Heating, ventilation and air-conditioning (HVAC) and lighting account for up to 60-70% of the energy consumed in buildings. It is therefore essential to design and operate these systems in an energy-efficient manner to meet low-energy targets. HVAC energy-demand is strongly related to the occupancy of the building due to the heat load and pollution generated by human metabolism, and their use of the building/equipment. In building management systems, current technologies such as infrared, CO2, temperature sensors, etc. fall short of providing accurate and reliable data about occupants' location, presence and actions which are essential for optimising building's performance. Next-generation buildings, however, are envisioned to be considerably more intelligent, with the ability to analyse utilisation of space, monitor occupants' comfort and building operation. To overcome these limitations, this research will develop a decentralised, data-driven and deep learning framework that can be integrated to building management systems, which is accurate, reliable, does not violate privacy and can provide the unique ability to capture valuable data about how and where occupants use a building.

Main objectives are:
1. Develop a framework that meets the above requirements for occupancy detection and prediction, satisfying conflicting, multiple, interdependent performance requirements.
2. Test and demonstrate the performance of proposed method for several building types and compare with existing methods.