruct the complete attack chain to locate the attacker's identity. Neither of these existing technologies has a coherent defense execution chain. They can only achieve local single-point protection and cannot form a complete closed loop from capturing unknown probes to autonomously optimizing defense capabilities. Summary of the Invention

[0003] The purpose of this invention is to address the significant shortcomings in the current network asset protection capabilities against persistent and covert scanning and detection.

[0004] To achieve the above-mentioned objectives and improve the above-mentioned problems, this invention provides a method for detecting and trapping attackers based on micro-variable traffic patterns, comprising the following steps: S1: Deploy lightweight honey spots with minor traffic variations around business assets and build virtual simulation port clusters and fake services; S2: Real-time monitoring of network-wide detection and scanning traffic, and targeted redirection to lightweight honey spots with minor traffic variations to achieve active trapping; S3: Lightweight honeypot collection of attacker multidimensional probing behavior data through traffic micro-variation to generate four-dimensional probing behavior raw collection dataset; S4: Based on the traffic trajectory correlation and source tracing calculation formula, the original dataset of four-dimensional probe behavior output by S3 is processed to trace the attacker's IP, device fingerprint, and complete probe attack chain. The formula is: ; The coefficient represents the homology matching coefficient, with a fixed value range of 0-1. , , , The preset fixed weight constants are set to values ​​of 0.4, 0.2, 0.3, and 0.1 respectively. This represents the normalized port access sequence value. This represents the normalized access frequency value per unit time. This is a normalized fingerprint encoding value for the scanning tool; This represents the normalized detection duration value. S5: Synchronizes the source tracing data obtained from S4 to the border firewall and web application protection wall to automatically block same-origin probe and scanning attacks; S6: Periodically summarize the original data set of four-dimensional detection behavior generated by S3, and iteratively optimize the flow micro-variation offset interval defined by S2 based on the data set.

[0005] As a preferred technical solution of this application, step S1 specifically includes: S101: Define the isolated network segment for the enterprise's core business hosts. Divide three independent network segments outward from the Web business host and the database business host as the honey spot's exclusive deployment area. Configure a one-way access isolation policy to prohibit the honey spot network segment's traffic from flowing back into the real business asset network segment. S102: Continuously collect all real business traffic packets for 72 hours, extract three types of data—packet fixed length, request-response delay, and transmission protocol identifier—to form a real business traffic baseline; based on this baseline, set a traffic micro-variation offset range, and set rigid quantization offset constraints: the packet fixed length range is ±3% of the baseline value, and the request-response delay range is ±5% of the baseline value; S103: Deploy a fixed virtual emulation port cluster within the honeypot's dedicated network segment.

[0006] As a preferred technical solution of this application, step S2 specifically includes: S201: Capture all inbound network packets from the enterprise border gateway in real time, and parse each packet for three types of identifiers: target access port, access frequency per minute, and total amount of data packets in a single request. S202: Set a fixed judgment threshold. When a single Internet Protocol address accesses 3 or more different ports within one minute, the traffic is marked as normal. If the number of ports accessed by a single IP in one minute is less than 3, it is judged as normal business traffic. S203: The marked complete traffic is forwarded to the virtual simulation port cluster, and normal business traffic is directly allowed to the real business asset network segment, realizing the targeted capture of all probe traffic.

[0007] As a preferred technical solution of this application, step S3 specifically includes: S301: After the honeypot receives the traffic redirected by S2, it strictly records the complete sequence of the attacker's port access time according to the order of access. S302: Fixed collection of four types of message characteristic data: message offset length, response round-trip time, static fingerprint encoding of scanning tool, and total duration of single detection; S303: Binds a millisecond-level precise timestamp to each probe behavior, integrates port time sequence, access frequency, four types of message characteristics, and timestamp, and uniformly generates a four-dimensional probe behavior raw data set.

[0008] As a preferred technical solution in this application, in step S4, , , , The numerical values ​​corresponding to the normalized four-dimensional feature vectors are based on... , , , Construct standardized behavioral feature vectors: ; A fixed threshold of 0.7 is set for the matching coefficient. If two or more detection behaviors are calculated... If the value is greater than 0.7, it is determined that the batch of probes was initiated by the same attacker; at this time, all related probe records are automatically linked together, and three types of source tracing data results are output: attacker IP, terminal device fingerprint, and complete probe access link.

