Themen

Fault-Tolerance in Cloud Computing Environments  – English only

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

Alexander Heß

The QUIC Transport Protocol  – English only

QUIC is a recently released UDP-based transport protocol, which promises to improve performance of connection-oriented web-applications that currently rely on TCP.
Currently, HTTP/3 is its most prominent use case, although other network protocols like DNS or SMB are also adapting QUIC as an alternative to TCP.
The goal of this seminar is to provide technical insights into the QUIC protocol architecture and to give an overview about possible further applications.

Alexander Heß

Egalitarian consensus protocols  – English only

Classic multi-consensus protocols are used to deliver client request in a deterministic order to replicas of a replicated state machine. Typically a leader acts as a sequencer and proposes the order of requests. However, the protocol is robust to tolerate a faulty leader and elect a new one without compromising the delivery order. Egalitarian multi-consensus protocols allow multiple leaders at the same time and do not provide a total order but a partial one so that independent requests could be executed in different order or even concurrently in replicas. The student is supposed to present one of the egalitarian algorithms with a focus on the dependency tracking and its correctness so that the partial order is achieved in every replica under all circumstances.

Franz J. Hauck

Deterministic STM – English only

Software-transactional memory, is a concept to run transaction-like concurrent executions, but preserve atomicity and isolation as a subset of the well-known ACID properties. For the application in the fault-tolerance domain, deterministic execution is a necessity. For STM there are some approaches available to achieve deterministic behaviour, i.e. each execution with the same input creates the same output and state. The task of this topic is to investigate different approaches (there are only a few), collect and understand them in order to present their functionality.

Franz J. Hauck

Misbehavior Detection in VANETs – English only

Vehicular ad hoc networks (VANETs) enable vehicles to communicate with each other and with infrastructure. In this way, vehicles can send messages, e.g., that there is a traffic jam ahead, which increases safety and efficiency. For the correct and safe operation of applications, it is important that the information sent by other vehicles is trustworthy or can be verified. In this seminar you will describe several approaches to detect misbehavior of vehicles in VANETs.

Artur Hermann

Intrusion Detection Systems in In-Vehicular Networks  – English only

Modern vehicles are managed by netwrked controllers, which present an attack surface for various types of cyberattacks. In the worst case, these attacks can disable safety-critical functions and cause accidents. To detect and protect against attacks on the in-vehicle network, an Intrusion Detection System  can be implemented. In this seminar, you will explore several approaches to IDS and how they are integrated into in-vehicle networks.

Artur Hermann

Comparative Analysis of CV2X and DSRC  – English only

This seminar paper aims to delve into the key differences between two prominent vehicular communication technologies, namely CV2X (Cellular Vehicle-to-Everything) and DSRC (Dedicated Short-Range Communication). As the automotive industry embraces the era of connected vehicles, it is crucial to understand the unique characteristics, functionalities, and potential applications of these technologies. By examining the fundamental dissimilarities between CV2X and DSRC, this paper intends to shed light on their respective advantages, limitations, and implications for the future of intelligent transportation systems.

Natasa Trkulja

Trusted Execution Environments – English only

The concept of trusted execution environments (TEEs) has gained significant attention in the field of computer security. TEEs provide a secure and isolated environment within a computer system, offering protection for sensitive data and enabling the execution of trusted code. This seminar paper explores the fundamental aspects, key technologies, applications, and challenges associated with trusted execution environments.

Natasa Trkulja

Smartphone Theft Protection  – English only

The goal of this work is to detail modern smartphone theft protection mechanisms. These mechanisms should be compared. Furthermore, it should be analyzed if an attacker bypass these protections.

Leonard Bradatsch

Network Security Breaches   – English only

The goal of this seminar is the outlining of popular network security breaches (1-3 examples). Subsequently, state-of-the-art protection or detection approaches against these presented breaches should be explained.

Leonard Bradatsch

Object Storages  – English only

Object storage is a key component in modern cloud infrastructure, providing scalable and flexible data storage services. This seminar aims to introduce this storage solution, it's characteristics, advantages and disadvantages, billing methods, common use cases and usage patterns. It should also provide an overview of common object storage solutions, including commercial offerings such as AWS S3, as well as alternative open source implementations such as MinIO. Finally, the seminar should provide a perspective on object storage security, including common pitfalls illustrated by real-world incidents.

Echo Meißner

Explanation Methods for Image Classifier Decisions  – English only

Deep Neural Networks are revolutionizing various industries, with image classification being one of the most prominent applications. Despite their impressive performance, these models often operate as "black boxes". This information gap makes it difficult to understand how they reach specific decisions. This seminar aims to demystify the inner workings of image classifiers by exploring various explanation methods.

Dennis Eisermann

Reasons for Natural Adversarial Examples  – English only

In the rapidly evolving field of machine learning, the phenomenon of adversarial examples poses significant challenges. These are inputs designed to deceive models into making incorrect predictions. Surprisingly, adversarial examples can also arise naturally in real-world data without any intentional manipulation. This seminar will delve into the reasons behind the emergence of these natural adversarial examples. Understanding these reasons is crucial for developing robust AI systems.

Dennis Eisermann

Privacy Guidelines for VR Use in Education  – English only

Virtual Reality (VR) technologies are increasingly being integrated into educational settings, offering immersive learning experiences that significantly enhance student engagement and understanding. However, the use of VR in education raises critical privacy concerns that must be addressed to protect students' sensitive information. This seminar will explore the unique challenges posed by VR technology in safeguarding student privacy. We will identify various educational applications of VR, examine the types of data collected by VR devices—including biometric and spatial information—and discuss potential risks related to user identification and data misuse. The goal is to find a set of privacy guidelines tailored for educational VR use, covering aspects such as transparency, user controls, data security, and responsible biometric data handling. Initial reading list:

• ieeexplore.ieee.org/document/8651847
• iacis.org/iis/2022/2_iis_2022_185-192.pdf
• www2.itif.org/2021-ar-vr-user-privacy.pdf
• www.sciencedaily.com/releases/2024/05/240506131541.htm

Mostafa Yehia

Privacy-Preserving Learning Analytics: Challenges and Techniques  – English only

Learning analytics (LA) refers to the process of collecting, analysing, and interpreting data from educational activities. It involves using this data to understand and optimize learning and the environments in which it occurs. Privacy-preserving learning analytics (PPLA) is a field aimed at balancing the benefits of LA with the protection of individuals' privacy in educational settings. Initial reading list:

• ieeexplore.ieee.org/document/7563858
• bera-journals.onlinelibrary.wiley.com/doi/10.1111/bjet.13234
• link.springer.com/chapter/10.1007/978-981-16-0882-7_53
• dl.acm.org/doi/10.1145/2567574.2567613
• dl.acm.org/doi/10.1145/3027385.3027414

Mostafa Yehia

Post-Quantum-Cryptography – English only

Since the build of small quantum computers (QC) in combination with an quantum-algorithm which can solve the discrete logarithm problem very fast, the need for new algorithms, which cannot be broken by QC, but are still feasible with our current CPUs. Your task is to take a dive into the topic, but you can choose the focus, e.g. an overview of current algorithms, a detailed explanation how one algorithm works

Michael Wolf