Invited Talk at University of Buffalo, NY

Dr. Arjuna Madanayake delivered an invited seminar at the University of Buffalo, NY, at the invitation of Dr. Josep Jornet. The title of the talk and abstract is given below.

Recent Progress in RF Multi-Dimensional Digital Signal Processing (MDSP

We provide a bird’s eye view of two on-going NSF Projects related to wireless communications at the FIU RAND Lab. We begin by providing a review of multi-dimensional spectral properties of propagating space-time planar waves, which are the far-field solutions to the wave equation, using Fourier and Laplace and Z-transform analysis in multi-dimensions. These mathematical properties are reviewed in such a way that it naturally leads to insightful explanations of key digital beamforming concepts/systems, such as, phased-arrays, timed-arrays, filter-sum-arrays, analog-digital-hybrid arrays, sub-arrays, side-lobes and grating lobes, for example. The talk then focuses on digital beamforming, and discusses multi-dimensional spatio-temporal circuit theory that has emerging applications in wideband fully-digital array processing.

Multi-beam beamforming is introduced and explained using properties of the spatio-temporal multi-dimensional spectrum of plane-waves, at RF, intermediate frequency (IF) and baseband, with a brief overview of some algorithms for achieving multiple simultaneous beams. The many challenges associated with wideband beamforming, such as beam squint and computational complexity, are discussed, followed by an introduction to approximate computing for digital beamforming, which leads to proposed solutions for low-squint ultra-low complexity digital array processing systems having practically acceptable RF performance suitable for low-SWaP applications.

The talk will embed engineering examples and details of real-world, fully-operational, experimental verifications, including an overview of high-speed RF digital signal processing platforms used in the RAND Lab, such as the ROACH-2 FPGA system, and the state-of-art Xilinx Zynq RF System on Chip (RF-SoC ZCU1275). The talk ends with an update on our 28 GHz fully-digital beamforming array that uses the Xilinx RF SoC as a RF DSP back-end.