[0009] As a preferred technical solution of this application, step S5 specifically includes... S501: Extract the two core features of attacker IP and device fingerprint from the source tracing results output by S4, encapsulate them into standardized interception command data packets, and remove all redundant collected data; S502: Issue two types of fixed blocking policies to the perimeter firewall and web application firewall: configure permanent blocking rules for attackers who continuously scan in batches, and configure 72-hour time-limited blocking rules for attackers who make single low-frequency attempts. S503: After receiving the interception command, the security device will block all inbound traffic from the same source attacker in real time and continue to block until the time limit rule expires or the permanent blocking policy is manually lifted.

[0010] As a preferred technical solution of this application, step S6 specifically includes: S601: Using a fixed statistical period of 7 days, extract the original data set of all four-dimensional detection behaviors within the period. Based on whether the honey spot successfully captured the corresponding detection traffic, divide the dataset into a successfully captured subset and a failed captured subset, and count the total number of samples in the two subsets respectively. S602: Substitute into the honey spot flow simulation fit calculation formula: ; in To improve the accuracy of honey spot flow simulation; This represents the total number of samples contained in the successfully captured subset within the period; The total number of samples contained in the failed trap subset within the period; when the fit... When the value is below 85%, the honey spot flow parameter optimization operation is triggered; S603: Extract the fingerprint features of the scanning tool stored in the decoy failure subset, and adjust the percentage boundary of the traffic micro-variation offset interval preset in S1 based on the features. After adjustment, the fixed length offset of the message still fluctuates within ±3%, and the request response delay offset still fluctuates within ±5%. The traffic control parameters of all traffic micro-variation lightweight honeypots are updated synchronously.

[0011] The system for detecting and trapping traffic micro-variations and tracing attackers includes a honeypot deployment and traffic baseline configuration module, a detection traffic diversion and trapping module, a detection behavior data acquisition and storage module, a traffic trajectory tracing and calculation module, a security device linkage countermeasure module, and a honeypot parameter iterative optimization module. The honeypot deployment and traffic baseline configuration module specifically includes: The network segment isolation configuration submodule completes the division of dedicated network segments for multi-layered honeypots and the configuration of one-way isolation policies; the traffic baseline and offset interval setting submodule collects 72 hours of business traffic to generate a real business traffic baseline, and generates a traffic micro-variation offset interval according to the quantitative constraints of packet length ±3% and response latency ±5%, without changing the transmission protocol identifier; the virtual simulation service deployment submodule deploys a specified simulation port and a fake service without business logic; The detection flow diversion and trapping module specifically includes: The border packet capture and parsing submodule extracts information such as port, access frequency, and total number of data packets of inbound packets from the border gateway in real time; the probe traffic judgment submodule distinguishes between probe access traffic and normal business traffic based on the threshold of the number of ports accessed by a single IP per minute; the traffic forwarding submodule forwards probe traffic to honeypots deployed by the preceding modules, while normal traffic is allowed to pass through to real business assets.

[0012] As a preferred technical solution of this application, the detection behavior data acquisition and storage module specifically includes: The port timing recording submodule fully preserves the attacker's original port scanning order; the multi-dimensional feature collection submodule uniformly collects four types of features: packet length, latency, tool fingerprint, and probe duration; the dataset encryption and storage submodule binds a millisecond timestamp to each probe behavior, integrates all information to generate a four-dimensional probe behavior original collection dataset, and stores it in encryption. The traffic trajectory tracing and calculation module specifically includes: The feature vector construction submodule reads the original dataset of four-dimensional detection behavior collected by the preceding module and generates standardized behavior feature vectors; the same-source behavior matching submodule completes the same-source attacker identification based on the features; the attack link restoration output submodule connects the same-source detection records, outputs the attacker's IP, device fingerprint and complete detection link, and transmits the output data to the security linkage countermeasure module.

