Bachelor's Theses

(Supervisor: Prof. Kenny Paterson, Joint Supervisor: Dr. Stefan Mangold)

Operational restrictions and limitations imposed by infrastructure and involved parties can pose challenges to reliable communication on humanitarian missions. This bachelor thesis aims to establish the foundations for a decentralized, peer-to-peer (P2P) mesh networking application, designed to improve communication reliability in such constrained environments. The application builds upon an analysis of existing solutions, such as Berty and Briar, to identify effective techniques and respective bottlenecks. By integrating these insights, the main functionalities of the application are developed. After having the core functionalities in place, this project focuses on the development and optimisation of Delay-Tolerant Networking (DTN) algorithms. This allows for asynchronous, store-and-forward messaging. The feasibility and effectiveness of proposed solutions are then assessed with a controlled testbed.

(Supervisor: Prof. Kenny Paterson, Joint Supervisor: Dr. Stefan Mangold)

This Bachelor's project propels the enhancement of a mesh networking prototype app, by incorporating group, broadcast, and unicast communication. Based on previously determined technical requirements, the project endeavors to analyze, design, and potentially modify elements of existing open-source messengers. The app leverages a decentralized, peer-to-peer (P2P) protocol. The addition of group and broadcast communication elevates the data transfer complexity, requiring improved routing mechanisms to address new scenarios. The solution aims for robustness, ensuring stable functionality despite interference and the presence of competing networks. Within this project, existing solutions are identified, analyzed, and documented. Performance metrics are established to facilitate the design and development of the enhanced application for both iOS and Android platforms. Furthermore, the project explores the possibilities and constraints of cross-platform communication. Communication performance is thoroughly analyzed and optimized wherever feasible, utilizing the available testbed with multiple devices. By addressing these complexities, the project aims to deliver a robust and reliable communication solution that meets the needs of users in various environments.

 

(Supervisor: Prof. Kenny Paterson, Joint Supervisors: Laura Hetz, Dr. Christian Weinert (Royal Holloway))

Cuckoo hash tables and probabilistic data structures like Cuckoo filters are crucial in cryptographic protocols addressing the private set intersection (PSI) problem, yet standard insertion methods can lead to inefficient restructuring or failure, especially amplifying communication costs to gigabyte levels in PSI contexts. This thesis aims to create new insertion strategies for these data structures to reduce communication costs while maintaining algorithmic efficiency. The strategies focus on minimizing the initial size of the hash table or filter, reducing relocations when adding items, and lowering failure or rehashing probabilities. These strategies will be theoretically analyzed and empirically tested to assess their performance, including runtime and memory overhead.