Indoor Occupancy Estimation Based on Synergy of Physical Modeling, Environmental Data Fusion, and Machine Learning Frameworks

In a modern society, driven by both economic interests and a concern for personal privacy, indoor occupancy, either at workspace or in other type of accommodation, became a matter of stressed research interest. Though various methodologies have been developed so far, the proposed solutions of a technical problem of occupancy estimation still suffer from narrow perspectives and partial solutions, usually formulated as an optimization of a chosen mathematized problem. This paper considers the broader context of the problem, particularly its physical background, basic assumptions for the effective use of the environmental variables in occupancy estimations and the viable frameworks, either conventional or modern machine learning ones, that are to meet those assumptions. The approach is experimentally verified in a case study.