Oberseminar Rechnernetze und Telematik (Sommer 2025)

In the oberseminar, talks are held on selected topics, as well as final presentations of master's & bachelor's theses, and projects. The seminar takes place hybridly:

• In room 051-02-008, as well as via
• Zoom - link (Meeting-ID: 879 6692 5056, password: WR6RriwYD)

Next Oberseminar Talks:

• 23.04.2025 14:00-14:30 Marcel Schuhmacher, Bachelor Project Final Presentation
  
  Title: Analyzing the oracle ILDARS problem - A casestudy

  Abstract: This project analyzes a previously unsolved instance of the oracle ILDARS problem involving four signals with labels  ( ), (w_1), (w_2), (w_1, w_2) and two reflecting walls (w_1, w_2). We focus on determining the distance p between receiver and source, as it remains the only relevant unknown  given the line-of-sight signal. By transforming the scenario using mirroring techniques, we derive two independent expressions for a related distance d_1, each depending only on known data and p. Equating these leads to an eighth-degree polynomial in p, which in all computed cases, yields a single verifiable, positive, real solution. This demonstrates that the chosen scenario is solvable.

• 23.04.2025 15:00-15:30 Mika Wilpert, Bachelor Thesis Final Presentation
  
  Title: Optimization of Bluetooth Low Energy Distance Measurements with Machine Learning

  Abstract: More and more devices require accurate wireless distance measurements with more potential applications to come. While Ultra Wide Band (UWB), the current state-of-the-art method, reaches centimeter level accuracies, its implementation remains expensive, leading to slow adaptation across devices. Bluetooth Low Energy (BLE) based distance measurements repre sent a promising solution to this as the technology is already present on most wireless devices. However, due to noisy measurements, BLE still struggles with accuracies, often only reaching meter level precision. The aim of this work is to test whether applying Machine Learning (ML) to BLE distance measurements could reduce this noise and hence increase accuracy. To test this hypothesis, a dataset of BLE-based distance measurements was collected and used to train and optimize various neural networks (NN) based on different distance measurements techniques such as Received signal strength indicator (RSSI) and phase slope (PS) as well as the combination thereof. The results were then compared to existing PS distance calculations without ML to assess whether ML could increase accuracy. Last, the best performing model’s capacity to extrapolate to new environments was tested with datasets of measurements performed in new environments and different distances than the ones in the training dataset. It was found that the use of ML significantly increased accuracy and models generalized well to new distances and environments. The best model achieved a mean absolute error of 0.051 m in the training environment and 0.243 m in new environments. The different NNs trained also offer choices in terms of complexity versus accuracy. This opens the path to bringing UWB level accuracies to BLE distance measurements, thus broadening the application range of wireless distance measurements.

Previous Oberseminar Talks:

• 16.04.2025 14:00-14:30 David Stark, Bachelor Thesis Kickoff Presentation
  
  Title: Analysis and Optimization of DTLS for Constrained Networks

  Abstract: Energy efficiency, security and data integrity are more important today than ever before. IoT devices and constrained networks such as Low Power Wide Area Networks (LPWANs) are also playing an increasingly important role in today's world. Of course, the usual factors such as security, reliability and scalability also play a crucial role in these networks. Due to limited resources such as energy and bandwidth, these networks need to use additional protocols and techniques to be competitive. The aim of the work is to transmit data securely in a constrained network (LoRaWAN) using DTLS and a suitable IP stack. Important components such as header compression and fragmentation (SCHC) as well as handshake optimization techniques are used for further optimization. The implementation is tested and evaluated with several devices, whereby key performance metrics such as energy consumption or transmission rate are monitored and optimized. 

• 02.04.2025 14:00-14:30 Viktor Göbel., Master Thesis Kickoff Presentation
  
  Title: Implementierung eines Trusted Platform Module 2.0 zur Absicherung der Boot- und Updateprozesse eines Mikrocontrollers