Integrating Contextual Data for Real-World Insights in Living Labs

Research has shown that access to building occupant behavior data can reduce energy consumption and improve occupants’ productivity, comfort, and well-being. However, behaviors can vary across cultural, geographic, building, environmental, and contextual settings. Therefore, to increase our understanding of the long-term naturalistic behavior of occupants, more living labs are emerging across different countries, offering an opportunity to address existing research gaps. With the growth of IoT and ubiquitous computing, it has become easier to replicate and validate short and long-term data across different contexts. However, selecting sensors' type, quantity, and position needs to be more cohesive with building information and activity simulation to avoid inaccurate, redundant, and privacy-intrusive sensing issues. In this dissertation, we tackle these critical challenges of living lab by demonstrating: 1) a methodology for integrating building simulation models to identify optimal light sensor placements with privacy-preserving sensing considerations, 2) a longitudinal in-hospital case study that integrates medical events data and environmental sensor streams to predict momentary patient sleep disruptions and 3) a novel methodology for integrating information extracted from building plans to support fault detection of long-term energy harvesting sensor deployments. Overall, the three chapters in this dissertation demonstrate contributions to three main pillars of living labs: instrumentation, utility, and maintenance.