Understanding how people actually move through a building and where they stay is essential for improving architectural design and operation. Yet indoor environments are difficult to observe continuously because GPS is unreliable indoors and high-precision sensing systems are often expensive. This research explores how human behavior in architectural space can be estimated from BLE beacon data, which are relatively easy to deploy in real settings.

　The data consist of sparse and noisy beacon reaction histories collected from rooms and circulation areas, which means that people’s movement cannot be reconstructed directly from the raw observations alone. We therefore propose a method that incorporates building geometry and the continuity of movement to recover meter-scale trajectories from coarse room-level data. Using the estimated trajectories, it becomes possible to analyze which parts of a building are heavily used, where people move quickly, and where they slow down or linger.

　When applied to large-scale data collected from real buildings, the method revealed usage patterns that reflect building composition and room arrangement, as well as differences in walking speed between open and enclosed areas. In other words, it becomes possible to read the relationship between architectural space and human behavior directly from data rather than only from plans. This research transforms noisy indoor sensor data into a useful basis for evaluating and improving architecture through actual patterns of use.

For more details: Peer-reviewed international conference paper