[0013] As a preferred technical solution of this application, the security device linkage countermeasure module specifically includes: The interception instruction encapsulation submodule extracts the core characteristics of attackers in the source tracing chain and generates standardized instructions; the hierarchical interception policy distribution submodule distributes two types of interception rules, permanent and 72-hour time-limited, to the perimeter firewall and web application protection wall; the traffic blocking execution feedback submodule drives the security device to block all inbound traffic from the same source attacker. The honey spot parameter iterative optimization module specifically includes: The periodic dataset classification and statistics submodule reads the original data collection dataset of four-dimensional detection behavior from the storage module every 7 days and divides it into two sub-datasets: successful trapping and failed trapping. The fitness determination submodule determines whether to trigger optimization based on the number of samples in the two categories. The offset interval parameter update submodule adjusts the boundary of the offset interval of honey spot flow micro-variation based on the fingerprint of failed samples and updates all honey spot flow control parameters simultaneously.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: In the scheme of this application: 1. By setting up dedicated network segments for each honeypot in a layered manner and configuring a one-way access isolation policy, a lightweight honeypot is constructed by combining a fixed simulation port with a fake service without real business logic. At the same time, a traffic micro-variation offset range with quantitative constraints is set based on the baseline of real business traffic. The core business fields of the packets are not modified throughout the process. This will not cause the honeypot traffic to intrude into real business assets, and the honeypot traffic will also fit the characteristics of real business. The deployment process has low resource consumption and can meet the internal network security audit specifications. There are no business occupation and insufficient compliance issues caused by heavy honeypots. 2. By using fixed judgment criteria to distinguish between probing traffic and normal business traffic, only probing traffic is forwarded to the honeypot to complete the trapping, while normal user access traffic is allowed to pass directly. This fully captures all scanning and probing behaviors without interfering with legitimate users' access to real business assets, thus achieving refined traffic diversion processing. 3. By performing full-dimensional information collection on the captured probing behavior, each probing record is bound with a millisecond-level timestamp and integrated into a unified encrypted storage four-dimensional probing behavior original collection dataset. This dataset completely preserves all behavioral traces of the attacker, such as the original scanning sequence, message characteristics, and probing duration. The data is complete and tamper-proof, providing complete and reliable data support for attacker tracing and honeypot iteration optimization. 4. By leveraging standardized behavioral characteristics to identify attackers from the same source, it can connect multiple scattered probe records to fully reconstruct the attack chain, accurately locate the attacker's IP and device fingerprint, and address the core shortcomings of existing defense methods that cannot trace the attacker's main body or trace the complete probe path. The tracing results can be synchronized to firewalls and web application firewalls, issuing differentiated blocking rules for scanning behaviors of different risk levels, automatically blocking all subsequent access traffic from the same source attacker, achieving proactive countermeasures without manual intervention, and reducing security operation and maintenance manpower costs. Furthermore, it periodically summarizes all collected datasets and distinguishes between decoy effectiveness samples, adjusting the traffic micro-variation offset range based on the scanning tool characteristics of decoy failure samples, continuously optimizing the honeypot simulation effect, and effectively improving the decoy adaptation capability to various new scanning tools, enabling autonomous iterative upgrades of defense capabilities. Attached Figure Description

[0015] Figure 1 The flowchart of the method and system for detecting micro-variable traffic behavior and tracing attackers provided in this application is shown. Detailed Implementation

[0016] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0017] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0018] It should be noted that, unless otherwise specified, the embodiments and features and technical solutions in the present invention can be combined with each other.

[0019] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0020] Please refer to Figure 1 The method for detecting subtle traffic variations and tracing attackers includes the following steps: S1: Deploy lightweight honey spots with minor traffic variations around business assets and build virtual simulation port clusters and fake services; S2: Real-time monitoring of network-wide detection and scanning traffic, and targeted redirection to lightweight honey spots with minor traffic variations to achieve active trapping; S3: Lightweight honeypot collection of attacker multidimensional probing behavior data through traffic micro-variation to generate four-dimensional probing behavior raw collection dataset; S4: Based on the traffic trajectory correlation and source tracing calculation formula, the original dataset of four-dimensional probe behavior output by S3 is processed to trace the attacker's IP, device fingerprint, and complete probe attack chain. The formula is: ; The coefficient represents the homology matching coefficient, with a fixed value range of 0-1. , , , The preset fixed weight constants are set to values ​​of 0.4, 0.2, 0.3, and 0.1 respectively. This represents the normalized port access sequence value. This represents the normalized access frequency value per unit time. This is a normalized fingerprint encoding value for the scanning tool; This represents the normalized detection duration value. S5: Synchronizes the source tracing data obtained from S4 to the border firewall and web application protection wall to automatically block same-origin probe and scanning attacks; S6: Periodically summarize the original data set of four-dimensional detection behavior generated by S3, and iteratively optimize the flow micro-variation offset interval defined by S2 based on the data set.

