A Relationship between Parameters of Two Representations of Discrete LTI ARMA System Model and the Corresponding Device Fingerprints

Device fingerprints has potential to identify the source of digital communication information. Although identifying transmitters with their fingerprints remains an arduous question, fusion identification with multiple device fingerprints is a feasible methodology. In this paper, a relationship between parameters of two representations of discrete-time linear time-invariant (LTI) autoregressive moving average (ARMA) system model is derived which has definite physical meaning. With the relationship, a novel linear device fingerprints is proposed which may enlarge the inter-class distance of the transmitters to be identified. The proposed device fingerprints can be used for fusion identification of network devices and the corresponding ARMA systems especially for that with slight differences.