Role-Based User Behavior Simulation in Enterprise IT Environments

Objective:

Develop lightweight simulation agents that mimic realistic user behavior (e.g., office staff, developers, admins) in a controlled lab setup. Agents should execute atomic tasks such as browsing, file access, or command usage, with rule-based or probabilistic scheduling.

Betreuer: Daniel Schlör

Key Tasks:

• Define atomic tasks per user role (e.g., open_browser, edit_file, ssh_login)
• Implement agent behavior scripts using Python or shell automation (e.g., using cron, pyautogui, expect, or similar tools)
• Simulate temporal patterns (e.g., morning login, lunch break, end-of-day)
• Generate logs (system, network) during simulations and validate basic realism (frequency, diversity)

Extension Directions (Master Thesis / Practica):

• Generative Modeling of User Behavior via Additive or Probabilistic Models
• Validation Metrics for Synthetic User Behavior in Security Research

• LLM-Based User Persona Simulation for Dynamic Workflows