A Lockdown Technique to Prevent Machine Learning on PUFs for Lightweight Authentication

Abstract

Abstract. We present a lightweight PUF-based authentication approach that is practical in settings where a server authenticates a device, and for use cases where the number of authentications is limited over a device's lifetime. Our scheme uses a server-managed challenge/response pair (CRP) lockdown protocol: unlike prior approaches, an adaptive chosenchallenge adversary with machine learning capabilities cannot obtain new CRPs without the server's implicit permission. The adversary is faced with the problem of deriving a PUF model with a limited amount of machine learning training data. Our system-level approach allows a so-called strong PUF to be used for lightweight authentication in a manner that is heuristically secure against today's best machine learning methods through a worst-case CRP exposure algorithmic validation. We also present a degenerate instantiation using a weak PUF that is secure against computationally unrestricted adversaries, which includes any learning adversary, for practical device lifetimes and read-out rates. We validate our approach using silicon PUF data, and demonstrate the feasibility of supporting 10, 1000, and 1M authentications, including practical configurations that are not learnable with polynomial resources, e.g., the number of CRPs and the attack runtime, using recent results based on the probably-approximately-correct (PAC) complexity-theoretic framework.

PUF Based Authentication Protocol for IoT

“…The main problems were vulnerability for DoS attacks, replay attacks, impersonation attacks, and synchronization problems. In the lightweight category of proposals are the Slender PUF, [14] noise bifurcation [15] and PUF lockdown protocol [16] retained, while in the non-lightweight category only Reference protocol II-A [13] and the protocol proposed by Sadegi et al [17]. The main difference between these protocols [13][14][15][16][17] and our PUF based protocol is that these protocols take noisiness of the PUF into account, while our protocol considers the usage of a strong and controlled PUF.…”

“…[14,15] also take countermeasures to offer resistance against machine learning attacks, although they cannot be completely excluded [13]. The proposed protocol in [16] prevents an attacker from querying a token for CRPs that have not yet been disclosed during a prior protocol run. The main weakness of [17] is that it does not scale well, as the server needs to exhaustively evaluate a pseudo random function for each registered token.…”

On the Physical Security of Physically Unclonable Functions

“…In an authentication scenario, the availability of the adaptive chosen challenges and repeated measurements can be limited by a proposed server managed CRP lockdown protocol [95]. While the lockdown protocol is primarily designed to prevent repeated measurements in the case of ML attacks with noise-side channel information [10], it is effective against the photonic SCA as well.…”

Defense Mechanisms against Machine Learning Modeling Attacks on Strong Physical Unclonable Functions for IOT Authentication: A Review

“…Mukhopadhyay [14] temporarily fused the access point to CRPs to increase the complexity to build ML model but the PUFs are still not directly secured against ML-MA. Another defense mechanism related to access control is developed by Yu et al [71]. The authentication using PUFs was performed by employing lockdown which requires server's permission to obtain new CRPs.…”