Secret-Key Generation from Private Identifiers under Channel Uncertainty
Vamoua Yachongka, Rémi A. Chou
TL;DR
This work tackles secret-key generation for device authentication from physical identifiers (PUFs) under channel uncertainty with multi-antenna parties. It develops inner and outer bounds for discrete sources and provides a complete Gaussian capacity characterization, using a two-layer random-coding strategy and Fisher-information-based converses to handle secrecy and privacy leakage. For Gaussian sources, the bounds coincide, yielding an exact region and revealing how the number of antennas at the decoder and Eve affects the secret-key, storage, and privacy trade-offs; numerical results illustrate when GS or CS strategies are advantageous depending on storage constraints. Overall, the paper advances robust, CSI-uncertain secret-key authentication in IoT, offering concrete design insights for balancing key rate, leakage, and public-data storage. Key contributions include novel treatments of compound channels in a source-type model and the incorporation of privacy leakage constraints, with practical implications for PUF-based security systems.
Abstract
This study investigates secret-key generation for device authentication using physical identifiers, such as responses from physical unclonable functions (PUFs). The system includes two legitimate terminals (encoder and decoder) and an eavesdropper (Eve), each with access to different measurements of the identifier. From the device identifier, the encoder generates a secret key, which is securely stored in a private database, along with helper data that is saved in a public database accessible by the decoder for key reconstruction. Eve, who also has access to the public database, may use both her own measurements and the helper data to attempt to estimate the secret key and identifier. Our setup focuses on authentication scenarios where channel statistics are uncertain, with the involved parties employing multiple antennas to enhance signal reception. Our contributions include deriving inner and outer bounds on the optimal trade-off among secret-key, storage, and privacy-leakage rates for general discrete sources, and showing that these bounds are tight for Gaussian sources.