Abstract:
Over the last few years, the sensing capabilities of devices have progressed significantly. Smarter, cheaper and more power efficient sensors are available today. Furthermore, measurement technologies that were once only used in applications with less stringent size and power requirements (e.g. automotive) are now within the reach of the IoT-enabled world. As such, the new use cases require a level of efficiency of embedded processing that goes beyond the capabilities of the current application processors (APs) or microcontrollers (MCUs), both of which usually require extensive engineering resources and time for specific application optimization. To alleviate this problem, some of the emerging device architectures now incorporate a dedicated sensor hub which is placed between the sensors' data streams and the core processing units (e.g. the AP and/or the MCU). The sensor hub takes care of the repetitive, always-on, and specific sensor tasks which are optimized for low power and low latency, thus offloading the main system controller to focus on tasks such as the graphical user interface.