Empirical baseline model for distance estimation using 868 MHz LoRa modulated radio
AI & Data
Semester programme:Master Applied IT
Project group members:Jan Fojtik
Project description
This project investigates whether RSSI measurements in the 868 MHz ISM band can serve as a reliable basis for distance estimation. Using LoRa modulated radio and a peer-to-peer node setup with fixed transmission parameters, a series of controlled outdoor experiments were conducted to characterize signal attenuation as a function of physical distance.
The study also examines the effect of human body occlusion on the propagation path and quantifies the uncertainty inherent in inverting the path loss model for distance prediction.
Context
Care facilities are exploring low-power indoor networks to support patient localization. Existing infrastructure in this sector commonly operates on the 868 MHz band, making it a practical candidate. Before a full localization system can be designed, the more fundamental question of whether RSSI can estimate distance reliably must be answered.
This study addresses that precursor problem in a controlled, interference-minimized environment to isolate the distance-related attenuation profile from confounding indoor effects.
Results
OLS regression over an open-field dataset produced a path loss exponent of 3.18 and a shadow fading variance of 4.43 dB, yielding the model RSSI(d) = -28.38 - 31.80 log(d) + X. This corresponds more closely to the log-distance path loss model than previously reported indoor measurements.
Human body occlusion showed modest attenuation at short ranges, with anomalous positive deviations at 15 m suggesting constructive ground reflection. Distance estimation uncertainty grows asymmetrically with range, limiting RSSI to coarse proximity indication beyond approximately 100 m.