[0021] Furthermore, S1 specifically includes: S101: Define the isolated network segment for the enterprise's core business hosts. Divide three independent network segments outward from the Web business host and the database business host as the honey spot's exclusive deployment area. Configure a one-way access isolation policy to prohibit the honey spot network segment's traffic from flowing back into the real business asset network segment. S102: Continuously collect all real business traffic packets for 72 hours, extract three types of data—packet fixed length, request-response delay, and transmission protocol identifier—to form a real business traffic baseline; based on this baseline, set a traffic micro-variation offset range and set rigid quantitative offset constraints: the packet fixed length value range is ±3% of the baseline value, the request-response delay value range is ±5% of the baseline value, and the transmission protocol identifier fully reuses the baseline without modification. Only the honeypot traffic parameters are allowed to take values ​​within the above quantitative range, without modifying the core business fields of the packets, ensuring that the honeypot traffic complies with the internal network security audit specifications; S103: Deploy a fixed virtual simulation port cluster within the honeypot's dedicated network segment, configuring four high-frequency attack ports: 80, 443, 3306, and 22, to build corresponding static fake web services, fake database services, and fake remote login services; the fake services only output standardized fixed response messages and do not carry real business operation logic.

[0022] Furthermore, S2 specifically includes: S201: Capture all inbound network packets from the enterprise border gateway in real time, and parse each packet for three types of identifiers: target access port, access frequency per minute, and total amount of data packets in a single request. S202: Set a fixed judgment threshold. When a single Internet Protocol address accesses 3 or more different ports within one minute, the traffic is marked as normal. If the number of ports accessed by a single IP in one minute is less than 3, it is judged as normal business traffic. S203: The marked complete traffic is forwarded to the virtual simulation port cluster, and normal business traffic is directly allowed to the real business asset network segment, realizing the targeted capture of all probe traffic.

[0023] Furthermore, S3 specifically includes: S301: After receiving the traffic redirected by S2, the honeypot strictly records the attacker's port access sequence in the order of access, without tampering with the original scanning order; S302: Fixed collection of four types of message characteristic data: message offset length, response round-trip time, scanning tool static fingerprint encoding, and total duration of a single probe, without deleting any collection dimension; S303: Binds a millisecond-level precise timestamp to each detection behavior, integrates port time sequence, access frequency, four types of message characteristics, and timestamp, and uniformly generates a four-dimensional detection behavior raw data set, which is stored in an independent encrypted storage partition; this dataset serves as the sole raw material for S4 source tracing operations and S6 iterative optimization.

[0024] Furthermore, in S4, , , , The numerical values ​​corresponding to the normalized four-dimensional feature vectors are based on... , , , Construct standardized behavioral feature vectors: ; A fixed threshold of 0.7 is set for the matching coefficient. If two or more detection behaviors are calculated... If the value is greater than 0.7, it is determined that the batch of probes were initiated by the same attacker. At this time, all related probe records are automatically linked together, and three types of source tracing data results are output: attacker IP, terminal device fingerprint, and complete probe access link. The output data is directly supplied to S5. in =0.4 (Port Access Sequence, Highest Weight) Attackers have fixed behavioral habits in port scanning. The order in which attackers traverse ports has strong personal / tool ​​characteristics. There is almost no high degree of overlap in the port scanning sequences of different attackers. Even if the attacker adjusts the scanning speed, the port traversal logic will not easily change. It is the most stable and most identifiable core feature to distinguish different attackers. Therefore, it is assigned the highest weight of 0.4. =0.2 (Access frequency per unit time, medium weight) Attackers can manually adjust the scanning packet sending rate. The same attacker can launch high-frequency batch scanning or switch to slow low-frequency asset mapping mode. The access frequency is controllable by humans and fluctuates greatly. It can only be used as an auxiliary distinguishing feature. Its identification is weaker than port sequence and tool fingerprint. The weight is assigned to 0.2. =0.3 (Scanning tool fingerprint encoding, second highest weight) Probing tools such as Nmap, Masscan, Xray, and custom scanning scripts carry unique packet fingerprint features. The same attacker uses the same scanning tool for a long time, and the fingerprint matching stability is second only to the port access sequence. There are only a few special cases where attackers temporarily change scanning tools, and the identification is slightly lower than the port sequence, so a weight of 0.3 is assigned. =0.1 (probing duration, minimum weight) The total duration of a single probing is affected by both environmental and human factors, such as external network jitter, honeypot response delay, and attacker interruption. The value has the strongest randomness. This feature alone cannot distinguish attackers and can only be used as a weak auxiliary reference. Therefore, the minimum weight is set to 0.1. A threshold of 0.7 represents the optimal balance between traceability completeness and accuracy in real-world testing scenarios. Thresholds that are too high or too low will result in significant defensive flaws. If the threshold is set below 0.7 (such as 0.5 or 0.6), the matching standard is too lenient. Even a small number of overlapping features can lead to a single entity being identified as the same, making it highly susceptible to cross-attacker mismatches. For example, if two unrelated attackers happen to use the same scanning tools and have similar scanning speeds, the system might mistakenly identify them as the same entity, triggering a bulk false block by the firewall, blocking normal access traffic and impairing business availability.

