• Adaptive and optimum secret key establishment for secure vehicular communications

      Bottarelli, Mirko; Karadimas, Petros; Epiphaniou, Gregory; Kbaier Ben Ismail, Dhouha; Maple, Carsten (Institute of Electrical and Electronics Engineers (IEEE), 2021-02-03)
      In intelligent transportation systems (ITS), communications between vehicles, i.e. vehicle-to-vehicle (V2V) communications are of greatest importance to facilitate autonomous driving. The current state-of-the-art for secure data exchange in V2V communications relies on public-key cryptography (PKC) consuming significant computational and energy resources for the encryption/decryption process and large bandwidth for the key distribution. To overcome these limitations, physical-layer security (PLS) has emerged as a lightweight solution by exploiting the physical characteristics of the V2V communication channel to generate symmetric cryptographic keys. Currently, key-generation algorithms are designed via empirical parameter settings, without resulting in optimum key-generation performance. In this paper, we devise a key-generation algorithm for PLS in V2V communications by introducing a novel channel response quantisation method that results in optimum performance via analytical parameter settings. Contrary to the current state-of-the-art, the channel responses incorporate all V2V channel attributes that contribute to temporal variability, such as three dimensional (3D) scattering and scatterers' mobility. An extra functionality, namely, Perturbe-Observe (PO), is further incorporated that enables the algorithm to adapt to the inherent non-reciprocity of the V2V channel responses at the legitimate entities. Optimum performance is evidenced via maximisation of the key bit generation rate (BGR) and key entropy (H) and minimisation of the key bit mismatch rate (BMR). A new metric is further introduced, the so-called secret-bit generation rate (SBGR), as the ratio of the number of bits which are successfully used to compose keys to the total amount of channel samples. SBGR unifies BGR and BMR and is thus maximised by the proposed algorithmic process.