Does it work?

The test method, the numbers so far, and what we don't know yet. Updated as beta drives come in.

What we claim, and what we don't

A radio scanner is easy to build and easy to oversell. The useful claim Wavelens makes is narrow: if a Bluetooth tracker or a phone is travelling with your car, the app will tell you within minutes, offline, with few false alarms. Everything else on the home page — plate readers, body cams, sensors — is a flag that says this maker's hardware is nearby, shown as Likely, never Confirmed. We don't claim to identify an officer, a camera model, or what a device is recording.

This page exists because that claim should be measured, not asserted. The numbers below start empty and fill in from real drives.

How it's measured

Each test is a logged drive with the app running as a passenger would use it: phone on the dash or in a pocket, screen off, default alert settings, own devices marked Mine first. The app's Rule hits table (More → Rule hits) records every signature that fired, how many separate passes, and how many packets each pass delivered, and the same counts go into the diagnostics report — so a drive past known hardware yields a hit rate, not a story. Three tests, repeated across testers, phones and roads:

• Planted tracker. An AirTag, SmartTag, Tile or Find My Device tag, separated from its owner, placed in or on the car before a drive of at least 15 minutes. Recorded: whether it alerted, and how many minutes after the drive began.
• Clean drive. A drive of at least 30 minutes with no planted device. Recorded: every alert, and whether it was justified (a real tracker or drone), excusable (a passenger's earbuds, a tag in an adjacent car at a long light) or a false alarm.
• Following. A second car with a known phone follows for at least 15 minutes with at least two deliberate separations. Recorded: whether the Following alert fired, when, and whether any other car was wrongly flagged.

Drone and camera hardware are scored opportunistically: a detection counts only when the thing was seen with eyes (a drone in the air; a camera pole or a body-worn camera) and the app's card is kept as evidence.

The numbers

Last updated 2026-10-21. Rows show the count of logged drives and the result across them; a dash means no logged drives yet.

Known weaknesses

• Scanner stalls. Some phones (Samsung One UI in particular) stop delivering Bluetooth results after a classic-Bluetooth inquiry. The October 2026 build bounces the scanner when that happens and no longer gives up after repeated stalls; uptime is now logged so this row above can be filled in.
• Vendor prefixes are not identities. The same radio module ships in a body cam, its charging dock and a transit bus. That is why these are capped at Likely — and why, since the October 22 build, fleet hardware (Samsara, Lytx, Zonar, transit AVL, and the Cradlepoint / Sierra / Peplink routers every fleet uses) has its own non-alerting category, and an optional second pass (Setup → Alerts; off by default since 4 November) holds a plate-reader, body-cam or sensor hit that rests on a vendor prefix alone to one soft tick and a place on the list, alerting in full only once it has been heard at the same spot on a second day. A bus is not at the same stop on two days at the moment you pass; a pole is. The residual false is a bus that shares your commute, which the known fixed logic then quiets.
• Wi-Fi is slow at speed. Android allows one Wi-Fi scan per 32 s unless scan throttling is turned off in Developer options; a hotspot passed at 50 km/h can be missed on the first pass. Routes you drive regularly fill in over repeated passes.
• Distance estimates are rough. The “~m” figure is a path-loss guess, good to about a factor of two.
• Tails are unproven. The following detector is designed to resist the car in the next lane, but its precision on real roads has not been measured yet. Its alerts are off by default for that reason.
• Oncoming vehicles at highway speed are mostly out of reach. A Bluetooth radio reaches a few hundred feet. Two cars closing at 150 mph cross that in one to two seconds — one packet at best, and the alert arrives after the car has gone. No phone app changes this, and we don't claim to. Where the physics is on your side: anything riding with you (closing speed zero), anything following you, fixed hardware you pass at surface-street speed, and anything you pass more than once. Road mode (first-packet alerts while charging and moving) and a companion radio with a directional antenna make the app hear more and sooner; they don't make an oncoming car stay in range longer.
• A camera with its radios off is invisible. Plate-reader poles talk to their makers over LTE, which no phone can fingerprint. Wavelens hears the ones that also beacon Bluetooth or Wi-Fi — battery packs, setup hotspots, accessory radios. A hardwired pole with nothing else on will not be heard by this or any phone app. The per-rule hit log in the app (More → Rule hits) is how we find out which poles do and don't.

Add a data point

Anyone running Wavelens can contribute any of the three tests above. Send the drive's diagnostics report (More → Share diagnostics; it contains no coordinates) and a line saying which test it was and what happened to wavelensapp@gmail.com. Every row on this page cites the number of drives behind it; yours moves the count.