Introduction
In the landscape of modern cybersecurity, human behavior remains the most significant vulnerability. Despite the deployment of advanced technical defenses, social engineering and simple negligence continue to serve as primary entry points for attackers. Traditional Information Security Awareness and Education Training (ISAET) programs often struggle to change behavior because they are perceived as generic, unengaging, and disconnected from the actual threats an organization faces. This paper proposes the EASY model (Engaging Stakeholders, Acceptable Behavior, Simple Teaching, Yardstick), a framework that utilizes real-time data from honeypots to create dynamic, relevant, and impactful security training.
Methodology: Low-Cost Threat Intelligence
To bridge the gap between abstract threats and organizational reality, the researchers deployed honeypots—decoy systems designed to attract and log unauthorized activity. These were hosted on low-cost Raspberry Pi Model B devices within a university environment. The hardware specifications for these deployment nodes are detailed in Figure 1.
Two specific honeypot types were utilized to gather diverse threat data:
- Dionaea: A low-interaction honeypot that emulates common protocols (SMB, HTTP, FTP) to capture malware samples.
- Kippo: A medium-interaction honeypot that logs SSH brute-force attempts and records the specific commands executed by attackers post-intrusion.
The EASY Training Model
The EASY model is grounded in Routine Activity Theory (RAT), a criminology framework which posits that a crime requires a motivated offender, a suitable target, and the absence of a capable guardian. The goal of EASY is to transform employees into those "capable guardians" through four pillars:
- Engaging Stakeholders: Involving everyone from executives to vendors to ensure organizational buy-in.
- Acceptable Behavior: Clearly defining and rewarding positive security hygiene.
- Simple Teaching: Using interactive visualizations and live data playbacks rather than static presentations.
- Yardstick: Implementing continuous measurement to track the effectiveness of the training.
Empirical Insights and Results
Data collection yielded significant insights into attacker behavior. The Dionaea dataset captured over 115,000 connections, while Kippo logged over a million login attempts.
Geographic and Temporal Trends
Analysis revealed a "workday" attack cycle, with activity peaking during business hours (9 AM to 5 PM GMT+8). This suggests that many attacks are either automated scripts running on office networks or are timed to blend in with legitimate traffic. Furthermore, geographic analysis showed a wide distribution of connection sources, as illustrated in the connection logs in Figure 2.
While the UK appeared as a major source in the Dionaea set, the authors clarify that these likely represent compromised "zombie" machines rather than the actual location of the human threat actors.
Attack Vectors and Tactics
- Port Targeting: Port 445 (SMB) was the target of 77.52% of hits, confirming its status as a primary vector for automated exploitation.
- Malware Persistence: The 2008 Conficker worm remained the most captured malware, highlighting the longevity of unpatched vulnerabilities in the wild.
- Credential Brute-Forcing: Attackers frequently used "admin" as a username and simple passwords like "123456" or common names. Over half of the attempted passwords were only six to nine characters long.
Conclusion
By integrating live honeypot data into the EASY framework, organizations can move away from "one-size-fits-all" training. Showing employees real-time playbacks of attackers navigating a system or guessing passwords provides a visceral understanding of risk that static materials cannot replicate. This approach not only increases the perceived risk for offenders but also empowers employees to become proactive guardians of the corporate network.
Summary updated