True random number generators as configware for mixed mode programmable systems on chip

The internet of things (IoT) brings security requirements to novel classes of devices, that are resource constrained and significantly specialized, so introducing new challenges in the provision of cryptographic primitives including random number generators (RNGs). Conventional solutions use pseudo-RNGs (PRNGs) or piggyback on the on-board sensors to harvest environmental entropy, leaving some security concerns open in either way. A recent trend offers the opportunity to fix this issue. In fact, the extremely various requirements of the IoT pushes manufacturers to shift the boundaries of programmability, marketing systems on chip that join micro/application processors with reconfigurable logic, routing and analog parts. This enables the development of true-RNGs (TRNGs) library parts that can be dropped in projects as firmware/configware whenever needed. As a demo, the design of a USB TRNG dongle capable of almost 400kbit/s is illustrated on a Cypress PSOC prototyping kit.