Watermarking of time‑series data for weather forecasts

Purpose and goal

Rainfall can be estimated from mmWave signal attenuation using existing telecom links as dense sensing infrastructure. Shared datasets need ownership and provenance protection because metadata/signatures are often lost during processing. Digital watermarking can embed imperceptible identifiers in data, but little work exists for 1D time-series and mmWave attenuation. This thesis will design, implement, and evaluate watermarking methods that protect ownership while preserving data validity.

Expected effects and result

The goal is to develop or identify a watermarking technique for radio attenuation measurements. The results include an MSc thesis, presentation, and possibly a conference paper. The method should demonstrate that ownership can be protected without compromising the scientific validity of rainfall calculations.

Planned approach and implementation

Conduct literature review, collect/preprocess mmWave data, develop and implement watermarking in Python/Matlab, test robustness against common operations, evaluate impact on precipitation quantification, optimize parameters, and document results for thesis, presentation, and possible conference publication.