Joint BS Deployment and Power Optimization for Minimum EMF Exposure with RL in Real-World Based Urban Scenario

TL;DR

The paper tackles the challenge of minimizing EMF exposure while preserving coverage in real-world urban BS deployments. It develops a high-fidelity, deterministic ray-based EMF model incorporating diffraction via a least-time SBR ray-launching algorithm that is validated with measurements. An adaptive grid refinement (AGR) scheme enables efficient area-wide exposure and coverage estimation, while a geometry-aware Nelder–Mead optimization accounts for practical network behaviors such as actual transmit power, intercell interference, CSI imperfections, and UE mobility. Compared with baselines using empirical channels, the proposed workflow yields more accurate EMF exposure estimates and provides actionable guidance for BS deployment and operation in urban environments.

Abstract

Base station (BS) deployment remains a critical task with successive wireless communication generations and increasing data rates demands, while the electromagnetic field (EMF) exposure is often underrated, yielding potential health implications. Therefore, this paper proposes a workflow that adjusts BS deployment and radiated power in a 3D urban scenario to jointly consider EMF exposure and coverage. To achieve this ambition, firstly, a novel least-time shoot-and-bounce ray (SBR) ray-launching (RL) tool is developed to improve computational efficiency, and simultaneously enhance diffraction modeling} for accurate EMF exposure calculation, validated with real-world measurements. To efficiently extend the computation across the target urban area, the adaptive grid refinement (AGR) algorithm is designed based on the spatial stability of the effective channel while accounting for BS beamforming, enabling global estimation of EMF exposure and signal coverage. Subsequently, to better represent real-world communication network behaviors, the actual maximum transmit power, intercell interference, and channel state information imperfection are incorporated on the BS side, while mobility over the EMF exposure averaging interval is captured on the user equipment side. Upon the aforementioned aspects, the coverage-guaranteed EMF exposure minimization problem is formulated in a realistic and accurate manner, and solved by a geometry-aware algorithm adapted to deterministic channel models, yielding the optimal BS deployment and power configuration. In comparison to a baseline that relies on an empirical channel model, the proposed workflow delivers more reliable estimation of EMF exposure and provides practical guidance for BS construction and operations.

Figures (6)

• Figure 1: Schematic of multipath scenario for different wave propagation phenomena with coordinates and vector definition.
• Figure 2: (a) Measurement scenario in Karlsruhe with an omni-directional Tx antenna • placed on the rooftop at a height of 19.18m, transmitting a continuous signal with $P_T = 10dBm$, $G_T = 5dBi$, at $f_0 = 2.5GHz$. The Rx is UMS400 universal monitoring system from Rohde & Schwarz rohde_schwarz with $G_R = 10dBi$ at height 2m and is located at different positions along $\rightarrow$ with 2m intervals. (b) Simulation scenario with $M_{\text{dim}} = 2.5$ billion and Rx sphere radius of 0.24m. The $d_{\text{tr,max}}$ is set to 5, and the maximum number of diffraction is 2. (c) Comparison of received power between measurement and simulation.
• Figure 3: Arieal view of 3D target area with different boundary definition. The grid center positions • form $\mathcal{U}_1$ with initial grid length $g_1$, • form the final $\mathcal{U}_0$ with minimum grid length $g_0$. The initial $\mathcal{B}^0$ consists of • and position origin is •.
• Figure 4: (a) 3D scenario as illustration of AGR algorithm in 1D case. The BS is at • with unit EIRP and the virtual UE follows . (b) Comparison for different scheme of EMF exposure estimation to $g_0$ of 1D case.
• Figure 5: (a) Simulated EMF exposure with $g_1$ and afterwards directly performs a 2D interpolation. (b) Simulated EMF exposure using Algo. \ref{['algo_2']}. (c) Simulated with finest $g_0$ as the reference simulation. All three subfigures shares one colorbar and the BS positions at • with unit EIRP. The area shown corresponds to the EMF exposure within the boundary $C_2$ depicted in Fig. \ref{['bound_def']}, while the blue areas represent buildings. The $x$ and $y$ axes represent coordinates along the west-east and north-south direction, respectively. Each grid in the subfigures corresponds to a distance of 1.25m, which is the minimum grid length $g_0$ along both the $x$ and $y$ axes.