Themen

✘ Practical Consensus in Real World Systems – English only

In many systems a shared source of truth is required for the system to function properly. In many such situations a consensus is formed within a group through different approaches: Complex algorithms such as Paxos and Raft, Proof-of-Work mechanisms or others. There is a vast amount of literature for consensus in distributed systems, but not everything makes it into real world systems.

David Mödinger

✘ Mechanical Sympathy and performance optimization in high level languages – English only

'Mechanical Sympathy', a term coined by a world-class Formula 1 driver, describes the concept of driver and machine working together in harmony.  This term can also be transferred to (software) engineers, who might not need to know how to build hardware from the scratch, but who can still gain performance advantages by having intimate knowledge of the inner workings of underlying hardware. By paying close attention to the pitfalls of modern CPU, GPU and memory hardware designs, developers who know how these components work can highly optimize their applications even in high-level languages like Java. This work should take a look at recent developments in the field of software optimization and summarize advances made in the last few years. 

✘ Individual Mobile Sensing in Healthcare - Trends and State-of-the-art – English only

Individual Mobile Sensing or Personal Mobile Sensing often focuses on the aspects of sensor data collection and analysis applied specifically for the pruposes of education, diagnosis, event detection, treatment, or monitoring. The aim of this seminar is to study recent trends and review state-of-the-art already reported in existing literature, specifically applied in the domain of healthcare.

✘ Mobile Sensing Systems - Technical barriers in context of Individual Sensing – English only

Many mainstream smart phones are equipped with sensors which enable sensing motion, environment, and position. However, these devices are not ideal for continous sensing, data accumulation, data processing, etc. mainly because of their limited resources. The aim of this seminar is to review scientific literature in order to better understand the technical barriers or limitations within the context of individual sensing.

✘ π-calculus – English only

As a process calculus the p-calculus describes the movement of a piece of data in exactly the same way as the tranfer of a message. Thus the p-calculus is used to define concurrent computations whose network configuration may change during the computation. The goal of this seminar is to look at motivation, theory and expressiveness of pi-calculus.

✘  Analysis of Distributed Denial of Service Attacks – English only

Distributed Denial of Service Attacks are a common problem in networks all over the world. They are hard to defend against as they affect services that usually are accessible to the public and exploit usual, standard conform behaviour. There are many different kinds of attacks. From the usual flooding attacks over slow HTTP attacks to reflective attacks, there are many ways a system can be taken down. Attackers mimic the behaviour of legitimate clients and thus make it hard to be identified as perpetrators. This seminar shall analyse the different kinds of attacks, how attackers can be identified, and how attacks can be mitigated.

✘ Network Function Virtualization – English only

Network function virtualization (NFV) is a very trendy new paradigm in computer network research. The idea of NFV is to transfer the functionality of network nodes (e.g. switches, routers, firewalls, load balancers, intrusion detection systems) to virtual machines. The benefit of this approach is said to be vendor independent solutions, increased flexibility, and lower power consumption compared. This seminar shall analyse the current state of research in NFV and assess future prospects of the new paradigm.

✘ Machine Learning Privacy – English only

Machine Learning has a number of very useful applications and offers great benefits. Machine learning algorithms can be used for recommendation systems, data analysis, or security applications. However, while machine learning can provide useful predictions and analysis, there are also privacy concerns. The aim of this seminar is to identify privacy issues, and survey possible solutions.

✘ Computer Game Security – English only

Video games have always been the kind of software that pushed the limits of existing programming platforms. Today, with the very popular online multiplayer games, thousands of players can share their experience with others online. But what about the security of video games? Can it keep up with cheating, fraud, and other malicious attacks? The goal of this seminar is to have a look at the motives and kinds of cheating, as well as to survey state-of-the-art protection mechanisms and how they are applied in computer games. 

✘ Recent Advances in Autonomous Driving – English only

Autonomous Driving as well as other AI-based applications have the potential to shape our future. The development of advanced and efficient machine learning algorithms made Autonomous Driving possible nowadays. In this seminar, you will address autonomous driving from both points of view: the theoretical (based on the literature) as well as the practical one. In particular, you will survey and compare the existing work in this field and try to answer the question which machine learning technique performs best in Autonomous Driving. Moreover, you need to address (based on the literature) the question of how V2X communication can enhance the autonomous driving experience.

✘ Trusted Execution Environment (TEE) – English only

Trusted Execution Environment (TEE) has been considered heavily in the last decade. The goal is to provide secure cloud computing as well as privacy-friendly data processing. The idea is summarized in introducing a secure enclave in a form of, e.g., a sandbox that is isolated from the normal space of the CPU. Intel has introduced its implementation of TEE as an extension to the 7th generation processors, namely Software Guard Extension (SGX). On the other hand, ARM has introduced TrustZone as an implementation of TEE. In this seminar, you will investigate TEE and its commercial implementations, the features of each, while tackling the problem of the shortcomings of using TEE-based implementation, i.e. cache attacks, relatively small TCBs, etc. Special focus will be on the privacy-friendly application development using TEE-based approaches.

