Electromagnetic emission-aware Machine Learning enabled scheduling framework for Unmanned Aerial Vehicles

Recently, there has been a notable increase in the number of User Proximity Wireless Devices (UPWD). This growth has significantly raised users’ exposure to Electromagnetic Field (EMF), potentially leading to various physiological effects. The use of Non-Terrestrial Networks (NTN) has emerged as an optimistic solution to improve wireless coverage in rural areas. NTN mainly consist of satellites, with High Altitude Platform Stations (HAPS) and Unmanned Aerial Vehicles (UAV) considered special use cases. It is well established that optimizing exposure over time (Dose), rather than dealing with a fixed value, plays a crucial role in reducing uplink EMF exposure levels. In this paper, for the first time, we showcase that the combined use of UAV and the Dose metric can help keep the regulated uplink EMF exposure level well below the required threshold. This paper employs a combination of Non-Orthogonal Multiple Access (NOMA), UAV technology, Machine Learning (ML), and the Dose metric to optimize EMF exposure in the uplink of wireless communication systems. The ML based technique consists of a combination of k-medoids-based clustering and Silhouette analysis. To further reduce uplink EMF exposure, a power allocation policy is developed by transforming a non-convex problem into a convex one for solution. The numerical results indicate that the proposed scheme, which integrates NOMA, NTN, and ML, achieves at least a 89% reduction in EMF contrast to existing methods.