An low-power microcontroller with accuracy-controlled signal-to-event converter for rare-event human activity-sensing applications

A specially designed event-driven sensor processor architecture is proposed to implement the low-power microcontroller for rare-event activity sensing applications. Te rare-event applications have a property of very long event-to-event distance like human-triggered events, for which the accuracy error of the signal detection is allowed. Te early evaluation for the fngerprint of the incoming sensor signal is performed by the proposed signal-to-event converter, which enables the entire post-processing unit only handles the atomic events. Te proposed architecture is implemented with additional 7500 NAND gates and 1 KB SRAM tracer in 0.18um CMOS process by adding the proposed building blocks on the commercial 8051-based microcontroller to reduce the sensor data-processing current, consuming only 20% compared the conventional discrete-time signal analysis.