42 - RF Vision
Description/Abstract/Artist Statement
This project explores the combination of machine learning (ML) and a software-defined radio (SDR) to improve wireless network performance. By leveraging a ML algorithm, the system processes real-time radio frequency (RF) data, then sends messages to a Wi-Fi access point (AP) to dynamically select the most efficient transmission channel. The study tackles a key issue in modern wireless communications which is the increasing congestion in the 2.4GHz Wi-Fi band, largely driven by the growing number of Internet of Things (IoT) devices. Through continuous RF environment monitoring using an SDR, the system collects extensive spectral data to support model training, testing, and validation. An edge computing architecture enables local processing of RF data, allowing for intelligent channel selection with minimal latency. The ML model evaluates signal properties, interference patterns, and channel usage metrics to determine the optimal transmission channel. Performance assessments show significant enhancements in throughput, latency, and connection stability compared to conventional static channel allocation. The system effectively adapts to varying RF conditions, maintaining reliable connectivity even in high-density wireless environments. This work exemplifies the practical application of ML for RF analysis, demonstrating the potential for intelligent, adaptive solutions in wireless communications. Additionally, the proposed methodology establishes a framework for broader use in spectrum management, interference reduction, and autonomous radio systems.
Faculty Advisor/Mentor
Dr. Otilia Popescu, Dr. Murat Kuzlu
Faculty Advisor/Mentor Department
Electrical Engineering Technology
College Affiliation
College of Engineering & Technology (Batten)
Presentation Type
Poster
Disciplines
Digital Communications and Networking | Electrical and Electronics | Signal Processing | Software Engineering | Systems and Communications
42 - RF Vision
This project explores the combination of machine learning (ML) and a software-defined radio (SDR) to improve wireless network performance. By leveraging a ML algorithm, the system processes real-time radio frequency (RF) data, then sends messages to a Wi-Fi access point (AP) to dynamically select the most efficient transmission channel. The study tackles a key issue in modern wireless communications which is the increasing congestion in the 2.4GHz Wi-Fi band, largely driven by the growing number of Internet of Things (IoT) devices. Through continuous RF environment monitoring using an SDR, the system collects extensive spectral data to support model training, testing, and validation. An edge computing architecture enables local processing of RF data, allowing for intelligent channel selection with minimal latency. The ML model evaluates signal properties, interference patterns, and channel usage metrics to determine the optimal transmission channel. Performance assessments show significant enhancements in throughput, latency, and connection stability compared to conventional static channel allocation. The system effectively adapts to varying RF conditions, maintaining reliable connectivity even in high-density wireless environments. This work exemplifies the practical application of ML for RF analysis, demonstrating the potential for intelligent, adaptive solutions in wireless communications. Additionally, the proposed methodology establishes a framework for broader use in spectrum management, interference reduction, and autonomous radio systems.