Beyond Motion: Affordable Human Presence Detection for Smart Homes

Abstract

Passive infrared sensors are widely used for lighting and building automation because they are inexpensive and low power, but they detect motion rather than presence. A person who sits still, reads, works at a desk, or sleeps can therefore disappear from the sensor state even though the room is still occupied. This thesis investigates whether a low-cost, battery-powered, privacy-preserving presence detector can reduce that failure mode within the product constraints of IKEA Home Smart.

The work evaluates candidate sensing technologies against cost, power, privacy, false-positive risk, false-negative behaviour, supply maturity, and coverage. The down-selection favours a hybrid architecture that combines PIR sensing with 24\,GHz mmWave radar. PIR provides the low-power entry and motion trigger, while radar maintains the occupied state during small human movements that a PIR sensor would miss. The proof-of-concept uses an EFR32xG24 development platform, a ZRE200GE PIR sensor, and a Hi-Link HLK-LD2410S 24\,GHz radar module. It reports a fused presence state to a custom test dashboard and was compared with IKEA MYGGSPRAY and Aqara FP300 under matched indoor test conditions.

The prototype improved on the PIR-only reference in short-range desk-work scenarios. At 3\,m, it maintained presence throughout all 8-minute desk-work trials, while MYGGSPRAY produced false negatives in every trial. At 5\,m, the prototype still performed better than MYGGSPRAY but produced false negatives in 60\% of the trials. During no-movement testing at 3\,m, the prototype remained weaker than the Aqara FP300 and lost presence in some trials when breathing and small involuntary movement were the only signals available. All tested devices failed to provide reliable static-presence detection at 8\,m. The 5-hour controlled empty-room test produced no false positives, but an informal through-door observation showed that 24\,GHz radar can detect adjacent-room activity through light materials.

The results support PIR + 24\,GHz radar as a practical low-cost path for improving IKEA's current PIR-based motion sensing, but they do not close the problem of true presence detection. The implemented prototype is an activity-dependent improvement, as it handles small body movement better than PIR alone, but it is not yet reliable for breathing-only stillness, long-range static occupancy, pets, or room-confinement edge cases. For a lowest-cost product, further work should focus on radar algorithm tuning, duty-cycled power validation, production hardware, and longer real-home testing. If the product requirement is robust detection of completely still occupants, 60\,GHz radar remains the stronger technical option, but belongs in a higher-cost product tier unless component pricing changes.

Popular Abstract

Many smart home systems depend on occupancy sensors to control lighting, heating, ventilation, and other automated functions. In low-cost products, this is often done with passive infrared sensors, commonly called PIR sensors. These sensors are inexpensive and use very little power, but they only detect movement. A person who sits still at a desk, reads on a sofa, or sleeps may therefore no longer be detected, even though the room is still occupied. In practice, this can cause lights or other systems to turn off too early.

The usual solution is to keep the room marked as occupied for a fixed time after the last detected movement. This reduces the risk of turning things off too soon, but does not solve the problem completely and may also keeps systems active after a person has left the room. A better sensor should therefore distinguish more reliably between an empty room and a room with a still person in it, without using a camera or collecting identifiable information.

This thesis investigates whether such a sensor can be built within the cost, power, and privacy constraints of a mass market IKEA Home Smart product. Several candidate technologies were compared, including radar, thermal sensing, ultrasound, acoustic sensing, Wi-Fi sensing, Bluetooth sensing, and time-of-flight sensing. The evaluation led to a prototype based on a combination of PIR and 24\,GHz mmWave radar. The PIR sensor provides a low-power way to detect entry and larger movements, while the radar is used to maintain the occupied state when the person makes smaller movements that a PIR sensor may miss.

Radar sensing works differently from a camera. It sends out radio waves and measures how the reflected signal changes over time. Small movements from a person, such as posture changes or breathing, can affect the reflected signal. This makes radar a useful candidate for presence detection in private spaces, because it can provide information about movement without producing an image of the room or the people in it.

The prototype was tested against IKEA MYGGSPRAY, a PIR-based motion sensor, and Aqara FP300, a high-end commercial presence sensor using a 60 GHz mmWave radar. The results showed that the prototype improved on the PIR-only reference in short-range desk-work scenarios. At 3\,m, the prototype maintained presence throughout all 8-minute desk-work trials, while MYGGSPRAY produced false negatives in every trial. At 5\,m, the prototype still performed better than MYGGSPRAY, but its reliability decreased. In tests where the subject remained almost completely still, the prototype sometimes lost presence, while the Aqara FP300 performed better. At 8\,m, none of the tested devices provided reliable static-presence detection.

The prototype did not produce any false positives during a 5-hour controlled empty-room test. However, this test was too short to prove long-term reliability in real homes. An additional observation showed that the 24\,GHz radar could detect activity through a wooden door. This may be useful in some situations, but it can also cause a sensor to report presence in the wrong room if it reacts to movement behind a door or thin wall.

The results show that PIR combined with 24\,GHz radar is a practical way to improve low-cost motion sensing. It can reduce the common problem where a stationary person is missed by a PIR-only sensor, especially during normal seated activity. The prototype is therefore a useful step toward better presence detection, but it does not fully solve the problem. It was not reliable for all forms of stillness, longer ranges or room-confinement cases.

For a lowest-cost product, the recommended direction is to continue developing the PIR plus 24\,GHz radar architecture, with further work on radar algorithms, power consumption, production hardware, and longer tests in real homes. If the goal is robust detection of occupants who are completely still, such as a person sleeping or only breathing, 60\,GHz radar is the stronger technical option. That solution will however be more expensive and fits in a higher-cost product category.

• author: Gummesson, Elias
• supervisor: Per Andersson
• organization: Department of Electrical and Information Technology
• course: EITM01 20261
• year: 2026
• type: H2 - Master's Degree (Two Years)
• subject: Technology and Engineering
• report number: LU/LTH-EIT 2026-1174
• language: English
• id: 9239490
• date added to LUP: 2026-06-23 10:54:45
• date last changed: 2026-06-23 10:54:45