Degree Projects

• Semantics-aware goal-oriented Communications
• Information freshness in distributed machine learning over wireless
• Neuromorphic computing
• Wireless Energy Harvesting Networks
• Physical layer security
• Network science

Contact: Nikolaos Pappas for more information.

Specific Projects

We will investigate efficient model synchronization mechanisms for enabling distributed data analytics/machine learning (ML) in wireless networks.

Strict latency requirements and unprecedented velocity of data are the main driving factors for the disruption of core analytics. Many services depend on near real-time decisions based on big streaming data, which render data aggregation and analytics at a central data center infeasible, due to high network delay.

Federated Learning (FL) brings the model directly to the user devices for local training, where an edge server periodically collects the trained parameters to produce an improved model and sends it back to the UEs. The communication takes place under limited resources and usually, only a small portion of the user devices can update their parameters upon each aggregation.

Thus, more efficient scheduling algorithms need to be developed to enable full implementation of FL. We target to investigate network-aware scheduling of FL model updates by learning individual network connectivity trends of the edge nodes in addition to the freshness and importance of their updates.

In this thesis, we will investigate scheduling policies by jointly considering the staleness of the received parameters and the network conditions to improve the efficiency of FL over wireless. Typically, staleness of updates impacts the convergence rates of distributed ML models such as the FL.

• Extended Reality (XR) over 5G

Page responsible: Nikolaos Pappas
Last updated: 2022-11-22