XR-RF Imaging Enabled by Software-Defined Metasurfaces and Machine Learning: Foundational Vision, Technologies and Challenges

TL;DR

This paper introduces iCOPYWAVES, a novel XR system leveraging programmable wireless environments and machine learning to enable low-latency, scalable RF imaging for immersive virtual experiences.

Contribution

The paper proposes a new XR approach using software-defined metasurfaces and RF wavefront copying, enabling low-latency and scalable immersive experiences.

Findings

Simulation demonstrates effective object reconstruction in XR-RF system

iCOPYWAVES achieves low-latency operation through physical layer control

Proof-of-concept validates the system architecture and workflow

Abstract

We present a new approach to Extended Reality (XR), denoted as iCOPYWAVES, which seeks to offer naturally low-latency operation and cost-effectiveness, overcoming the critical scalability issues faced by existing solutions. iCOPYWAVES is enabled by emerging PWEs, a recently proposed technology in wireless communications. Empowered by intelligent (meta)surfaces, PWEs transform the wave propagation phenomenon into a software-defined process. We leverage PWEs to i) create, and then ii) selectively copy the scattered RF wavefront of an object from one location in space to another, where a machine learning module, accelerated by FPGAs, translates it to visual input for an XR headset using PWEdriven, RF imaging principles (XR-RF). This makes for an XR system whose operation is bounded in the physical layer and, hence, has the prospects for minimal end-to-end latency. Over large distances, RF-to-fiber/fiber-to-RF is employed to provide intermediate connectivity. The paper provides a tutorial on the iCOPYWAVES system architecture and workflow. A proof-of-concept implementation via simulations is provided, demonstrating the reconstruction of challenging objects in iCOPYWAVES produced computer graphics.