A preprint describes a live augmented-reality display that places measured WiFi information over a camera view. It is designed to show three parts of a wireless signal at once: beamspace power, which indicates how the measured signal is distributed across arrival angles; relative delay, which compares the timing of different paths; and polarization, represented through the electric field's behavior. In selected demonstrations, the overlay made angular propagation, multipath and field-orientation effects visible in the physical scene.
The work is a technical feasibility demonstration. Its evidence comes from selected propagation and polarization scenes in which the system overlaid measured quantities in physical coordinates. The display is therefore a radio-derived visualization of measured beamspace, delay and polarization, not a direct optical image of the radio field.
The array behind the camera view
At the receiver, the setup used a receive-only, phase-coherent ESPARGOS array operating in the 2.4 GHz WiFi band. The receiver had a 2 x 4 patch layout, and each element offered two switchable feeds with plus or minus 45-degree slant polarization. In processing, receiver gain was corrected and the phases across receivers were calibrated before the configured antenna mapping was applied.
For the demonstration rig, four ESPARGOS boards were combined into a 4 x 8 array containing 32 antenna elements, alongside a webcam. Channel-state information, or CSI, was streamed over Ethernet to a laptop, where the measured channel data were prepared for the camera-aligned overlays.
Turning radio measurements into a scene
After those corrections, the measured channel was transformed into beamspace with a two-dimensional Fourier transform. In ordinary terms, beamspace is an angular map of measured signal power: it organizes the radio measurement by the directions from which signal components arrive.
That angular map was registered to the camera view, and the system used the camera-covered subset of what the array could see. The array provided a wider angular field of view than the camera, so the radio measurement could cover directions outside the displayed image even though the overlay was restricted to the camera's visible portion.
The renderer is split between the CPU and GPU. The paper describes the display as video-like and interactive, with continuously updating overlays at a high frame rate. This presentation brings the angular, delay and polarization information into the camera scene as part of the live display.
Reflections and field rotation
One demonstration focused on multipath, the condition in which a signal reaches the receiver along more than one route. The visualization separated a direct component and a ground-reflected component from a later reflection at a building corner, with the groups separated in both angle and relative delay. Relative delay was estimated from phase increments between adjacent subcarriers, or neighboring frequency slices, under the assumption that each resolved beam is dominated by a single propagation delay.
For polarization, the two feeds, labelled R and L, were converted into separate vertical and horizontal electric-field components using an empirically determined Jones matrix, a calibration step for the conversion. The renderer displayed the result as animated electric-field phasor dots and traces, creating distinct motion patterns for several polarization states. This gave the camera overlay a way to show field orientation alongside arrival angle.
In the reported polarization scenes, the system separated propagation components by angle and polarization. It showed changes in rotation between direct line-of-sight and reflected circularly polarized paths, and a wire rack acted as a polarization-selective object. Together, those examples put several properties of the measured propagation scene into one camera-registered display.
A demonstration with clear limits
The limits of the evidence are visible in the way the results are reported. The paper presents representative screenshots from live visualization rather than a quantitative benchmark or user study. It does not report numerical registration error, angular or delay resolution, latency, frame rate, or a polarization-estimation or classification metric. The demonstrations therefore show that the pipeline can render these overlays in the selected scenarios, not how accurately or robustly it will perform across broader environments.
The calculations also rely on specific assumptions. Delay visualization treats each resolved beam as dominated by a single propagation delay, while polarization estimation assumes the relative vertical and horizontal field components remain approximately constant over the observed band. Camera registration assumes aligned array and camera boresights and ignores the small parallax between the camera and array center. Those conditions set boundaries on how the overlay should be interpreted.
The document is identified as arXiv:2608.25996v1, dated 26 August 2026, and is a preprint. The work is supported by the German Federal Ministry of Research, Technology and Space within the SENSATION project, grant no. 16KIS2532.
Paper data and sources
Original title: Visualizing Wireless Propagation and Polarization in Augmented Reality with ESPARGOS
Authors: Florian Euchner, Stephan ten Brink
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text