Changes are needed in IoT sensor design.
By Neill Ricketts, chair of the ultra-low-power CO2 sensor developer, GSS.
Legislation is being introduced in many states to limit carbon dioxide (CO2) exposure. The reason is not because CO2 is directly toxic, but because it is a key correlative indicator for other pollutants and viruses— including COVID-19.
Here in the US, many of the states that have already implemented such legislation have focused efforts on schools, hospitals, and similar public buildings.
Legislation will affect energy usage
The legislation requires air to be refreshed regularly, with heating or cooling by HVAC systems (heating, ventilation, and air conditioning). These are already responsible for approximately 40% of the energy used in an average building, and refreshing air to increase in-building air quality will increase energy costs and pollution.
IoT systems can mitigate this through the constant monitoring of localized CO2 levels, which allows fewer, smaller, and better targeted air refreshes to be made.
IoT CO2 sensors already exist. However, there is a challenge in implementing them as an IoT device in such situations… and that challenge is power. If we were to dismantle a typical nondispersive infrared (NDIR) sensor, we’d find an incandescent bulb used to generate a broadband IR source. These use considerable amounts of energy. While this might be manageable in modern buildings where access to mains power is easier, much of the legislation enacted so far focuses on schools, which tend to be in older buildings without this access. Building of this kind therefore require the use of battery-powered IoT systems, with data transmitted over LoRaWAN or Bluetooth LE.
the duty cycle and signal processing algorithms to reduce the power consumed by orders of magnitude
Alternative sensor topologies
Swapping incandescent IR emitters for solid-state LED or MEMS emitters (both of which exist) is not enough. Doing so would lower the power consumption (and improve mean time to failure), but even with LED-based sensors, power consumption would be in the region of 5 to 150mW.
Make energy savings in CO2 sensors
It is certainly possible to meet the required power levels and deploy an IoT system that mitigates many of the unintended consequences of such legislation, but to do so we need to look holistically at the design. This means incorporating more energy-efficient components and deploying better-optimized signal processing algorithms and hardware to reduce computational load.
However, the key change lies in the duty cycle, with dedicated IoT systems needing to run far less frequently than they might in safety-critical applications. For air quality applications, this could be every 10-15 minutes or even less, entering ultra- low-energy sleep modes between samplings.
Through these techniques, sensors such as the COZIR-Blink can run at sub 30mJ per measurement, giving 100μW average power for 5-minute interval measurements. This enables sensors that are not just in the range of AA or coin cell operation, but also in the range of energy harvesting supplies, which would allow the sensors to be positioned and then forgotten about.