Oberseminar Rechnernetze und Telematik (WiSe 2024/2025)
In the oberseminar, lectures are held on selected topics, as well as final presentations of master's and bachelor's theses, as well as projects. The seminar is hybrid in room 051-02-007 as well as over this Zoom-link.(Meeting-ID: 879 6692 5056, Kenncode: WR6RriwYD)
Next oberseminar(s):
- 16.10.2024, 14:00-14:30 Steven Kirschenbauer, Master Thesis intermediate presentation
Title: Analysis of a paxcounter for non-invasive estimation of crowd density and pedestrian flow
Abstract: This thesis analyzes a Bluetooth/BLE/WiFi-probe system ("paxcounter") for modeling and estimating crowd density and pedestrian flow in public spaces. It is part of the FreiburgRESIST project, which focuses on developing resilient digital strategies for emergency situations in urban areas, with an emphasis on real-time crowd monitoring and GDPR compliance. -
16.10.2024, 15:00-15:30 Katsiaryna Mironava, Bachelor Project final presentation
Titel: Analyse und Umsetzung von OFI in Kryptografie als Vorbereitung zum Rezertifizierungsaudit des ISMS bei der JobRad GmbH -
Abstract: Im Rahmen meines Bachelorprojekts bei der JobRad GmbH lag der Fokus darauf, die Sicherheit und Effizienz der Speicherung kryptografischer Schlüssel angesichts aktueller Cyberbedrohungen zu optimieren. Zusätzlich untersuchte ich, ob die Implementierung eines Hardware-Sicherheitsmoduls (HSM) oder alternativer Module/Verfahren sinnvoll wäre. Ich prüfte die Aktualität der Richtlinie “Kryptografie”, um sicherzustellen, dass sie den aktuellen Anforderungen und Standards entspricht. Darüber hinaus wurden zusätzliche Standards und Leitfaden in diesem Bereich berücksichtigt. Anschließend erfolgte die Analyse der bestehenden Implementierung, um Abweichungen von den Anforderungen zu identifizieren. Im nächsten Schritt suchte ich nach einer angemessenen Lösung, um die Abweichungen zu beheben und die Implementierung zu optimieren. Letztendlich entschied ich mich aufgrund dieser Prüfung für eine alternative Optimierungslösung zur Speicherung kryptografischer Schlüssel, anstelle einer Integration des HSMs.
Last oberseminar(s):
- 25.09.2024, 14:00-14:45.: Ahmet Bulut, Master Thesis final presentation
Title: Efficient ZK Argument for Shuffle Implementation in RustAbstract:
Abstract: A shuffle operation in cryptography is an operation that takes a committed, anonymous series of values and returns the original series modified with a permuted order. It is an important operation in many real world scenarios (e-voting, mental card games). Due to the plaintexts or data being encrypted for privacy, the correctness of a shuffle of commitments is not straightforward to verify. While there are algorithms to construct such arguments, we are providing the first comprehensive pure Rust library for the Correctness of a Shuffle Operation.
The library is built over elliptic curve prime order groups to commit and encrypt data for privacy, while exploiting the homomorphism of the elliptic curves for efficiency.
This argument for correctness combines two separate arguments (Multi-exponentiation Argument, Product Argument) to produce a Shuffle Argument for correctness. Utilizing the Rust programming language, which has enormous support from the cryptography community, with its efficiency in runtime and security in memory, we aim to provide an extensive and easy to use zero-knowledge proof framework that can be seamlessly incorporated and used by other proof schemes, or used to construct complex arguments. - 18.09.2024 14:00-14:45: Alex Lenitschek, Bachelor Thesis final presentation
Title: Design, Implementation, Simulation and Evaluation of a Density based Clustering Algorithm for Indoor Localization based on Directed and Reflected Signals
Abstract: The Indoor Localization based on Directed and Reflected Signals (ILDARS) project aims to create a device capable of locating sound-emitting sources within unknown room environments by analysing sound emissions like direct sounds or reflected sounds and the time difference between their arrival.
This thesis focuses on a critical early stage of the ILDARS pipeline: clustering signals to determine wall positions. Specifically, a for the ILDARS adapted version of the Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) algorithm is designed and implemented to effectively cluster the input data. The motivation behind this work lies in the crucial role accurate clustering plays in the overall ILDARS pipeline, as all subsequent algorithms rely on the precision of these initial clustering results for accurate source localization.
To achieve this, the HDBSCAN algorithm is designed, implemented, simulated, and its performance across diverse virtual environments evaluated. The analysis considers various factors, such as room size, shape, and the presence of noise in the signal data, to assess the algorithm's robustness and effectiveness in clustering the data. - 18.09.2024 14:45-15:15: Peter Gillessen, Bachelor Project final presentation
Title: Investigating HTTP/3 Request Smuggling Vulnerabilities in nginx
Abstract: This project investigates potential HTTP/3 request smuggling vulnerabilities in the popular reverse-proxy nginx. Following the methodology proposed by Pisu et al., tests were conducted using an environment that forces HTTP/3 to HTTP/1.1 conversions. Out of 203 malformed HTTP/3 requests, successfully rejected or modified 201, adhering to RFC 9114. However, two requests containing disallowed ASCII characters (0x20 and 0x9) in header values were forwarded unmodified to the backend, indicating a potential risk for request smuggling. These findings highlight the need for further investigation into specific edge cases, despite the overall robustness of nginx in handling such vulnerabilities. -
11.09.2024 13:00-13:30: Emre Bayazıtoglu, Bachelor Project final presentation
Title:
Comparative Analysis of Load Balancing Algorithms in General Graphs
Abstract:In this work, we study the load balancing problem comparing the Push-Pull Sum protocol proposed in Nugroho et al. [1] to the Single Proposal Load Balancing protocol proposed in Dinitz et al. [2]. In undirected graphs, nodes can transfer loads to their neighbors, aiming to achieve a balanced state in the network. These loads may represent computational tasks related to CPU usage, memory utilization, or even internet traffic. Balancing loads and therefore challenging over- and underloading helps improve the efficacy of distributed systems and prevent system and performance errors from occurring. In cloud computing, load balancing algorithms are crucial for improving response times, ensuring system stability, and therefore contributing to customer satisfaction.
To provide a comprehensive analysis, we implement the aforementioned load balancing algorithms and evaluate their performance through simulations. Simulations are conducted using the PeerSim simulation tool, where we compare the progression of the mean squared error across multiple computation rounds. The simulations are performed on various topologies with different characteristics to identify the limitations and strengths of each algorithm.
[1]: S. Nugroho, A. Weinmann, and C. Schindelhauer, Adding Pull to Push Sum for
Approximate Data Aggregation. Springer, 2023
[2]: Y. Dinitz, S. Dolev, and M. Kumar, Local Deal-Agreement Algorithms for Load
Balancing in Dynamic General Graphs. Springer, 2022.
There are currently no items in this folder.