Empirical Bounds on the Rate of Secret Bits Extracted from Received Signal Strength

Symmetric cryptography relies on pairs of identical secret keys shared by the legitimate communicating parties. To implement a symmetric-key algorithm for cryptography, a major concern is to develop secure methods for distribution of the secret key. In securing the wireless fading channels by symmetric-key algorithms, the physical layer properties of the channel can be exploited for distribution of the secret keys. In this approach, the channel state provides a common randomness which is shared by the legitimate users but is mostly unknown to an eavesdropper. by means of signal processing techniques, this common randomness is extracted into random secret keys. This paper establishes an information theoretic upper bound on the rate at which the secret keys can be extracted. Instead of the conventional approach that relies on mathematical models for the wireless channel, this paper adopts an experimental approach to estimate this bound from empirical data. A set of signal processing techniques is developed here to numerically estimate this bound for a pair of received signal strength (RSS) recorded by indoor commercial radios.