pipette
ENEnglish

Digital Twin-Conditioned RSS-Fingerprint Localization in Cell-Free Millimeter-Wave Networks

Varun S. Advani, Khalid Kanaan, Ahmed Nasser, Mohamed Y. Selim, Ahmed M. Eltawil

PreprintUso en el mundo real

En palabras de los autores

Digital twins (DTs) can generate dense synthetic fingerprints for localization via ray tracing, yet the sim-to-real domain gap limits their direct use, particularly in cell-free millimeter-wave (mmWave) deployments where distributed geometries and blockage-induced sparsity widen the mismatch. We propose DT-conditioned (DTCond), a framework that fuses real and DT beam-domain fingerprints as complementary inputs through per-domain encoders, learnable per-access point (AP) attention, and calibrated uncertainty. We derive oracle and pipeline Cram\'er-Rao bounds (CRBs) for DT-aided fingerprint localization, proving that DT information strictly tightens the positioning bound whenever spatial structure is preserved, with a heteroscedastic refinement that tightens the bound by a further 9.1%. On a real 28 GHz cell-free testbed, the deployable two-stage pipeline, which reads the DT fingerprint at a coarse k-nearest neighbor (k-NN) estimate, achieves 0.284 m mean error against 0.313 m for real-only k-NN. With the DT fingerprint read at the true position (oracle input), DT conditioning achieves 0.186 m mean error, validated through a ten-method ablation against domain-adaptation (DA) and transfer-learning baselines, and a 48% improvement on unseen regions. Multi-user self-attention provides 31-41% per-user gain for emulated simultaneous users, and progressive calibration reduces the measurement budget by 50%. As a use case, localization drives network configuration in closed loop. Between beam sweeps, a DT updated from measured RSS and read at the tracked position estimate selects the 3 of 81 beam pairs that each AP trains. With 49 calibration anchors in the update, the mean beam-selection loss stays below 1 dB over 2 m of motion, whereas reusing the last sweep loses 2.7 dB after 1 m.

Resultado principalEl resumen no menciona limitaciones.

Apareció: viernes, 16 de octubre. arXiv. Preprint, todavía sin revisión por pares.

Comentario de los autores: 16 pages, 14 figures, 7 tables. Under review at IEEE Journal on Selected Areas in Communications