[0025] If the threshold is set higher than 0.7 (such as 0.85 or 0.9), the matching criteria are too strict, and only when all four features are highly consistent can they be determined to be from the same source. When the same attacker slightly adjusts the scanning packet sending speed or temporarily changes to a simple scanning tool, the matching coefficient will be lower than the threshold. The system will be unable to connect the attacker's probing behavior launched in multiple time periods, breaking the complete attack chain, losing source tracing data, and rendering the proactive source tracing defense function ineffective.

[0026] The balancing advantage of a threshold of 0.7 has been verified through 720 hours of real-world testing on enterprise public web assets: when When the value is ≥0.7, at least two types of high-weight features, namely port access sequence and scanning tool fingerprint, are highly overlapping among the four features, and the probability of cross-attacker false match is less than 0.3%. At the same time, it allows attackers to slightly adjust the access frequency and probe duration, without interrupting the continuous probe behavior carried out by the same attacker in segments. It can completely restore the entire attack chain under the premise of extremely low false interception rate, taking into account both security accuracy and traceability integrity.

[0027] Furthermore, S5 specifically includes S501: Extract the two core features of attacker IP and device fingerprint from the source tracing results output by S4, encapsulate them into standardized interception command data packets, and remove all redundant collected data; S502: Issue two types of fixed blocking policies to the perimeter firewall and web application firewall: configure permanent blocking rules for attackers who continuously scan in batches, and configure 72-hour time-limited blocking rules for attackers who make single low-frequency attempts. S503: After receiving the interception command, the security device will block all inbound traffic from the same source attacker in real time and continue to block until the time limit rule expires or the permanent blocking policy is manually lifted.

[0028] Furthermore, S6 specifically includes: S601: Using a fixed statistical period of 7 days, extract the original data set of all four-dimensional detection behaviors within the period. Based on whether the honey spot successfully captured the corresponding detection traffic, divide the dataset into a successfully captured subset and a failed captured subset, and count the total number of samples in the two subsets respectively. S602: Substitute into the honey spot flow simulation fit calculation formula: ; in To improve the accuracy of honey spot flow simulation; This represents the total number of samples contained in the successfully captured subset within the period; The total number of samples contained in the failed trap subset within the period; when the fit... When the value is below 85%, the honey spot flow parameter optimization operation is triggered; S603: Extract the fingerprint features of the scanning tool stored in the decoy failure subset, and adjust the percentage boundary of the traffic micro-variation offset interval preset in S1 based on the features. After adjustment, the fixed length offset of the message still fluctuates within ±3%, and the request response delay offset still fluctuates within ±5%. Simultaneously update the traffic control parameters of all traffic micro-variation lightweight honeypots to improve the honeypot's decoy adaptation capability to new scanning tools.

