PIR vs mmWave vs Wi-Fi vs acoustic: which sensor for which room

What does each sensor actually sense?

PIR: heat moving across a lens

A passive infrared (PIR) sensor detects a change in infrared radiation from a warm body crossing its view, so it sees motion, not presence. Vendor guides list PIR's "Cost: Very affordable", and say battery units can "run on batteries for years" (Homey wiki, published 31 July 2025, updated 15 September 2025).

The catch: when a person stops moving, a PIR stops seeing them, and "will falsely detect that the room is empty" (Novelic, vendor blog, undated).

PIR is also easy to fool with heat: it "can sometimes be falsely triggered by sources of rapid heat change like direct sunlight or heating vents" (Homey wiki, published 31 July 2025, updated 15 September 2025).

mmWave radar: small movement of a body, in the dark, through curtains

A millimetre-wave radar sends a low-power radio signal and reads what bounces back. Because it reads movement rather than heat, it keeps seeing a person who is sitting or lying quietly, and it does not need light. It "can pass through many light materials such as drywall and curtains" (Homey wiki, published 31 July 2025, updated 15 September 2025).

Radar has its own blind spots. It is "immune to heat but can be triggered by other moving objects like ceiling fans or rustling curtains if not configured properly", and draws more power than PIR, so units "typically require a constant power source". "Some 24 GHz sensors support detection zones" (Homey wiki, published 31 July 2025, updated 15 September 2025), which lets you mask a fan or a curtain.

Wi-Fi sensing: ripples in the router signal

Wi-Fi sensing reads how a body disturbs the radio signal between a router and its devices (channel state information, or CSI). The appeal is that it needs no new hardware. One of Verizon's new Fios routers in the US ships with Origin Wireless "human presence detection" (MIT Technology Review, 2024). The limit is resolution: it is good at "is someone home" and weak at "which room, doing what".

Acoustic: sound above a threshold, classified by software

Acoustic monitoring puts a microphone in a resident's room, and software classifies sounds above a set level. Earzz, a UK vendor, says its system recognises footsteps, thuds, screaming and coughing, and "keeps digital records of all the sounds in each resident's room" (Earzz, vendor page). Ally Cares, another UK vendor, says its monitor detects "changes in their sleep through sound and motion during the night" (Ally Cares, vendor page). These systems are built for care-home night shifts: staff at a hub hear an alert and decide whether to go in. The vendors with published results are UK-based, so check Canadian support and where the audio is stored.

Where does each one fail?

Our summary of the sources above. Notes on fogged glass come from a forum thread, not a study.

Table sources: DigiKey Maker.io (2023), Homey wiki (published 31 July 2025, updated 15 September 2025), Novelic (vendor, undated), Home Assistant Community (2024, anecdotal), MIT Technology Review (2024). Links in the Sources list.

Which sensor for which room?

Hallway and front door. PIR or a door contact. People move through hallways, which is what PIR does well, and a door contact tells you if someone left at 2 a.m. A hallway PIR plus a bedroom radar is a common layered setup.

Bedroom at night. Radar, if the question is "is my parent in bed, and have they been still for too long". A PIR times out when someone lies down. A 2013 British Journal of Nursing paper says research into sleep in care homes "indicates that care-giving processes at night disturb vital sleep", but also found staff hesitant to rely on technology: "the personal assessment of a resident's wellbeing is perceived as best when performed by care-home staff" (Eyers, Carey-Smith, Evans and Orpwood, British Journal of Nursing, 2013). Our own view, not the paper's: a sensor that saves a needless check is worth having.

Acoustic suits a different question: can staff hear a resident call out. The trade-off is a microphone in the bedroom, with records kept, which needs a plain consent conversation.

Bathroom. Radar, placed with care. PIR is a poor fit, because people sit still in a bathroom, and cameras and microphones are unwelcome there. One Home Assistant user says an LD2410C module can "see through the 9.75 mm thick glass doors" of a shower, while an Aqara FP1 "will lose me when the doors fog up" (Home Assistant Community, 2024). Expect to tune sensitivity. The practical output is a dwell-time alert: someone has been in far longer than normal.

Living room and kitchen. Radar, for dwell time in an armchair; a PIR resets when someone sits down to watch TV. The kitchen is trickier: a range hood fan, a blind or a dog will all test your calibration.

Shared rooms. Every sensor family struggles here, and care homes are where consent is most complex.

What does the evidence say (and not say)?

Acoustic: large before/after reports, no control group

Acoustic monitoring is the only one of the four with big, published care-home numbers. A Care England report on six Quantum Care homes says hourly checks meant "nearly 94,000 physical night checks and over 8,000 staff hours" in three months. Afterwards, "routine room entries reduced by 90%". It also says: "Evidence from providers shows that modern monitoring approaches can significantly reduce both the number of falls and the escalation of incidents, reporting a 74.5% falls reduction, of which 64.5% being true falls and 9% being incidents thought to have been a fall that led to a 999 call out" (Care England, 2026). It does not attribute that figure to a named provider.

