SAMPLIFI: a Secure signal Authentication scheme based on non-linear aMPLIFIcation
Abstract
As communication systems become increasingly exposed to security threats, strengthening user and device authentication requires protection mechanisms that can operate directly at the physical layer, complementing conventional cryptographic solutions. In this context, this paper introduces SAMPLIFI, a physical-layer authentication mechanism that leverages controlled amplifier-induced nonlinearities to create intrinsic ambiguity in the transmitted signal representation. By intentionally operating a radio-frequency power amplifier in the compression region, SAMPLIFI realizes a structured noninjective mapping in which distinct logical authentication states may generate the same physical waveform, thereby preventing a unique inversion of the observed signal. Authentication security is further reinforced by a seed-driven evolution of the modulation cardinality and by a pseudo-random codeword offset within a fixed-length authentication frame, so that the state governing each successful access is locally reproduced by the legitimate entities but is never transmitted over the air. We analyze passive observation followed by impersonation under black-, gray-, and white-box adversarial models. Under an ideal fresh-output model for the seed-derived variables, we show that, conditioned on the enrollment parameter, previous observations provide no predictive information about the next fresh authentication state, and we derive a closed-form finite-frame upper bound on the white-box impersonation probability. We further characterize gray-box credential uncertainty and establish that the proposed non-injective representation cannot increase the information or optimal guessing capability available to an adversary relative to a matched injective representation. The analytical results are complemented by numerical evaluations and controlled RF hardware measurements assessing authentication security, legitimate detection reliability, and practical implementation behavior.
Related
- Chaotic map based efficient anonymous authentication and key agreement scheme for VANETS
- ECPD-MIE: An adaptive entropy-controlled chaotic permutation-diffusion scheme for secure multi-image encryption of grayscale and color images
- A class of asymptotically optimal authentication codes with secrecy
- Near-Field Physical-Layer Authentication Under Impersonation Attacks
- Authentication over Arbitrarily Varying Channels with Causal Adversaries
BibTeX
@article{biagi2026samplifi,
title = {{SAMPLIFI: a Secure signal Authentication scheme based on non-linear aMPLIFIcation}},
author = {Mauro Biagi and Valeria Loscri},
journal = {IEEE Transactions on Communications},
year = {2026},
month = sep,
url = {https://hal.science/hal-05771067v1},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - JOUR
TI - SAMPLIFI: a Secure signal Authentication scheme based on non-linear aMPLIFIcation
AU - Mauro Biagi
AU - Valeria Loscri
PY - 2026
DA - 2026/09/30
JO - IEEE Transactions on Communications
UR - https://hal.science/hal-05771067v1
AB - As communication systems become increasingly exposed to security threats, strengthening user and device authentication requires protection mechanisms that can operate directly at the physical layer, complementing conventional cryptographic solutions. In this context, this paper introduces SAMPLIFI, a physical-layer authentication mechanism that leverages controlled amplifier-induced nonlinearities to create intrinsic ambiguity in the transmitted signal representation. By intentionally operating a radio-frequency power amplifier in the compression region, SAMPLIFI realizes a structured noninjective mapping in which distinct logical authentication states may generate the same physical waveform, thereby preventing a unique inversion of the observed signal. Authentication security is further reinforced by a seed-driven evolution of the modulation cardinality and by a pseudo-random codeword offset within a fixed-length authentication frame, so that the state governing each successful access is locally reproduced by the legitimate entities but is never transmitted over the air. We analyze passive observation followed by impersonation under black-, gray-, and white-box adversarial models. Under an ideal fresh-output model for the seed-derived variables, we show that, conditioned on the enrollment parameter, previous observations provide no predictive information about the next fresh authentication state, and we derive a closed-form finite-frame upper bound on the white-box impersonation probability. We further characterize gray-box credential uncertainty and establish that the proposed non-injective representation cannot increase the information or optimal guessing capability available to an adversary relative to a matched injective representation. The analytical results are complemented by numerical evaluations and controlled RF hardware measurements assessing authentication security, legitimate detection reliability, and practical implementation behavior.
SN - 0090-6778
ER -