🛡️
CySecTH Rosenheim

Research

All publications can be found on ORCID: 0000-0002-5597-3913.

Prof. Dr. Kevin mayer is the head of the Security Analytics for Embedded Devices research group. The group is located at the Technische Hochschule Rosenheim and Friedrich-Alexander-Universität Erlangen-Nürnberg.

Research in Digital Vehicle Forensics investigates how data from modern and legacy vehicles can be acquired, interpreted, and used as reliable digital evidence. Vehicles increasingly act as complex cyber-physical data sources, containing traces from infotainment systems, ECUs, sensors, communication interfaces, and connected services. I focus on forensic acquisition methods, evidential reliability, interpretation challenges, and the impact that forensic procedures may have on vehicle systems and stored data.

In Embedded Forensics & Security, I study the security and forensic analysis of embedded systems used in constrained, heterogeneous, and safety-critical environments. This includes firmware analysis, hardware interfaces, memory acquisition, secure system design, and the forensic reconstruction of events on devices with limited transparency and resources. A central goal is to develop methods that make embedded evidence more understandable, reproducible, and trustworthy while accounting for the technical limitations of real-world devices.

Our work in Security Analytics focuses on detecting, explaining, and investigating security-relevant behavior in complex IT, OT, and cyber-physical infrastructures. I combine log analysis, anomaly detection, threat detection, system monitoring, and forensic reasoning to support both operational security and post-incident investigations. A particular focus lies on methods that are not only effective, but also interpretable and useful for analysts who must understand why a system behavior is suspicious.

Ongoing projects/topics

Research topics

  • The Sortitute: Embedded edge device running a pruned and quantized model to detect different types of material to sort trash
  • TinyBrains: Embedded edge device to help deaf people in ASL and GSL live translation. The installed model is pruned and quantized
  • ClickBait: Pruned and quantized model running on an embedded edge device that can detect different key’s pressed based on their sound pattern
  • Security and Forensic Analysis of the Abrites MODI Applications
  • Development of a resilient security concept for networked medical devices

Final theses

  • Formalizing Interpretability and Understandability for AI Systems in Digital Forensics: An Extension of the ExF Framework (MA)

  • Security of Apple CarPlay (MA)

  • Security implications of extracting microservices from a .NET Modulithe (MA)

  • Investigation of Various Approaches to Non-Destructive Digital Forensic Data Acquisition from a Vehicle Infotainment System (MA)

  • Security Analysis of the Communication Layer of Android Auto (BA)

  • Adaptive Bias Control for RF GaN Power Ampliers Using On-Device TinyML Inference (BA)

  • Design and Evaluation of a Snapshot-Based Method for Detecting Web Defacment Attacks (BA)

  • Real-Time In-Vehicle Cyberthreat Detection for Vehicle Fleets (BA)

  • Evaluating the Impact of Structural Pruning and Low-Bit Quantization on the Safety Alignment of SLMs (BA)

  • Automated detection of Web-Defacement-Attacks (BA)

PhD projects

  • Vehicle Security Operations Center
  • Threat Modeling for Automotive Systems