LLM-Driven Red Team Agent Platform for Adversarial Drills and Real-World Scenario Simulation
基於 LLM 驅動的紅隊代理攻防演練與場域模擬實戰平台
Research projects and co-authored publications by Te-En Wei (Wilbur Wei). Published work covers AI agents, zero trust and risk assessment. Future directions include drone and robot communication security, and PUF hardware security applications.
基於 LLM 驅動的紅隊代理攻防演練與場域模擬實戰平台
Building on existing AI security and zero-trust research, I plan to deepen work in communication security and hardware security. The directions below are prospective research plans, open to discussion on problem definition, experimental design and industry collaboration.
How can devices, controllers and services establish trust in commands and data?
Planned work examines authentication, command authorisation and communication protection between drones, robots and control systems. Threat scenarios such as impersonation, replay and unauthorised control will inform evaluation of defences and their effects on latency, availability and operational safety.
How can hardware characteristics support device identity and key protection?
Explore Physical Unclonable Functions (PUFs) for device identification, authentication and key generation. Planned evaluation considers reliability under environmental variation, resource constraints and resistance to attacks, as well as connections to device and communication security.
These are prospective directions. Published papers and awards are listed separately.
AHAF
Read the awarded paper ↗HMELF
Read the awarded paper ↗STIE-ZTA
Read the awarded paper ↗A zero-trust authorisation framework for AI agents. Controls across role, resource and action are combined with human confirmation and anomaly detection to address excessive privileges.
A study of encrypted lateral movement using Ligolo-ng tunnelling attacks. HMELF combines network traffic features with endpoint behaviour in a hybrid ensemble-learning framework.
A multi-agent penetration-testing framework using retrieval-augmented generation and shared memory. Reconnaissance and vulnerability analysis were evaluated on VulnHub targets; automated exploitation remains limited by environmental factors.
An account-level AD risk model combining risk assessment and attack graphs. An APT29 case study examines privilege, impact and attack paths beyond a simple high-privilege ranking.
Zero-trust architecture requires continuous verification of every subject that requests access. The practical difficulty is that enterprises already run many security tools — intrusion detection, firewalls, endpoint protection — each producing its own data, with no unified way to combine them into a judgement about whether a subject can be trusted. STIE-ZTA is a policy-based, self-adaptive trust inference framework that aggregates and analyses this heterogeneous data to evaluate each subject's behavioural trust level continuously, so that organisations can make timelier and better-grounded access decisions while getting more out of the infrastructure they already have.
An AD-focused study that turns unstructured threat intelligence into feature networks and combines risk analysis with language models to support vulnerability assessment and patch prioritisation.
AI 方法建構 APT 駭客家族攻擊情資特徵拓樸及風險評估架構-以 AD 伺服器攻擊為案例研究
Journal article and citation ↗The journal page provides the English title, romanized author names and bibliographic details.
This is the journal extension of the CISC 2025 conference study.
Related versionAn AD server attack case study examining AI methods for constructing APT-family attack intelligence feature topology and a risk matrix.
This is the conference version. The later journal version has a revised title and author list; each record retains its own bibliography.
Related versionTitles and author order follow each publication version. The seven records include the conference and journal versions of the APT intelligence study; these are related publications, not seven independent studies. For research collaboration or further bibliographic details, please use the academic email.
wilbur.wei@saturn.yzu.edu.twShare your event or team context, the problem to address and the expected timeline.
Prepare an inquiry, then copy it or choose an email option to send it.