Comparison Between Asymmetric and Symmetric Channel-Based Authentication for MIMO Systems

Abstract

Authentication is a key element of security, by which a receiver confirms the sender identity of a message. Typical approaches include key-based authentication at the application layer or the physical layer authentication (PLA) where a message is authentic if it appears to have gone through the legitimate channel. In both cases a source of randomness is needed, whereas for PLA the random nature of the communication channel is exploited. In this paper we compare the various approaches using in all cases the channel as a source of randomness. We consider a parallel additive white Gaussian noise (AWGN) channel model (which models for example a multiple-input multiple-output (MIMO) system with beamforming) and simple auto-regressive (AR) models for its evolution as well as the relation ofthe legitimate and attacker channel. In this setting the attacker can either predict the key used for key-based authentication or forge the channel estimated at the receiver for PLA. We compare the schemes in terms of false alarm and missed detection probability for the various approaches. The analysis includes both symmetric and asymmetric key-based authentication and the possibility of renewing the key or update the channel estimate (in PLA) to improve the authentication reliability against repeated multiple attacks.