Deploying Machine Learning in Baseband and Antenna Processing Algorithms
Virginia Tech, USA
Machine learning has become an incredibly powerful tool for synthesizing and optimizing signal processing algorithms directly from data and simulation across a wide range of vision and voice applications and more recently radio applications. In this talk we highlight key enablers, approaches, techniques and results in the area of multi-element array processing and baseband radio signal processing. We will also highlight recent and ongoing work at DeepSig on improving performance in 5G and other wireless systems through increased density, computational complexity reduction, and enhanced non-linear baseband algorithm design. Finally we will discuss ML driven object recognition and sensing approaching leveraging machine learning on single- and multi- element RF apertures and their ability to provide an unprecedented level of low-latency, broad-band, and highly-sensitive spectral awareness as an enabler to a wide variety of RF sensing and spectrum sharing systems.
- Tim O'Shea
Tim O’Shea (SM’06–M’07–SM’13) is the CTO at DeepSig and a Research Assistant Professor at Virginia Tech where he is focused on building future wireless systems for 5G and beyond by leveraging machine learning and data-centric design at the physical layer. He has led applied research in software radio, cognitive radio and security at VT, USG, Hawkeye360, and Federated Wireless for over 10 years, is a core developer of the GNU Radio project, and a senior member of the IEEE. His research and publications have focused on leveraging deep learning techniques to advance core problems in wireless communications and he is currently working on applied efforts to rapidly advance the performance, capabilities and adaptive nature of communications systems at the physical layer and under high degrees of freedom. He received his BS and MS degrees in 2007 from NC State University and PhD in EE in 2017 from Virginia Tech.