Computation with ricocheting Wi-Fi waves

Traditional digital electronic processors increasingly struggle to meet the unceasing demand for faster processing of bigger data with less energy, nowadays driven in particular by artificial intelligence: fundamental thermal limits prevent engineers from squeezing even more transistors into these units. A promising alternative paradigm is to perform analog computational operations with waves as they interact with carefully designed materials – but achieving the necessary fabrication accuracies is to date impossible.

In a paper recently accepted for publication in Physical Review X, Philipp del Hougne from Institut Langevin (CNRS & ESPCI Paris) and co-author Geoffroy Lerosey from Greenerwave (CNRS-funded start-up incubated at ESPCI Paris) now propose a counterintuitive solution. Rather than relying on intricate material designs, they introduce a scheme simply using materials that randomly scramble the waves instead, at the much lower cost of shaping the waves before they interact with the material. The authors demonstrate the ease of implementing their proposal with an experiment that resembles ricocheting Wi-Fi waves in an indoor environment. The latter’s irregular geometry completely scrambles the wave field. By placing simple reconfigurable metasurfaces in the room that allow them to tune how the waves are reflected off the walls, the authors are able to perform the desired computational operation – by having Wi-Fi waves reverberate in a seemingly arbitrary but tamed manner.

The scheme, likely in a miniaturized version using higher-frequency waves, is poised to become an integral part of more complicated wave-based computation schemes, notably high-speed, low-power implementations of artificial neural networks. Previously, Philipp del Hougne had been awarded the First Place in the Student Paper Competition at the Metamaterials 2018 Congress in Espoo (Finland) for the work now accepted for publication in Physical Review X.

Contact: Philipp del Hougne (philipp.delhougne (arobase) espci.fr)

Reference: del Hougne, P. and Lerosey, G. (2018). Leveraging Chaos for Wave-Based Analog Computation: Demonstration with Indoor Wireless Communication Signals. Physical Review X, in press.

Link: https://journals.aps.org/prx/accepted/dc07aKdcFa91ea06d2949139dac733fa62ce1c02c

Top