Toward a universal framework for wave control in next-generation extremely tunable microwave systems
The next generation of microwave systems for communications, sensing and wave-based signal processing requires extreme tunability to realize completely different transfer functions with the same system, leading to a massive system parametrization with hundreds or more of tunable elements. Such systems defy conventional design and deployment wisdom, notably because of their high-dimensional design space involving intertwined effects of tunable elements due to mutual coupling. As a concrete example, I will discuss our recent efforts to achieve agile free-form signal filtering and routing with chaotic-cavity-backed non-local programmable metasurfaces, forgoing the limitation of traditional filter synthesis to spatially disjoint resonators.
The first step toward mastering wave control in an extremely tunable microwave system consists in compactly representing and efficiently navigating its design space. I will present our recent efforts toward this goal, and some of their ramifications. Specifically, I will discuss how to formulate compact physical models and how to calibrate their parameters to describe a given experimental tunable system, even without knowing the latter’s design details and purely based on non-coherent detection. Next, I will discuss ambiguities in this parameter estimation problem, and how they can be lifted by additionally constraining the problem. This leads to the concept of the “Virtual Vector Network Analyzer” that leverages tunable elements as “virtual ports” to the system and determines the full augmented scattering matrix (comprising actual and virtual ports) free of any ambiguity, without ever inputting or outputting waves via the “virtual ports”. Finally, I show how the described model formulation and calibration efforts yield a solution to the decade-old problem of optimally focusing on a moving target inside an unknown complex medium without knowing the target’s location.
Livestream the event on Zoom (Yale login required)