CellScope — Cell Signal Geolocation Engine

The Problem

A telecom analysis application needed a module that could estimate a device's likely location from signal records, without relying on expensive commercial tools.

The Approach

Designed a confidence-scoring algorithm combining RSRP signal-strength weighting, timing-advance (TA) distance rings and sector-angle validation. Used Polars + DuckDB instead of pandas for performance, and made results exportable as KML/GeoJSON/Excel/PDF.

Engineering Highlights

• Algorithm converting TA values into physical distance rings per LTE/GSM/UMTS generation.
• RSRP-weighted centroid calculation plus ConvexHull coverage-area polygon.
• 0-100 location confidence score, multi-format export (KML/GeoJSON/Excel/PDF).
• Real pytest unit tests plus an end-to-end scenario via FastAPI's TestClient.