Always-On Sparse Event Wake-Up Detectors: A Review

Recently there has been a significant increase in demand for wireless sensor networks, as their field of application is rapidly expanding, driven mostly by the growing importance and prevalence of the Internet of Things concept. To enable practical application of wireless sensor networks, sensor nodes must be inexpensive, small, light-weight, intelligent and autonomous. This paper presents a review of low-power always-on wake-up detectors used to reduce a sensor node’s power consumption and enable continuous detection of sparse events. We describe the wake-up concept, discuss its advantages and present the wake-up detector’s power consumption, detection accuracy and false positive rate as parameters of interest. We present a state-of-the-art analysis of wake-up detectors, grouping them based on their power consumption into zero-power and near-zero-power detectors and active detectors, which we further separate based on their implementation into digital detectors and mixed-signal and analog detectors. This analysis shows state-of-the-art wake-up detectors operate with detection accuracy over 90% and a wide range of power consumptions, spanning from a few nW to few tens of
$\mu \text{W}$
, because of their diverse applications and sensor modalities. Additionally, it shows that active mixed-signal wake-up detectors are currently the most utilized implementation, with emphasis on acoustic transducer modality. It also shows potential trends for future detector design utilizing more MEMS and NEMS to further lower detector power consumption. We also show techniques and ideas not yet integrated in the wake-up concept, to potentially improve the wake-up detector concept and present the wake-up detectors’ common applications and their specifics.