Filterbank Secret Key Generation Rates in Multipath Channels

The sixth generation of wireless networks (6G) is expected to support the deployment of Internet of things (IoT) devices in massive scales. Finding lightweight and decentralized secret-key distribution primitives is therefore a challenge. Secret-key generation (SKG) from wireless channel coefficients is seen as a possible solution. It allows the extraction of secret keys using the channel randomness observed at the physical layer, without a centralized key distribution server. In this work an SKG approach suitable for wideband IoT devices is proposed. We investigate a filterbank-based SKG method, in which secret bits are generated through power measurements over different frequencies. To minimize dependencies and correlation among frequencies the quantile and the Karhunen-Loeve transforms are used. Finally, we perform a numerical evaluation of the achievable SKG rates, in the form of mutual-information (MI) estimates, using 3GPP channel models. Our numerical evaluation shows that the achievable SKG rate depends on the channel statistics, and, hence, to optimally harvest the information, devices need to be channel-aware.