Sensor-based random number generator seeding

Abstract

Random number generators (RNGs) are the foundation of strong security and privacy measures. With an increasing number of smart devices being connected to the Internet, the demand for secure communication will only increase. An important outgrowth of Internet-connected devices is the embedding of sensors. Yet, there remains a paucity of good protocols to provide sensor-based secure RNG seeds. In their raw form, sensor data are a weak source of RNG seeds for two reasons: 1) adversarial control - a malicious party gaining control of the sensor and generating a known data sequence and 2) collinearity across sensors - inherent correlated sequences generated because sensors are embedded in the same device. We propose a new seeding technique that leverages sensor data to provide secure seeds for RNG. Given the current proliferation of sensors and Internet-connectivity on smart devices, this technique could increase cybersecurity in a variety of domains, without additional cost.