Source-linked AI summary
Visualizing Wireless Propagation and Polarization in Augmented Reality with ESPARGOS
Florian Euchner, Stephan ten Brink
TL;DR
Wireless channel measurements contain rich spatial, delay, and polarization information but are difficult to interpret because radio-frequency fields are invisible. The paper converts phase-coherent ESPARGOS WiFi measurements into a camera-registered augmented-reality overlay, showing angular spectrum, relative delay, and polarization. The resulting visualization makes measured propagation effects visible in the physical scene for communication, teaching, and debugging.
Problem
Coherent multi-antenna measurements reveal multidimensional radio-environment structure but are difficult to interpret through abstract plots or conventional processing.
Method
The system applies FFT-based beamspace processing, camera registration, and Jones-calculus polarization processing to phase-coherent ESPARGOS CSI.
Results
The overlay displays measured beamspace power, relative delay, and polarization in the optical scene, while preserving relative phase information across beams.
Takeaways & Limitations
The system provides a live visual interface for making angular propagation, multipath, and field-orientation effects visible in the physical coordinate frame.
Abstract
from arXiv · showhide
Wireless multipath propagation, beamforming, and polarization are central concepts in radio systems, but they are difficult to observe directly because radio-frequency fields are invisible to humans. This paper presents an augmented-reality visualization system that turns phase-coherent WiFi channel measurements from the ESPARGOS antenna array into a live camera overlay. The system estimates a two-dimensional beamspace representation, registers it with the optical camera view, and overlays the resulting angular spectrum onto the physical scene. Additional visual layers show relative path delay and polarization, the latter computed using Jones calculus from channel measurements captured from two separate antenna feeds. The result is an intuitive ``WiFi camera'' for science communication, teaching, and experimental debugging, while all displayed quantities remain directly derived from measured channel data.
I. INTRODUCTION
The paper turns phase-coherent WiFi channel measurements into a live, camera-registered visualization of wireless propagation. Its real-time pipeline exposes angular structure, delay, and polarization directly from measured CSI.
- Motivation: Coherent multi-antenna measurements reveal signal arrival directions, reflection delays, and propagation-induced polarization changes.These measurements are information-rich but usually inspected through abstract plots or downstream algorithms.
- Contribution: The system uses CSI as the basis for a live camera-registered visualization rather than only as input to an offline algorithm.
- Contribution: The overlay displays measured RF spatial information, including angular structure, delay, and polarization, at the apparent directions of sources or scatterers.
- Processing pipeline: The processing chain uses an FFT-based beamspace method because it is simple and fast enough for real-time visualization.The open-source implementation also supports spatial-spectrum estimators such as MUSIC.
A. ESPARGOS Receiver Setup
ESPARGOS is a receive-only, phase-coherent WiFi array built from synchronized subarrays with dual slant-polarized feeds. Its processing calibrates and combines CSI so both feed responses can support polarization visualization.
- Array hardware: ESPARGOS uses a 2 × 4 patch array with two switchable ±45° slant-polarized feeds per element in the 2.4 GHz WiFi band.Several subarrays can be combined into a larger rectangular aperture.
- Synchronization: Shared clock and phase-reference distribution synchronize the WiFi receivers across the array.
- CSI preparation: Raw CSI is gain-corrected, phase-calibrated across receivers, and arranged according to the configured antenna mapping.The measured frequency-domain channel H_m,n,k indexes array elements and OFDM subcarriers.
- Dual-feed acquisition: Randomly switched R/L feeds are combined from a short backlog under a quasi-static-scene assumption to recover dual-feed measurements.Packet averaging improves SNR, while iterative global phase alignment compensates unknown CFO-induced phase offsets.
- Polarization processing: The dual-feed snapshot is transformed into separated vertical and horizontal field components for polarization visualization.An empirically determined Jones matrix models feed response, cross-polarization, and antenna orientation.
B. From CSI to Angular Spectrum
The system transforms calibrated CSI into beamspace using a two-dimensional Fourier transform. The resulting beam- and frequency-dependent polarization response is averaged over subcarriers, with delay-domain sparsity accelerating computation.
- Beamspace coordinates: For a planar half-wavelength array, plane-wave arrivals produce approximately linear phase progressions across the two array axes.Beamspace coordinates represent neighboring-element phase differences, and the visible region maps to physical azimuth and elevation.
- Beamspace transformation: A two-dimensional Fourier transform converts the measured channel into the beamspace domain.The transform is evaluated on discretized beamspace coordinates using zero-padding and a two-dimensional FFT.
