High-Fidelity Propagation-Centric Digital Twin With Admissible-Space Filtering for Indoor Wireless Coverage Optimization

Abstract

Maximizing indoor wireless coverage in underground and industrial facilities is challeng-ing due to severe blockage, scattering, and highly non-uniform attenuation caused by heterogeneous materials and complex geometric layouts. In such environments, conventional statistical channel models often fail to provide reliable coverage estimates because they do not account for geometry- and material-dependent radio-frequency (RF) propagation effects. Digital twins (DTs) provide a promising alternative by enabling site-specific RF propagation modeling that captures accurate geospatial structures, material properties, and three-dimensional object placements. In this work, we present SCOPE-DT (Spatial Coverage Optimization Enabled by Digital Twin), a DT-empowered, spatially consistent, cost-effective, and computationally efficient framework for maximizing wireless coverage in challenging indoor environments, built entirely on open-source tools. SCOPE-DT integrates: (i) a high-fidelity, planning-oriented offline DT construction pipeline combining LiDAR-based spatial acquisition, Blender and Mitsuba for geometry and material modeling, and NVIDIA SionnaTM for spatially consistent RF propagation modeling; and (ii) DT-guided meta-heuristic optimization for access-point (AP) placement. To address the challenge of optimizing AP placement over geometrically irregular and non-convex indoor layouts, a multi-stage spatial filtering approach is introduced that refines the feasible search space by excluding physically infeasible candidate AP locations and suppressing artificially inflated coverage contributions from geometrically inconsistent grid cells. The proposed framework is validated using site-specific received-power measurements collected with a USRP-based experimental platform. To further assess generalizability, the spatial filtering pipeline is evaluated across nine diverse indoor layouts of varying scale and geometric complexity, spanning diverse environment types including residential, office, and industrial layouts, confirming consistent and geometry-adaptive behavior across all tested scenes. Extensive simulations and experiments demonstrate that SCOPE-DT (a) achieves superior coverage accuracy compared with standard 3GPP and COST Multi-Wall models, and (b) significantly improves both the optimality and convergence of meta-heuristic AP-placement algorithms when multi-stage spatial filtering is applied.