Description

Calibration is any procedure that ties a sensor's output to a ground-truth scale or that refits a model's parameters to a new environment. The thesis core hypothesis — BLE-calibrated CSI crowd sensing — is built around periodic calibration campaigns in which BLE-derived trajectories supply ground-truth density and flow against which CSI inference is anchored. This note is the umbrella; specific variants (data-assimilation, domain-adaptation, fingerprint re-survey) are the implementation tactics.

When it's used

  • Periodic re-anchoring of CSI inference from BLE ground truth
  • Indoor-positioning fingerprint re-survey
  • Crowd-model parameter fitting (e.g. social-force-model parameters from real trajectories)
  • Sensor-fusion bias correction

Limitations

  • Calibration is only as good as the ground-truth signal feeding it
  • Re-calibration cadence vs cost is a deployment trade-off
  • Drift between calibrations is the residual error budget

Source Papers

  • wolinski2014_f409 — calibrating crowd simulators against data
  • chen2018_894a — SFM parameter calibration
  • zhong2022_7cb2 — calibration in CSI sensing
  • koo2026_a08d — calibration in WiFi sensing pipeline
  • duives2013_3924 — calibration in crowd-modelling review

1 vault paper use this method

Titles and DOIs only — no abstracts, no analyses.

  • A Survey on Green Wireless Sensing: Energy-Efficient Sensing via WiFi CSI and Lightweight Learning 2026 DOI ↗