✘ Range Proofs – English only

Confidential transactions in crypto currencies require range proofs to detect integer overflows. Any output amount of a transaction has to be a positive integer. As storage is valuable on block-chains, the goal is to reduce the size as much as possible. Recent advances in bulletproofs reduce the size significantly. The paper should compare the different existing methods and point out how the improvements are achieved.

✘ Chameleon Signatures – English only

Common signatures require a private key and a public key and provide authenticity for the signed data. The signed data can not be changed, without resulting in a new signature. Chameleon signatures extend this structure and allow the owner of a secret to sign alternative data keeping the same signature. This could be useful to change illicit transactions on a ledger, where the signature output is immutable. The paper should give an in-depth explanation of the internal mechanisms of chameleon signatures.

✘ Machine Learning on Event Streams – English only

Machine Learning approaches and algorithms can be utilized to analyze event streams and time-series data and provide forecasts. This has shown very promising results in the domain of predictive maintenance and stock market prediction. However, current research is not limited to these domains. The goal of this seminar is to give an overview of current approaches, algorithms, possibilities, and limitations of machine learning on event streams and time-series data.

✘ Applications of data provenance and data lineage – English only

Modern systems often have a requirement for data provenance and data lineage, which is the notion of retaining the origin of data, how it changes, and where it moves over time. This trend can be observed in various domains, such as version control systems to track changes in source codes, versioning file systems, journaling file systems, and systems that apply event sourcing. The goal of this seminar is to give an overview of approaches for data lineage/provenance and their current fields of application.

✘ Securing Cooperative Adaptive Cruise Control – English only

Cooperative Adaptive Cruise Control (CACC) is an application of vehicular ad-hoc networks that essentially tries to improve highway throughput by enabling higher speeds and smaller safety distances. This is done through a combination of algorithms for self-driving vehicles, sensors, and vehicle-to-vehicle communication. A malicious actor may attempt to abuse this functionality to cause accidents, either by transmitting falsified messages, disrupting the communication between specific vehicles, or by distributing malware to cause vehicles to behave differently. In this seminar topic, the central question is whether CACC can be done in a secure way. To answer this question, you will describe the different attack strategies, and review some approaches to detect and prevent such misbehavior.

✘ Securing Smartphones – English only

In this day and age, almost everyone owns a smartphone and takes it with them wherever they go. These devices contain a lot of personal data; thus, securing these devices is very important.

The goal of this seminar is to give an overview of the security architectures and mechanisms implemented in modern smartphones (for example based on iOS and/or Android) and to research solutions and proposals from academic literature.

✘ TLS - a history of failures? – English only

Transport Layer Security and its predecessor - the Secure Socket Layer - is a fundamental security technology of the modern Internet. It secures almost everything starting from your social media account, over your private emails to your online banking. Yet, the protocol itself and its most widely used implementations have a long history of problems. Some are related to the way cryptography is applied in TLS, others arise from the compatibility that TLS aims to achieve (downgrade attacks).
In your presentation and paper, you should provide an overview of the major attacks on TLS and its common implementations and cluster them into categories. What major lessons can we learn for other cryptographic protocols?

✘ Spectre, Meltdown & Co - How your processor cache leaks sensitive information – English only

In Bioinformatics, the search for Motifs discovers specific but probably unknown sequences of DNA. This concept is potentially also applicable to other kinds of sequences. This topic is intended to determine whether Motif finding algorithms are applicable to network protocols. The goal of such an application would be to identify distinctive substrings of messages. Moreover, using Motifs, format and type of unknown messages could be identified.

In particular, this topic should answer based on the literature: how motif finding works, how the known approaches differ, and which, if any, of the concepts could work to detect structure also in unknown network protocols.

✘ Slide attacks – English only

Moderne symmetrische Blockverschlüsselungen wie AES oder Blowfish basieren auf der vielfachen Wiederholung einer einfachen Rundenfunktion. Da differentielle und lineare Kryptoanalyse sich auf die Fortpflanzung der Zufälligkeit über die Runden kozentrieren war es lange Zeit selbstverständlich, dass jede Verschlüsselung mit genügend Runden sicher gemacht werden kann. Slide attacks haben diesem Denken ein Ende gesetzt, da sie unabhängig von der Rundenzahl durchführbar sind und die Rundenfunktion selbst angreifen. In dieser Arbeit sollen Slide attacks beschrieben werden, sowie die nötigen kryptographischen Grundlagen erarbeitet werden.