Sökning: id:"swepub:oai:DiVA.org:miun-41578" >
A2Event :
Abstract
Ämnesord
Stäng
- Internet of Things (IoT) sensors are becoming more and more intelligence and unobtrusive, surrounding us in everyday life. A key requirement for those devices is to be “always-on” performing “smart” detection to catch only the events of interest when they occur. A recent and trend is to design low power circuit to extract features close to the sensors without an analog-digital conversion. The goal is to achieve place-and-forget deployment with perpetual operation: this poses the challenge of achieving micro-watt power consumption of the always-on analog event detector and of harvesting energy from environmental sources, to supply the smart sensor. This paper presents A2Event (Analog to Event), a microwatt programmable pattern recognition circuit for analog sensors, with up to 128 simultaneous time-frequency features exploiting mixed-signal low power design, which is able to generate an event when a targeted pattern is recognized. Moreover, the paper proposes a self-sustaining always-on smart audio detector that combines A2Event with energy harvesting to detect events of interests perpetually. Experimental results show that A2Event consumes only 26.89 μW in always-on mode, during the time-frequency feature-extraction, while the whole system consumes only 63 μW during pattern recognition including the power for a commercial MEMS microphone and the energy harvesting subsystem. We demonstrate that the whole smart sensor operates perpetually when powered with a small form factor flexible photovoltaic panel in indoor lighting conditions, achieving a detection accuracy of 100% in the detection of two different audio streams. © 2019 Elsevier Inc.
Nyckelord
- Acoustic Sensors
- Always-on Sensing
- Energy Harvesting
- Event-Driven Sensors
- Low-power Design
- Self-Sustaining Sensors
- Electric power supplies to apparatus
- Internet of things
- Microphones
- Pattern recognition systems
- Photovoltaic cells
- Smart sensors
- Event-driven
- Harvesting energies
- Internet of Things (IOT)
- Photovoltaic panels
- Time frequency features
- Low power electronics
Publikations- och innehållstyp
- ref (ämneskategori)
- art (ämneskategori)
Hitta via bibliotek
Till lärosätets databas