Grant: Research Tools and Instruments

The proposed experimental platform aims to optimize federated learning (FL) in wireless networks to make it practical for applications such as smart cities, healthcare, and industrial Internet of Things. FL allows devices to collaboratively train machine learning models without sharing data, but its implementation in wireless networks is challenged by bandwidth variability, latency, and energy constraints. This project addresses these challenges through adaptive optimization strategies that ensure efficient communication, energy utilization, and robust model performance under diverse conditions. The experimental platform includes edge devices, software-defined radios (SDRs), a central server dedicated to FL training, and a spectrum analyzer, allowing for realistic simulations of FL under dynamic wireless environments. The project will explore strategies like adaptive learning to respond to changing network conditions and evaluate scalability using multi-tier architectures. These efforts are designed to advance resource-efficient, scalable FL models that enhance privacy and performance in edge computing. The acquisition of specialized equipment is crucial for enabling experimental evaluations and performance optimization. The project also contributes to the training of Highly Qualified Personnel (HQP), who will gain hands-on experience in cutting-edge technologies related to wireless communication, machine learning, and edge computing. Ultimately, the research will push the boundaries of Artificial Intelligence-driven edge computing, contributing significantly to the next generation of wireless communication networks and their integration with distributed machine learning, thereby supporting a wide range of practical, privacy-sensitive applications.