Choosing smart home sensors: types, protocols, battery life and accuracy

Choose a smart home sensor in three steps: decide what the automation needs to know (temperature, an open door, movement, someone sitting still, water on the floor, stale air), pick the protocol your controller already speaks, and only then compare models on battery life, accuracy and how they report. A sensor that measures the right thing, in the right place, over a radio your hub handles well, beats a more precise one that drops off the network.

There is no single best sensor for a home, because the trade-offs differ by type. A temperature sensor lives for years on a coin cell; a presence radar needs mains power. A cheap contact sensor is as good as an expensive one; a cheap air quality sensor may report a number that is not a concentration at all. This guide goes through the types, the protocol and battery questions, how to read a specification, and where to put each sensor.

Which sensor types are there, and what do they measure?

Sensor	What it detects	Usual power	Watch out for
Temperature and humidity	Air temperature, relative humidity	Coin cell or AAA	Placement, update interval
Contact	Door or window open or closed	Coin cell	Magnet gap on uneven frames
PIR motion	Movement of warm bodies	Coin cell or AA	Misses people sitting still
mmWave presence	Presence, including still people	Mains or USB	Needs constant power, tuning
Leak	Water across two probes	Coin cell or AAA	Must sit at the lowest point
Air quality	CO2, VOC index, particulates	Mains or USB, often	What the number actually means

Contact and leak sensors are simple switches and differ mainly in build quality, battery access and whether they also sound a local alarm. Temperature and humidity sensors differ in the measuring chip and in how often they report. Motion, presence and air quality sensors are where the measuring principle really matters.

Should you use PIR motion or mmWave presence?

A passive infrared (PIR) sensor detects changes in the infrared radiation reaching it, which happens when a warm body moves across its field of view. It is cheap, uses very little power and is reliable for “someone walked into the hallway”. Its weakness follows from the principle: a person reading on the sofa or working at a desk stops producing a change, and the lights go off after the timeout.

A millimeter-wave (mmWave) radar sensor emits a radio signal and measures the reflections, which lets it detect small movements such as breathing and therefore a person who is sitting still. Aqara’s FP2, for example, uses millimeter-wave radar, detects presence without movement and supports multiple zones in a room, and the maker describes it as a wired sensor that should be constantly powered. That is typical: radar needs continuous power, so plan for a USB supply or a wired location. Radar can also react to fans, curtains or pets, so expect to tune sensitivity and zones. Many homes combine both: a PIR for fast “turn on” triggers and a radar sensor to keep lights on while someone is in the room.

Which protocol should a sensor use?

Use the protocol your controller already handles well. For battery sensors that means a low-power mesh: Zigbee or Thread. Both let battery devices sleep most of the time and wake to report, while mains-powered devices route their traffic; Matter vs Zigbee compares the two ecosystems. Thread sensors are usually sold as Matter devices and work with any Matter controller that has a Thread border router; Zigbee sensors need a Zigbee coordinator, and before buying one for Home Assistant, check the model in the Zigbee2MQTT device list or the ZHA documentation to see which of its features are supported.

Wi-Fi suits mains-powered sensors such as radar presence or air quality monitors, where power is not a constraint. A battery Wi-Fi sensor has to rejoin the network every time it wakes up, which costs much more energy, so such sensors usually report less often or need larger batteries. Bluetooth sensors can be read by a nearby hub or by a Bluetooth proxy, but range is limited to a room or two per receiver.

What determines battery life?

A common coin cell holds little energy: Energizer’s CR2032 datasheet gives a typical capacity of 235 mAh at 3 V. A sensor that lasts years on that budget spends almost all its time asleep. Four factors decide how long it actually lasts:

Reporting frequency. Every transmission costs energy. A temperature sensor that reports every small change uses far more than one that reports on larger changes or at fixed intervals.
Radio conditions. A sensor far from its parent router retries more often and spends longer transmitting. Adding a mains-powered router nearby can extend battery life as well as reliability.
Temperature. Batteries deliver less in the cold, so outdoor and fridge sensors drain faster than the label suggests. Lithium cells tolerate it better: Energizer rates its lithium AA for operation down to -40 °C and its alkaline AA down to -18 °C.
Polling and features. A sensor with a display, a buzzer or frequent check-ins uses more than a plain one.

Manufacturer battery-life claims assume a particular reporting rate and conditions. Treat them as an upper bound and check the battery level entity in your controller after a few weeks; the trend tells you more than the box. The glossary entry on battery life covers the general arithmetic.

How do you read an accuracy specification?

Read the chip’s datasheet or product page, then remember that the finished sensor also depends on its enclosure and placement. Three distinctions matter.

Typical versus maximum. Sensirion’s SHT40, a humidity and temperature chip found in many sensors, specifies a typical accuracy of ±1.8 %RH and 0.2 °C, with a maximum humidity error of ±3.5 %RH. “Typical” describes most units; “maximum” is the bound you can rely on.

Over which range. Accuracy is usually specified over part of the operating range. CO2 sensors state accuracy as a fixed amount plus a percentage of the reading, over a stated span: Sensirion’s SCD41, a photoacoustic NDIR sensor, specifies its range as 400 to 5,000 ppm. A sensor that measures CO2 directly this way is different from one that estimates “eCO2” from other gases.

Concentration or index. Some air quality outputs are not concentrations. Sensirion’s SGP40 VOC sensor provides a VOC Index from 0 to 500 for judging indoor air quality, for example to control a fan. That index is useful for trends and automations, not for comparing your home with a limit value in ppm.

Resolution is not accuracy: a display showing 21.37 °C is not accurate to a hundredth of a degree. For automations, consistency often matters more than absolute accuracy; a sensor that reads half a degree high, but always by the same amount, controls a heater well once you know the offset.

Where should sensors go?

Placement affects readings as much as the chip does. Put temperature and humidity sensors at about seated or standing height, away from direct sun, radiators, kitchen appliances and outside walls, and give them a few minutes after mounting before trusting them. Mount PIR sensors so that people walk across their field of view rather than straight toward them, and keep them away from heaters and windows that see direct sun. Put leak sensors at the lowest point under the appliance, where water collects first. Keep CO2 sensors away from windows, vents and directly in front of people’s faces, in the room where people spend time. Fit contact sensors on the frame and the moving part with the magnet gap the maker specifies.

Where to go next

Local smart home automation covers which sensors keep working without the internet.
ZHA vs Zigbee2MQTT covers device support for Zigbee sensors in Home Assistant.
Zigbee explained covers routers, end devices and why battery sensors sleep.
Home Assistant explained covers how sensors become entities and automations.

Topics: smart home, sensors, Zigbee, Thread, battery life, Home Assistant