Abstract
Amid worry for 5G and 6G, the layman's question arises: "How much exposure do I realistically experience when I walk down the street?" Focusing on mmWave radio-frequency electromagnetic field exposure with Distributed Massive Multiple-Input Multiple-Output (DMaMIMO) technology, a new state-of-the-art (SOTA) numerical method is proposed to enable accurate exposure assessment almost anywhere on Earth. Google Earth 3D photorealistic tiles provide high-level-of-detail and high-coverage photogrammetry. We semantically classify the meshes with an SOTA deep learning model. The path of a pedestrian is first ray-traced at 28 GHz with either 6G DMaMIMO or realistically deployed 5G antenna systems as the transmitter. The large-scale fading parameters are extracted and form the input for the QuaDRiGa tool, which finely models the small-scale fading features of the channel along the full path with an omnidirectional User Equipment (UE) as receiver. The resulting channel is used in a hybridization procedure with a Huygens' box that models the Electromagnetic Fields (EMFs) around the UE. The surface-absorbed power density (S (ab)) exposure metric is computed along the path using FDTD simulations of a realistic anatomical phantom. A case study in Helsinki finds that the cell-free MaMIMO free-space exposure range is 20 dB more uniform than collocated MaMIMO. A case study in New York City finds that users experience, on average, 20 dB higher values in exposure compared to non-users. The small-scale fading hotspot phenomenon in realistic environments is studied in detail, showing on average a 12 dB electric field increase w.r.t. the background and a specific shape with up to 3 sidelobes, which is characterized quantitatively. The S (ab) is less than 1% of the ICNIRP guidelines during all simulations at realistic Tx powers.