Research Projects
Decentralized Federated Learning in Mobile Networks
Resolving Mobility-Induced Catastrophic Forgetting in DFL under mobile wireless networks
- Resolving catastrophic forgetting caused by mobile client departure from the neighborhood.
- Utilizing Synthetic Sample Distillation and Enriched Knowledge Distillation through Relayed Mobile Caching Mechanism.
Mobility-Assisted Decentralized Federated Learning: Convergence Analysis and A Data Driven Approach
- Investigating the impact of mobility on Decentralized Federated Learning (DFL) systems.
- Deriving onvergence analysis for DFL systems under wireless mobile networks and establishing convergence guarantees.
- Designing a mobility-aware DFL framework to combat network sparsity and data heterogeneity.
Distribution-Aware Mobility Assisted Decentralized Federated Learning
- Developing heterogeneity-aware trajectories for DFL under mobile wireless networks to resolve the performance degradation under data heterogeneity.
Deep Learning Based 2D Direction of Arrival Estimation using L-shaped Arrays
- Developed a deep learning-based method for Direction of Arrival (DOA) estimation operating based on observation matrix of transmitted signals by RF sources in a multi-source scenario, outperforming state-of-the-art methods in accuracy and robustness.