Agile Free-Form Signal Filtering with a Chaotic-Cavity-Backed Non-Local Programmable Metasurface

Abstract: Filter synthesis is an inverse problem that is traditionally approached rationally by considering spatially disjoint resonators, approximating them as lumped elements, and engineering the coupling of selected pairs. This approach strongly limits the design space, making it challenging to build extremely tunable filters. Here, we demonstrate agile free-form signal filtering with an alternative purely-optimization-based design paradigm using a programmable system with many spatially overlapping modes. We back a programmable metallic metasurface with a quasi-2D chaotic cavity, inducing strong non-local interactions between all meta-elements and the connected ports. Thereby, the metasurface efficiently controls the transfer function between the ports. Our all-metallic device has unique advantages: ultra-wideband (UWB) tunability (7.5-13.5GHz), low loss, compactness, guaranteed linearity under high signal-power levels. First, we experimentally confirm theoretical predictions about reflectionless and transmissionless scattering modes; we also experimentally observe transmissionless exceptional points. Second, we impose diverse types of transfer function zeros at desired frequencies within an UWB range. Third, we achieve low-loss reflectionless programmable signal routing. Fourth, we investigate the trade-off between routing fidelity and bandwidth, achieving 20dB discrimination over 10MHz bandwidth. Fifth, we demonstrate UWB tunable multi-band filters that reject (<-24dB) or pass (>-1dB) signals in specified bands whose centers, widths and number are reprogrammable.