HB+DB: Distance bounding meets human based authentication

Abstract

Authentication for resource-constrained devices is seen as one of the major challenges in current wireless communication networks. The HB + protocol by Juels and Weis provides device authentication based on the learning parity with noise (LPN) problem and is appropriate for resource-constrained devices, but it has been shown to be vulnerable to a simple man-in-the-middle attack. Subsequent work has focused on modifying the cryptographic properties of the original protocol to mitigate this problem. We propose that this attack could be mitigated using physical layer measures from distance-bounding protocols and simple modifications to devices' radio receivers. We take the HB + as a reference protocol and combine it with distance-bounding techniques. This hybrid solution, the HB + DB protocol is shown to provide resistance against the man-in-the-middle attacks on HB + as a result of the additional physical-layer mechanisms. We analyze the security of the proposed HB + DB protocol against active man-in-the-middle attacks and present experiments showing how it is practically possible to limit the success of a practical man-inthe-middle attack. We also briefly discuss the possibility that HB + DB could provide some resistance to basic threats scenarios meant to be mitigated by distance-bounding protocols. We make a practical implementation to verify that our proposed method is feasible. Finally, we discuss a proof-of-concept channel for our scheme implemented on a platform equivalent in resources to a contactless smart card/NFC device.

Breaking and fixing the HB+DB protocol

“…The HB+DB protocol . Depicted in Figure 2, the HB+DB protocol [15,16] uses two public parameters: η ∈ (0, 1/2) for LPN noise and the mean ψ ∈ [0,1/2] of a Bernoulli distribution for channel noise. The prover P and the verifier V share three k-bit keys x,y,z.…”

“…The HB+DB protocol [15,16] is essentially a white-box composition of an authentication protocol (i.e., HB + ) and a proximity-checking phase. This section shows, however, that HB+DB provides neither secure authentication, nor secure distance bounding.…”

“…The false acceptance rate (FAR) is the success probability of an adversary committing a MiM attack, MF, or DF by sending (early) random responses; the latter was indeed stated by [15,16] as the best strategy for successful DF against HB+DB. Since HB+DB uses k-bit challenges and 1-bit responses, attackers should aim to guess responses, rather than challenges.…”

“…LPN-hardness. Section 3.4 refutes the claim that HB+DB's passive-security reduces to the hardness of LPN [16]. Fixing HB+DB.…”

“…DB insecurity. In Section 3.1, we show how the LPN noise demanded by [16] yields very successful attacks, despite using many proximity-checking rounds.…”