[0029] The system for detecting and trapping traffic micro-variations and tracing attackers includes a honeypot deployment and traffic baseline configuration module, a detection traffic diversion and trapping module, a detection behavior data acquisition and storage module, a traffic trajectory tracing and calculation module, a security device linkage countermeasure module, and a honeypot parameter iterative optimization module. The honeypot deployment and traffic baseline configuration module specifically includes: The network segment isolation configuration submodule completes the division of dedicated network segments for multi-layered honeypots and the configuration of one-way isolation policies; the traffic baseline and offset interval setting submodule collects 72 hours of business traffic to generate a real business traffic baseline, and generates a traffic micro-variation offset interval according to the quantitative constraints of packet length ±3% and response latency ±5%, without changing the transmission protocol identifier; the virtual simulation service deployment submodule deploys a specified simulation port and a fake service without business logic; The detection flow diversion and trapping module specifically includes: The border packet capture and parsing submodule extracts information such as port, access frequency, and total number of data packets of inbound packets from the border gateway in real time; the probe traffic judgment submodule distinguishes between probe access traffic and normal business traffic based on the threshold of the number of ports accessed by a single IP per minute; the traffic forwarding submodule forwards probe traffic to honeypots deployed by the preceding modules, while normal traffic is allowed to pass through to real business assets.

[0030] The detection behavior data acquisition and storage module specifically includes: The port timing recording submodule fully preserves the attacker's original port scanning order; the multi-dimensional feature collection submodule uniformly collects four types of features: packet length, latency, tool fingerprint, and probe duration; the dataset encryption and storage submodule binds a millisecond timestamp to each probe behavior, integrates all information to generate a four-dimensional probe behavior original collection dataset, and stores it in encryption. The traffic trajectory tracing and calculation module specifically includes: The feature vector construction submodule reads the original dataset of four-dimensional detection behavior collected by the preceding module and generates standardized behavior feature vectors; the same-source behavior matching submodule completes the same-source attacker identification based on the features; the attack link restoration output submodule connects the same-source detection records, outputs the attacker's IP, device fingerprint and complete detection link, and transmits the output data to the security linkage countermeasure module.

[0031] The safety equipment linkage countermeasure module specifically includes: The interception instruction encapsulation submodule extracts the core characteristics of attackers in the source tracing chain and generates standardized instructions; the hierarchical interception policy distribution submodule distributes two types of interception rules, permanent and 72-hour time-limited, to the perimeter firewall and web application protection wall; the traffic blocking execution feedback submodule drives the security device to block all inbound traffic from the same source attacker. The honey spot parameter iterative optimization module specifically includes: The periodic dataset classification and statistics submodule reads the original data collection dataset of four-dimensional detection behavior from the storage module every 7 days and divides it into two sub-datasets: successful trapping and failed trapping. The fitness determination submodule determines whether to trigger optimization based on the number of samples in the two categories. The offset interval parameter update submodule adjusts the boundary of the offset interval of honey spot flow micro-variation based on the fingerprint of failed samples and updates all honey spot flow control parameters simultaneously.

[0032] The usage process of the traffic micro-variation detection behavior trapping and attacker tracing method and system provided by this invention is as follows: pass.

[0033] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection, an electrical connection, or a connection that allows communication between them; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0034] Obviously, the embodiments described above are merely some embodiments of the present invention, not all embodiments. The accompanying drawings show preferred embodiments of the present invention, but do not limit the patent scope of the present invention. The present invention can be implemented in many different forms; rather, these embodiments are provided to provide a more thorough and complete understanding of the disclosure of the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the patent protection scope of this invention.

Claims

1. A method for detecting and trapping attackers based on micro-variable traffic patterns, characterized in that, The following steps are included: S1: Deploy lightweight honey spots with minor traffic variations around business assets and build virtual simulation port clusters and fake services; S2: Real-time monitoring of network-wide detection and scanning traffic, and targeted redirection to lightweight honey spots with minor traffic variations to achieve active trapping; S3: Lightweight honeypot collection of attacker multi-dimensional probing behavior data through traffic micro-variation to generate a four-dimensional probing behavior raw collection dataset; S4: Based on the traffic trajectory correlation and source tracing calculation formula, the original dataset of four-dimensional probe behavior output by S3 is processed to trace the attacker's IP, device fingerprint, and complete probe attack chain. The formula is: ; The coefficients are homologous matching coefficients, with a fixed value range of 0-1; , , , The preset fixed weight constants are set to values ​​of 0.4, 0.2, 0.3, and 0.1 respectively. This represents the normalized port access sequence value. This represents the normalized access frequency value per unit time. This is a normalized fingerprint encoding value for the scanning tool; This represents the normalized detection duration value. S5: Synchronizes the source tracing data obtained from S4 to the border firewall and web application protection wall to automatically block same-origin probe and scanning attacks; S6: Periodically summarize the original data set of four-dimensional detection behavior generated by S3, and iteratively optimize the flow micro-variation offset interval defined by S2 based on the data set.

