Evaluation of Temperature-Based Occupancy Detection in Residential Buildings: An Experimental Study in a Lab Environment
This master thesis aims to investigate the potential of detecting room or apartment occupancy using only indoor temperature measurements. Through an experimental campaign conducted in a controlled laboratory environment (KTH Granryd Lab), the thesis will establish a benchmark for how effective temperature-based methods are in estimating occupancy. The results will be valuable in supporting future applications of low-cost sensor networks and AI-based control strategies in residential heating and cooling systems.
Background
The integration of AI and advanced numerical methods into building energy systems is accelerating, especially in the areas of smart monitoring and adaptive control. One critical component of this is the ability to accurately detect occupancy in indoor spaces. Traditional methods rely on sensors such as CO₂ monitors, PIR sensors, and thermal cameras, which may increase system complexity and costs.
A growing research interest has emerged around using only indoor temperature data to infer occupancy levels. The key advantage is that temperature sensors are inexpensive, non-intrusive, and already widely installed in many buildings. However, the accuracy of this method remains uncertain and highly context-dependent.
This master thesis will explore this approach by conducting experiments in a climate-controlled chamber at the Granryd Lab (Department of Energy Technology, KTH). By mimicking typical residential scenarios and collecting high-resolution data across multiple sensor types, the thesis will analyze the extent to which occupancy can be reliably detected using only temperature data, and under what conditions it fails or succeeds.
Task description
The thesis work will be structured in the following phases:
- Literature review
- Survey of occupancy detection methods, with emphasis on temperature-based approaches.
- Summary of advantages, limitations, and current knowledge gaps.
- Experimental setup and data collection:
- Design of occupancy scenarios in the climate chamber simulating residential use (1 or more persons).
- Deployment of multiple sensors
- Indoor temperature sensors (multiple types and placements)
- CO₂ sensors, PIR sensors, Thermal cameras
- Execution of controlled tests under varied conditions (e.g., activity levels, ventilation).
- Data Analysis and Evaluation:
- Comparison of temperature-based occupancy detection against more established methods (e.g., CO₂ or PIR).
- Analysis of reliability, time resolution, sensitivity to noise, and sensor placement.
- Establishment of a benchmark for the use of temperature data alone.
- Final Report
Learning outcomes
Upon successful completion, the student will:
- Gain hands-on experience with experimental work in a climate laboratory
- Understand the principles of sensor-based occupancy detection and indoor environmental monitoring
- Learn to process, interpret, and evaluate multi-sensor datasets
- Develop knowledge regarding the trade-offs between sensor cost, data quality, and control precision in energy systems
Prerequisites
- Basic knowledge in building energy systems, thermodynamics, and indoor climate.
- Experience with data analysis and scripting (e.g., Python, MATLAB).
- Interest in experimental work and sensor technology.
- No Swedish language requirement.
Research Area/specialization/track
Applied Thermodynamics / Building Energy Systems / Smart Monitoring and Control
Duration
This thesis is expected to be completed over a six-month period. It includes laboratory work, data analysis, and reporting. The project can start as early as January 2026 or by agreement.
How to apply
Students interested in this project should send their CV and transcript of records to the supervisor listed below. Applications will be reviewed continuously.
Supervisors
Examiner