A pilot across six homes in Dorset and the BCP council area reported total falls down 49.2%, unwitnessed bedroom falls down 58.2%, hospital transfers down 79.3% and ambulance callouts down 63.7%, and "two thirds of the homes reduced scheduled night checks as a result" (BCP Council, 2026).

Both are before/after reports with no control group. The Care England report was produced with a vendor, Adaptive Care, whose own site says acoustic monitoring is "evidenced to prevent falls by 55%" (Adaptive Care vendor page, archived 13 June 2024) — a different number again. A UK conference abstract on a "living lab" evaluation says acoustic monitoring "has the potential to identify active residents and reduce the risk of falls, although it cannot offer fall prediction or prevention functionality" (IJTAHC supplement via PMC, 2025). Promising, and the best data of the four, but not proven effect sizes.

Radar and PIR: a Canadian null result, and thin evidence overall

A multiprovincial Canadian randomized controlled trial of passive remote monitoring used "motion sensors, cameras, and door alarms" with 313 patient-caregiver pairs in Ontario and Nova Scotia. In Ontario the monitored group had a 30% lower risk of moving to higher care, but the result was not statistically significant; in Nova Scotia there was no difference (Ontario HR 0.7, 95% CI 0.3 to 1.4, P=.30; Nova Scotia HR 1.1, 95% CI 0.3 to 3.7, P=.90) (JMIR Aging, 2025). A null result.

A Canadian-led scoping review of the international literature screened 486 articles and kept 14; most used motion sensors (10) and door sensors (8). Its conclusion: "More high-quality research on this topic is needed" (Canadian Journal on Aging, Read et al., 2023). We could find no radar-specific outcome trial in older adults. If a vendor tells you radar reduces falls, ask for the study.

Wi-Fi sensing: small trials, worse at home

A 2025 trial recruited 12 volunteers aged 60 and over; the paired accuracy results come from eight measurements in seven of them. Breathing-rate accuracy in homes was 89.25%, against up to 96.90% previously reported under controlled laboratory conditions. Heart-rate error averaged 9.26 bpm. The authors call the study underpowered, and note that two to five researchers were in the room during home tests, which "may have introduced some minor disruptions" (IEEE J. Translational Engineering in Health and Medicine via PMC, 2025).

As of February 2024, MIT Technology Review reported that "we have yet to see a commercially viable Wi-Fi device for tracking breathing or detecting falls". Sam Yang, who runs the sensor startup Xandar Kardian, told the magazine "the signal is just not clean enough". In a home, "Grandma falling looks a lot like a child jumping off the couch" (MIT Technology Review, 2024).

Privacy and Canadian rules in one minute

Cameras are the least accepted option. A scoping review of 34 studies raised concerns about cameras and visual data in 15 of them (44.1 per cent), and two of the 34 (5.9 per cent) "found that activity sensors were more acceptable than cameras". Sound recording appeared in 26.4 per cent of studies (Ageing & Society, 2026).

Microphones need a consent conversation, and some acoustic vendors keep sound records. The Office of the Privacy Commissioner says PIPEDA "applies to private-sector organizations across Canada that collect, use, or disclose personal information in the course of a commercial activity", and that Ontario, New Brunswick, Nova Scotia and Newfoundland and Labrador "have also adopted substantially similar legislation regarding the collection, use, and disclosure of personal health information" (PIPEDA in brief, OPC, accessed 6 September 2026). Which law reaches a particular care home depends on the province and on the operator; in Ontario, ask counsel about PHIPA (consolidation current to 1 January 2026) and the Fixing Long-Term Care Act, 2021 (consolidation current to 11 December 2025) as well. This is general information, not legal advice.

Radar itself is licence-exempt. Presence sensors fall under what ISED RSS-210 calls "field disturbance sensors", and Issue 11 (June 25, 2024) sets limits for these in several bands, including 24.075 to 24.175 GHz and 57 to 71 GHz (RSS-210, ISED Canada, 2024). Those are certification limits for the device class, not a safety approval of any product; RF exposure is a separate standard.

Health Canada decides whether software is a medical device from its intended use, judged from "the manufacturer's labelling, including instructions for use manuals, websites, promotional material" (Health Canada SaMD guidance, 2019). If a vendor's website promises fall detection or vital signs, ask whether the product is licensed as a medical device.

A simple decision checklist

1. Does the person sit or lie still in this room? If yes, PIR goes blind. Choose radar or acoustic.
2. Is a microphone acceptable to the resident and family, in writing? If no, rule out acoustic.
3. How many people use the room? Two or more, and every sensor gets fuzzy.
4. What does the vendor's website say the product does? "Detects falls" and "measures breathing" are medical claims. Ask for the evidence and the licence.

Each family has a directory page listing what is sold here, with the Canadian availability and the vendor wording we could verify: radar presence sensors, radar fall-detection sensors, Wi-Fi sensing and acoustic monitoring.