Semi empirical path-loss modeling and cross-validation of industrial links at 5 GHz

Abstract

Introduction: Wireless links in industrial plants are exposed to metallic structures, absorbing materials, machinery, and multipath conditions that can reduce the accuracy of general propagation models. Objective: To develop and validate a semi-empirical path-loss model for 5 GHz point-to-point links in a paper and sugar industrial environment. Methods: Two line-of-sight links, Panasa–ISC and Panasa–Bomba 2, installed at approximately 15 m height, were analyzed. Eight spatial points were defined on each link, and eight received-power measurement series were collected using an RF analyzer, a Wi-Fi monitoring application, and a drone. Mean values were compared with the Friis free-space model and two industrial reference parameterizations. The α, β, and A path-loss parameters were fitted in MATLAB and evaluated using root mean square error and cross-validation between links. Results: The model fitted with Panasa–Bomba 2 yielded α=1.73, β=2.09, and A=0.8417. Its RMSE against the mean series was 1.20 dB on the source link and 2.87 dB when applied to the Panasa–ISC link. In both scenarios, it outperformed the Friis model and the absorbing-industrial reference, whose errors ranged from 4.11 to 5.93 dB. Conclusions: Local measurement-based semi-empirical fitting improved received-power estimation and showed transferability between two nearby industrial paths. The model can support preliminary link planning under line-of-sight conditions, distances below 500 m, and operation near 5 GHz.