- Beamspace output: The FFT produces a beam- and frequency-dependent Jones vector whose vertical and horizontal responses describe the beamspace channel.
- Power estimation: Beamspace-domain power is averaged over the included subcarriers to obtain the angular-spectrum power representation.K denotes the number of subcarriers included in the average.
- Acceleration: The implementation accelerates the spatial FFT by retaining a short delay window around the synchronized first-arrival tap before transforming back to frequency.This remains faster than computing the beamspace transform for every subcarrier.
C. Camera-Registered Augmented Reality Overlay
The augmented-reality overlay maps beamspace measurements to camera pixels through angle conversion and perspective projection. Alignment offsets compensate small orientation errors between the antenna array and camera.
- Angle mapping: Beamspace coordinates are first converted into azimuth and elevation angles for camera registration.
- Camera projection: Assuming aligned boresights and negligible parallax, the angles are projected into normalized camera coordinates with a pinhole model.The projection uses the camera’s horizontal and vertical field-of-view angles.
- Calibration: Fixed azimuth and elevation offsets compensate small manual alignment errors between antenna and camera orientations.
D. Delay and Polarization Visualization
The system estimates relative delay and polarization for resolved beams, then renders these measured properties as camera-registered animated visual patterns. Polarization processing preserves relative beam phases, allowing distinct propagation components to remain distinguishable.
- Delay Visualization: Relative delay is estimated from adjacent-subcarrier phase increments under a single-delay-per-beam assumption.Delay is mapped to color hue, while beamspace power controls brightness.
- Polarization Visualization: Polarization estimation assumes that the relative vertical and horizontal field components remain approximately constant over the observed band.The polarization state is represented by a Jones vector after fixing its otherwise arbitrary phase.
- Polarization Visualization: A common phase rotation stabilizes the scene-wide animation phase across frames without changing relative phases between beams.The rendering gauge makes the animation temporally stable while preserving inter-beam phase relationships.
- Polarization Visualization: Electric-field phasor trajectories are rendered as animated dots with trailing traces, producing distinct motion patterns for different polarization states.Vertical, horizontal, slanted, circular, and elliptical polarization appear as different trajectories.
- Polarization Visualization: Out-of-phase beams with similar polarization remain distinguishable through their different starting phases.This follows from preserving relative phases between beams in the visualization.
E. Hardware-Accelerated Rendering
The rendering pipeline divides work between the CPU and GPU to combine beamspace-derived information with the camera feed and draw polarization traces.
- CPU Processing: The CPU computes low-resolution textures for beamspace power and optional delay or polarization information.These textures contain discretized P(Ψx, Ψy), τ̂(Ψx, Ψy), or η̂VH(Ψx, Ψy).
- GPU Rendering: GPU shaders interpolate camera-to-beamspace coordinates, blend texture samples with the video feed, and draw electric-field traces.A vertex shader performs interpolation, while a fragment shader handles blending and polarization rendering.
III. EXPERIMENTS AND RESULTS
Experiments use a combined 4×8 ESPARGOS array to capture and render all results live through the described visualization pipeline. The display updates continuously at a high frame rate and is presented through representative static screenshots.
- Experimental Setup: The experiments use the combined 4×8 ESPARGOS array shown in Fig. 3.All results are captured and rendered live on a laptop running the visualization pipeline.
- Live Visualization: The interactive display updates continuously at high frame rate, although the paper presents representative screenshots rather than live video.The screenshots in Fig. 4 document the live visualization operation.
A. Angular Spectrum: Beamspace and Augmented Reality
The augmented-reality overlay uses the camera-covered subset of the array’s visible beamspace region to place measured angular spectra onto the optical image. The array observes a wider angular field of view than the camera.
- Angular Spectrum and Camera Coverage: The full visible beamspace region is larger than the smaller region covered by the optical camera.The overlay therefore maps the camera-covered subset of measured angular information onto the optical scene.
B. Visualizing Delay
ESPARGOS visualizes multipath components by relative delay in a camera-registered augmented-reality display, showing when different propagation paths arrive.
- Direct and ground-reflected components appear earliest, while a building-corner reflection appears later.
- The augmented-reality demo presents these delay visualizations as representative screenshots from the live system.
C. Visualizing Polarization
ESPARGOS separates propagation components by angle and polarization while displaying measured field information in the optical scene.
- Two dipoles transmitting the same signal are resolved with different polarization and phase.
- Circular polarization rotates differently between line-of-sight and reflected paths.
- A wire rack is shown acting as a polarization-selective object.
- The pipeline overlays measured beamspace power, relative delay, and polarization onto the optical scene.