2. The method for detecting and trapping attackers by detecting micro-variables in traffic flow and tracing attacker origins according to claim 1, characterized in that, Specifically, S1 includes: S101: Define the isolated network segment for the enterprise's core business hosts. Divide three independent network segments outward from the Web business host and the database business host as the honey spot's exclusive deployment area. Configure a one-way access isolation policy to prohibit the honey spot network segment's traffic from flowing back into the real business asset network segment. S102: Continuously collect all real business traffic packets for 72 hours, extract three types of data—packet fixed length, request-response delay, and transmission protocol identifier—to form a real business traffic baseline; based on this baseline, set a traffic micro-variation offset range, and set rigid quantization offset constraints: the packet fixed length range is ±3% of the baseline value, and the request-response delay range is ±5% of the baseline value; S103: Deploy a fixed virtual emulation port cluster within the honeypot's dedicated network segment.

3. The method for detecting and trapping attackers by detecting micro-variables in traffic flow and tracing attacker origins according to claim 2, characterized in that, Specifically, S2 includes: S201: Capture all inbound network packets from the enterprise border gateway in real time, and parse each packet for three types of identifiers: target access port, access frequency per minute, and total amount of data packets in a single request. S202: Set a fixed judgment threshold. When a single Internet Protocol address accesses 3 or more different ports within one minute, the traffic is marked as normal. If the number of ports accessed by a single IP in one minute is less than 3, it is judged as normal business traffic. S203: The marked complete traffic is forwarded to the virtual simulation port cluster, and normal business traffic is directly allowed to the real business asset network segment, realizing the targeted capture of all probe traffic.

4. The method for detecting and trapping attackers by detecting micro-variables in traffic according to claim 3, characterized in that, Specifically, S3 includes: S301: After the honeypot receives the traffic redirected by S2, it strictly records the complete sequence of the attacker's port access time according to the order of access. S302: Fixed collection of four types of message characteristic data: message offset length, response round-trip time, static fingerprint encoding of scanning tool, and total duration of single detection; S303: Binds a millisecond-level precise timestamp to each probe behavior, integrates port time sequence, access frequency, four types of message characteristics, and timestamp, and uniformly generates a four-dimensional probe behavior raw data set.

5. The method for detecting and trapping attackers by detecting micro-variables in traffic according to claim 4, characterized in that, In S4, , , , The numerical values ​​corresponding to the normalized four-dimensional feature vectors are based on... , , , Construct standardized behavioral feature vectors: ; A fixed matching coefficient threshold of 0.7 is set. If two or more detection behaviors are calculated... If the value is greater than 0.7, it is determined that the batch of probes was initiated by the same attacker; at this time, all related probe records are automatically linked together, and three types of source tracing data results are output: attacker IP, terminal device fingerprint, and complete probe access link.

6. The method for detecting micro-variable traffic patterns and tracing attackers according to claim 5, characterized in that, Specifically, S5 includes S501: Extract the two core features of attacker IP and device fingerprint from the source tracing results output by S4, encapsulate them into standardized interception command data packets, and remove all redundant collected data; S502: Issue two types of fixed blocking policies to the perimeter firewall and web application firewall: configure permanent blocking rules for attackers who continuously scan in batches, and configure 72-hour time-limited blocking rules for attackers who make single low-frequency attempts. S503: After receiving the interception command, the security device will block all inbound traffic from the same source attacker in real time and continue to block until the time limit rule expires or the permanent blocking policy is manually lifted.

