Wi-Fi is no longer just for internet access. A new technology uses existing Wi-Fi signals to identify and track people.
This system which was developed by the researchers at the University of California does not require any dedicated hardware on the user’s end instead of relying on standard Wi-Fi receivers.
A new system named WhoFi uses the human body to alter radio waves to identify a person with up to 95.5 percent accuracy.
It works on consumer level equipment and still works in poor light. This approach relies on radio measurements and not on cameras.
A WhoFi deployment captures Channel State Information (CSI) which is a detailed measurement of how a Wi-Fi signal is affected by its environment.
However, it turns those changes into a compact biometric identification that is unique to that individual.
The remarkable discovery comes from Danillo Avola of the Sapienza University of Rome and their team built and tested the pipeline on public data.
CSI is primarily a matrix of amplitudes across antennas and in simple terms it shows how a Wi-Fi signal travels through a space.
CSI sequences feed a deep network that learns a person's specific embedding.
The significant results came from a Transformer encoder that excels at predictable sequence patterns in the signal.
The researchers tested WhoFi’s accuracy using the NTU-Fi Human ID benchmark which contains recordings of 14 subjects.
The WhoFi system achieved a high level of accuracy with 95.5 percent rank 1 identification rate.
The teams have kept the system purely academic. They trained it with public data, documented every step, and compared encoders in a replicable way.