Demo: A Batteryless Chemical Sensor (NES Group)

Powering massive number of IoT sensors is one of the grand challenges of the IoT revolution. As batteries are short-lived, hazardous, and costly, the future IoT sensing substrate must be powered by ambient energy. Although great advances have been made in supporting computation on batteryless devices, the sensing side has been entirely neglected. Most sensors, such as miniaturized chemical sensors, do not work on intermittent power.

Before a sensor has warmed up and is ready to sense the power is out. To solve the problem, we developed effective predictive models to estimate the measurement from a few transient samples. We would now like to build a demonstrator which runs these models locally on an embedded device and compare the prototype batteryless sensor to its continuously outlet powered version in different settings.


Download as PDF

Student Target Groups:

  • - Students in ICE and Computer Science.

Thesis Type:

  • Master Project / Master Thesis

Goal:

  • Supported by A. Gomez (see the paper reference), assemble the hardware platform and connect an SGP30 or SGP40 sensor.
  • Port the necessary drivers to communicate with the sensor over I2C. Integrate a sensor calibration routine and the predictive machine learning model to run locally and efficiently on the target hardware.
  • Extend the provided smartphone app to obtain and display measured sensor values.
  • Compare the obtained prototype performance to a continuously powered sensor in various environments..

Recommended Prior Knowledge:

  • Affinity to assemble embedded hardware and software, building demonstrators, system tests and getting things done;
  • Interest in optimizing machine learning models for operation on embedded devices;
  • Programming skills in Python and C/C++.

Start:

  • a.s.a.p.

Contact: