Themen

Privacy in Medical Chatbot Assistants  – English only

Medical chatbot assistants are increasingly used to provide healthcare guidance, triage symptoms, and support patient communication. However, their reliance on sensitive health data raises significant privacy concerns, including data security, patient confidentiality, and compliance with regulations like GDPR and HIPAA. This seminar will explore the privacy risks associated with medical chatbots, examining data collection practices, potential vulnerabilities, and ethical implications. We will also discuss strategies for ensuring secure data handling, user consent mechanisms, and AI-driven privacy safeguards, aiming to assess current challenges and propose guidelines for responsible implementation in healthcare settings.

Mostafa Yehia

Force/Haptic Feedback Gloves in Virtual Reality Surgical Simulation and Training  – English only

Haptic/Force feedback gloves are increasingly being used in virtual reality (VR) simulations to provide realistic tactile sensations, enhancing experience and immersion. These gloves can simulate touch, resistance, and texture, allowing users to interact with their virtual environment in a more immersive way. This seminar will explore Haptic/Force feedback gloves and their underlying technology, effectiveness in skill acquisition, and impact on surgical education. We will also examine key challenges such as hardware limitations, cost barriers, and the difficulty of replicating realistic tissue responses, aiming to assess their potential and future research directions in VR-based medical training.

Mostafa Yehia

Security Challenges and Solutions in V2X Networks  – English only

Vehicle-to-Everything (V2X) communication plays a crucial role in the operation of autonomous vehicles, enabling interaction with other vehicles and infrastructure. This seminar focuses on the security challenges in V2X networks, such as eavesdropping, message spoofing, and denial-of-service attacks. It examines various mechanisms used to ensure the authentication, integrity, and confidentiality of V2X communications.

Artur Hermann

Remote Attestation in Automotive Systems to Ensure Vehicle Security  – English only

As modern vehicles become increasingly connected, ensuring the integrity and trustworthiness of their electronic control units (ECUs) and software components is critical. This seminar paper provides an overview of remote attestation, a security mechanism that allows a verifier to check the integrity of an in-vehicle system remotely. It explores different attestation approaches, including challenge-response and runtime monitoring, and discusses their applicability in the automotive domain.

Artur Hermann

Security of User-Agent Policy Enforcement Points  – English only

This seminar topic examines software and hardware mechanisms to enhance the security of user-agent PEPs to prevent them from being bypassed and compromised. It focuses on prevention of compromise, rapid detection and response strategies.

Janek Schoffit

Incremental Policy Compilation in microsegmented Zero Trust Network Architectures  – English only

This seminar topic explores how policy deployment within microsegmented Zero Trust Architectures can be accelerated by using incremental compilation techniques to update only impacted nodes. It focuses on minimizing update cycles and enhancing adaptability in dynamic network environments.

Janek Schoffit

Accelerating PyTorch Image Models for Real-Time Automotive Applications  – English only

This seminar focuses on the comparative analysis of 3D object detection architectures for automotive applications. It examines state-of-the-art models leveraging sensors such as LiDAR, cameras, and radar, analyzing their accuracy, efficiency, and robustness. Special attention is given to trends in multi-modal sensor fusion, scalability, and performance in challenging environments.

Dennis Eisermann

Adversarial Attacks on Machine Learning Systems in Autonomous Vehicles  – English only

The seminar topic explores the vulnerabilities of machine learning models used in autonomous and connected vehicles. It focuses on how adversarial attacks, such as perturbations, evasion, and physical-world exploits, can compromise the security of a modern vehicle.

Dennis Eisermann

Explainable AI  – English only

Artificial Intelligence is transforming industries and everyday life, but its "black box" nature often leaves users and stakeholders in the dark about how decisions are made. Imagine an AI system denying a loan or diagnosing a critical health condition - how can we trust it if we can’t understand it? This seminar paper's goal is to explore Explainable AI (XAI): an emerging field focused on creating transparent, interpretable AI models without sacrificing performance. You'll delve into cutting-edge techniques like SHAP (Shapley Additive Explanations), LIME (Local Interpretable Model-Agnostic Explanations), and interpretable neural networks, and assess their real-world applications in finance, healthcare, and beyond.

Natasa Trkulja

Cybersecurity in Internet of Things  – English only

The Internet of Things (IoT) has revolutionized our world, connecting billions of devices - from smart homes to industrial systems. However, with great connectivity comes great vulnerability. This seminar paper focuses on the topic of Cybersecurity in IoT, examining the unique challenges posed by constrained devices, lack of standardization, and massive attack surfaces. You'll analyze different types of attacks on IoT systems, explore cutting-edge solutions such as lightweight encryption protocols, and propose strategies for building more resilient IoT.

Natasa Trkulja

Securing Digital Patient Data: Privacy Analysis and Protection Strategies for Electronic Health Records  – English only

Electronic Health Records (EHR) are becoming increasingly popular in healthcare systems worldwide. In Germany, the recently introduced electronic health record (“elektronische Patientenakte”, ePA) has faced significant privacy concerns. Security researchers discovered vulnerabilities that could potentially expose sensitive medical data, highlighting the balance between accessibility and data protection. This raises questions about authentication mechanisms, access control, data encryption, and secure data sharing. Analyze privacy and security challenges in electronic health records, discuss attack vectors, and present possible solutions.

Lukas Pietzschmann

Unclogging the Internet: Maximizing Flow though a Network  – English only

The Maximum Flow Problem is a key concept in network theory, focusing on how to send the maximum amount of data (e.g. packets in the internet) from a source to a destination, subject to capacity constraints on each edge. Over the years many algorithms have been developed to solve this problem efficiently, with the Ford-Fulkerson and Edmonds-Karp algorithms being two examples. Outline the Maximum Flow Problem, its theoretical and practical applications, and explore different algorithms to solve it. Master students can also try to implement one algorithm of their choice for a little demo at the end.

Lukas Pietzschmann

Platforms for analysing BFT consensus protocols  – English only

Consensus protocols are heavily used in state-machine replication and blockchain systems in order to deliver deterministic input to replicated services. These protocols are typically very complex and hard to analyse and implement. Recently researchers came up with platforms that allow to implement and analyse these protocols, e.g. the platform Bedrock. Students can start with the Bedrock paper as a central publication but should investigate what other similar platforms are available and present their capabilities.

Franz J. Hauck

Composition of replicated state-machines  – English only

State-machine replication (SMR) is a well-known means to build fault-tolerant services by replicating them. Recent work gives some ideas how to compose SMR-based services to a new SMR-based service. Based on this the problems and possible solutions of composed SMR-based services should be collected and presented. A central publication forms the basis for this topic, but we expect the student to also seek and look at related work.

Franz J. Hauck

Data Consistency in Distributed Databases  – English only

Consistency models allow to reason about the state of a distributed system, where data is stored and processed across multiple nodes. A given consistency model defines rules for the order of arrival and appearance of updates, while the strictness of the rule set always provides a trade-off between consistency guarantees and implementation effort. The goal of this seminar is to analyze different database systems and their approaches to data replication and consistency.

Alexander Heß

Fault-Tolerance in Cloud Computing Environments  – English only

Cloud computing has emerged as a cost-effective solution for acquiring scalable computing resources. Providers rely on fault-tolerance mechanisms to cope with hardware failures, software errors, or interrupted network connections in order to ensure high availability and reliability of their infrastructure. The goal of this seminar is to examine different fault-tolerance mechanisms used by established cloud computing providers such as Amazon Web Services or Microsoft Azure.

Alexander Heß