An Attack-Driven Incident Response and Defense System (ADIRDS)

Anthony Cheuk Tung Lai, Siu Ming Yiu, Ping Fan Ke, Alan HoarXiv 2025 · arXiv · OpenReview

Abstract One of the major goals of incident response is to help an organization or a system owner to quickly identify and halt the attacks to minimize the damages (and financial loss) to the system being attacked. Typical incident responses rely very much on the log information captured by the system during the attacks and if needed, may need to isolate the victim from the network to avoid further destructive attacks. However, there are real cases that there are insufficient log records/information for the incident response team to identify the attacks and their origins while the attacked system cannot be stopped due to service requirements (zero downtime online systems) such as online gaming sites. Typical incident response procedures and industrial standards do not provide an adequate solution to address this scenario. In this paper, being motivated by a real case, we propose a solution, called "Attack-Driven Incident Response and Defense System (ADIRDS)" to tackle this problem. ADIRDS is an online monitoring system to run with the real system. By modeling the real system as a graph, critical nodes/assets of the system are closely monitored. Instead of relying on the original logging system, evidence will be collected from the attack technique perspectives. To migrate the risks, realistic honeypots with very similar business context as the real system are deployed to trap the attackers. We successfully apply this system to a real case. Based on our experiments, we verify that our new approach of designing the realistic honeypots is effective, 38 unique attacker's IP addresses were captured. We also compare the performance of our realistic honey with both low and high interactive honeypots proposed in the literature, the results found that our proposed honeypot can successfully cheat the attackers to attack our honeypot, which verifies that our honeypot is more effective.

Introduction: The Challenge of Zero-Downtime Incident Response

Traditional Incident Response (IR) methodologies often rely on a "detect-isolate-analyze" workflow. While effective for standard enterprise environments, this approach falters when applied to critical systems requiring zero downtime, such as online gaming platforms or healthcare databases. In these contexts, taking a system offline for forensic analysis results in unacceptable financial or operational losses. Furthermore, sophisticated attackers frequently employ techniques that leave minimal footprints in standard system logs, rendering traditional forensic tools ineffective.

Table of incidents where typical IR methodologies fail to identify root causes or halt active attacks.

As shown in the table above, incidents involving modified database transactions or native web server backdoors often bypass standard logging, making it impossible for typical IR to decide the attack root cause while the system remains live.

The ADIRDS Framework

To address these gaps, we propose the Attack-Driven Incident Response and Defense System (ADIRDS). ADIRDS is designed to operate alongside production systems, providing real-time monitoring and active defense without requiring service interruption.

The ADIRDS architecture, featuring SIEM integration, a Honeypot Deployment Center, and a Defense Command Center.

The system consists of three primary components:

  1. SIEM (Security Information and Event Management): Collects logs and events for real-time analysis.
  2. Honeypot Deployment Center: Dynamically deploys "realistic" honeypots that mimic the production environment's business context.
  3. Defense Command Center: Orchestrates the response based on intelligence gathered from both the production system and the honeypots.

Methodology: Attack-Driven Evidence Collection

The core of ADIRDS is the shift from passive log review to active, "attack-driven" evidence collection.

  • Graph-Based Monitoring: The system models the network as a graph, identifying "red lines" or suspicious connections between trusted assets and untrusted external entities.
  • MITRE ATT&CK Integration: Instead of searching for generic errors, ADIRDS looks for specific indicators of known attack techniques, such as credential dumping or unauthorized file operations.
  • Realistic Honeypots: A key innovation is the use of "business-context aware" honeypots. Unlike traditional low-interaction traps, these honeypots use real services, follow production naming conventions, and simulate active user sessions. By appearing as high-value, slightly vulnerable production targets, they lure advanced attackers into revealing their origins and methods.

Experimental Results and Case Study

The system was evaluated through a 35-day deployment of a "Web Poker" casino honeypot. The results demonstrated the superior luring capability of context-aware traps:

  • High Engagement: The system recorded over 27,000 accesses and identified 38 unique attacker IP addresses.
  • Effectiveness: In head-to-head comparisons, the realistic honeypot successfully induced attackers to download deceptive configuration files and attempt logins, whereas traditional low and high-interaction honeypots failed to capture any such activity in the same timeframe.
  • Real-World Application: In a live case study involving an online gaming company, ADIRDS identified a native IIS module backdoor that had evaded standard anti-virus and Windows Defender protections.

Conclusion

ADIRDS provides a robust framework for managing cyber-attacks on systems where downtime is not an option. By integrating realistic honeypots and attack-driven monitoring, organizations can identify and mitigate threats in real-time. Future work will focus on automating the design and deployment of these context-aware environments to further reduce the manual effort required by security teams.

Summary updated

Abstract

One of the major goals of incident response is to help an organization or a system owner to quickly identify and halt the attacks to minimize the damages (and financial loss) to the system being attacked. Typical incident responses rely very much on the log information captured by the system during the attacks and if needed, may need to isolate the victim from the network to avoid further destructive attacks. However, there are real cases that there are insufficient log records/information for the incident response team to identify the attacks and their origins while the attacked system cannot be stopped due to service requirements (zero downtime online systems) such as online gaming sites. Typical incident response procedures and industrial standards do not provide an adequate solution to address this scenario. In this paper, being motivated by a real case, we propose a solution, called "Attack-Driven Incident Response and Defense System (ADIRDS)" to tackle this problem. ADIRDS is an online monitoring system to run with the real system. By modeling the real system as a graph, critical nodes/assets of the system are closely monitored. Instead of relying on the original logging system, evidence will be collected from the attack technique perspectives. To migrate the risks, realistic honeypots with very similar business context as the real system are deployed to trap the attackers. We successfully apply this system to a real case. Based on our experiments, we verify that our new approach of designing the realistic honeypots is effective, 38 unique attacker's IP addresses were captured. We also compare the performance of our realistic honey with both low and high interactive honeypots proposed in the literature, the results found that our proposed honeypot can successfully cheat the attackers to attack our honeypot, which verifies that our honeypot is more effective.
BibTeX
@misc{lai2025attack,
  author = {Anthony Cheuk Tung Lai and Siu Ming Yiu and Ping Fan Ke and Alan Ho},
  title = {{An Attack-Driven Incident Response and Defense System (ADIRDS)}},
  year = {2025},
  eprint = {2502.02230},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2502.02230},
}
Figure