Reverse Body-Coupled Authentication for Smart Rings
A simulation study asking whether a smart ring can be powered by the door handle it authenticates against — and what such a link would have to defend against.
Bachelor thesis · COMSOL Multiphysics 6.0 · LTspice · Python · pytest + Hypothesis
- Built a 2D axisymmetric FEM model of body–handle–ring coupling in COMSOL, extracting the full Maxwell capacitance matrix, to evaluate batteryless smart-ring operation.
- Developed a validation pipeline combining closed-form electrostatics, FEM and circuit simulation, backed by 385 automated tests and a provenance register on every constant.
- Corrected the existing channel model, changing predicted coupling by 19×, and established an exposure-limited energy bound that rules out batteryless operation by two orders of magnitude.
Read the technical analysis
- Modelled a smart ring powered by the door handle it touches — inverting the usual body-coupled link, which puts the transmitter on the wearable — and asked whether that architecture is feasible at all.
- Taught myself COMSOL, including attending COMSOL Day, and built the model as a validation ladder: closed-form electrostatics predicts the FEM before it is trusted, each FEM reproduces the previous tier to better than 1%, and the circuit simulation closes its own energy balance.
- The channel correction below is the first contribution: solving the network rather than combining reciprocals brackets effective capacitance from 6.43 fF at a 1 mm gap to 124.8 fF in galvanic contact.
- Proved batteryless operation is excluded by two orders of magnitude via an exposure-limited energy bound, independent of the coupling bracket.
- Identified and quantified an energy-domain denial-of-service threat with an attacker-optimal standoff distance — no prior report of it exists in the HBC literature reviewed.
- Enforced machine-readable provenance on every constant, keeping a register of thirty-three open unknowns, and withdrew ten of my own claims when their support failed.
Interactive · Thesis result
The same channel, read four ways. Only one of them is the network.
End-to-end coupling
6.43fF
as modelled
All four values are from the thesis. The reciprocal-sum reading was the received one; refuting it,
and explaining why it nonetheless landed close, is the first contribution of the work.