7. The method for detecting and trapping attackers by detecting micro-variables in traffic according to claim 6, characterized in that, Specifically, S6 includes: S601: Using a fixed statistical period of 7 days, extract the original data set of all four-dimensional detection behaviors within the period. Based on whether the honey spot successfully captured the corresponding detection traffic, divide the dataset into a successfully captured subset and a failed captured subset, and count the total number of samples in the two subsets respectively. S602: Substitute into the honey spot flow simulation fit calculation formula: ; in To improve the accuracy of honey spot flow simulation; This represents the total number of samples contained in the successfully captured subset within the period; The total number of samples contained in the failed trap subset within the period; when the fit... When the value is below 85%, the honey spot flow parameter optimization operation is triggered; S603: Extract the fingerprint features of the scanning tool stored in the decoy failure subset, and adjust the percentage boundary of the traffic micro-variation offset interval preset in S1 based on the features. After adjustment, the fixed length offset of the message still fluctuates within ±3%, and the request response delay offset still fluctuates within ±5%. The traffic control parameters of all traffic micro-variation lightweight honeypots are updated synchronously.

8. A system for detecting and tracing traffic micro-variations, employing the method for detecting and tracing traffic micro-variations as described in any one of claims 1-7, comprising a honeypot deployment and traffic baseline configuration module, a detection traffic diversion and trapping module, a detection behavior data acquisition and storage module, a traffic trajectory tracing and calculation module, a security device linkage countermeasure module, and a honeypot parameter iterative optimization module, characterized in that, The honeypot deployment and traffic baseline configuration module specifically includes: The network segment isolation configuration submodule completes the division of dedicated network segments for multi-layered honeypots and the configuration of one-way isolation policies; the traffic baseline and offset interval setting submodule collects 72 hours of business traffic to generate a real business traffic baseline, and generates a traffic micro-variation offset interval according to the quantitative constraints of packet length ±3% and response latency ±5%, without changing the transmission protocol identifier; the virtual simulation service deployment submodule deploys a specified simulation port and a fake service without business logic; The detection flow diversion and trapping module specifically includes: The border packet capture and parsing submodule extracts information such as port, access frequency, and total number of data packets of inbound packets from the border gateway in real time; the probe traffic judgment submodule distinguishes between probe access traffic and normal business traffic based on the threshold of the number of ports accessed by a single IP per minute; the traffic forwarding submodule forwards probe traffic to honeypots deployed by the preceding modules, while normal traffic is allowed to pass through to real business assets.

9. The traffic micro-variation detection behavior trapping and attacker tracing system according to claim 7, characterized in that, The detection behavior data acquisition and storage module specifically includes: The port timing recording submodule fully preserves the attacker's original port scanning order; the multi-dimensional feature collection submodule uniformly collects four types of features: packet length, latency, tool fingerprint, and probe duration; the dataset encryption and storage submodule binds a millisecond timestamp to each probe behavior, integrates all information to generate a four-dimensional probe behavior original collection dataset, and stores it in encryption. The traffic trajectory tracing and calculation module specifically includes: The feature vector construction submodule reads the original dataset of four-dimensional detection behavior collected by the preceding module and generates standardized behavior feature vectors; the same-source behavior matching submodule completes the same-source attacker identification based on the features; the attack link restoration output submodule connects the same-source detection records, outputs the attacker's IP, device fingerprint and complete detection link, and transmits the output data to the security linkage countermeasure module.

10. The traffic micro-variation detection behavior trapping and attacker tracing system according to claim 7, characterized in that, The security equipment linkage countermeasure module specifically includes: The interception instruction encapsulation submodule extracts the core characteristics of attackers in the source tracing chain and generates standardized instructions; the hierarchical interception policy distribution submodule distributes two types of interception rules, permanent and 72-hour time-limited, to the perimeter firewall and web application protection wall; the traffic blocking execution feedback submodule drives the security device to block all inbound traffic from the same source attacker. The honey spot parameter iterative optimization module specifically includes: The periodic dataset classification and statistics submodule reads the original data collection dataset of four-dimensional detection behavior from the storage module every 7 days and divides it into two sub-datasets: successful trapping and failed trapping. The fitness determination submodule determines whether to trigger optimization based on the number of samples in the two categories. The offset interval parameter update submodule adjusts the boundary of the offset interval of honey spot flow micro-variation based on the fingerprint of failed samples and updates all honey spot flow control parameters simultaneously.