LoRaWAN Gateway & Coverage Calculator

Path loss — Okumura-Hata for 865–868 MHz, then corrected against Macnman field data. Uncalibrated, Hata assumes flat, clutter-free terrain with full Fresnel clearance and over-predicts LoRaWAN badly: at 866 MHz over 8 km the first Fresnel zone needs about 26 m of mid-path clearance, and a 30 m gateway talking to a 2 m sensor only gives about 15 m. Our open-terrain test reached 8 km, i.e. 37.8 dB more loss than the model predicts, and that correction is applied per environment.

Reliable vs best case — cell sizes use a fade margin, so they are the range you can design around. The best-case figure drops the margin and matches what a clear-day range test produces.

RSSI and SNR — median signal at the cell edge. SNR is RSSI above the thermal noise floor (−174 dBm/Hz + 10·log₁₀(BW) + 6 dB noise figure = −117 dBm at 125 kHz), checked against the published demodulator limits, which reproduce the sensitivity table to within 0.1 dB at every SF.

Device-to-device — Hata assumes a raised base station, so MacTalk uses a log-distance model. With both radios near the ground the earth blocks the Fresnel zone along the whole path, so the exponents sit in the two-ray regime near 4 even over open ground.

Radio horizon — every distance is capped at 4.12·(√h₁+√h₂) km. No link budget beats the curvature of the earth.

Airtime — Semtech AN1200.13, including the 13-byte LoRaWAN header and mandatory low-data-rate optimisation at SF11/SF12.

Capacity — a 10% channel-utilisation ceiling across 8 uplink channels, since LoRaWAN is pure ALOHA.

Planning estimates only — confirm with a site survey before committing to a design.