Empirical Model of RadioWave Propagation in the Presence of Vegetation inside Greenhouses Using Regularized Regressions
Metadata
Show full item recordAuthor
Cama Pinto, Dora; Damas Hermoso, Miguel; Holgado Terriza, Juan Antonio; Gómez Mula, FranciscoEditorial
Mdpi
Materia
Wireless propagation model Precision agriculture COST235 FITU-R ITU-R Weisbberger model Propagation model Regularized regressions
Date
2020-11-19Referencia bibliográfica
Cama-Pinto, D., Damas, M., Holgado-Terriza, J. A., Arrabal-Campos, F. M., Gómez-Mula, F., Martínez Lao, J. A., & Cama-Pinto, A. (2020). Empirical model of radio wave propagation in the presence of vegetation inside greenhouses using regularized regressions. Sensors, 20(22), 6621. [doi:10.3390/s20226621]
Abstract
Spain is Europe’s leading exporter of tomatoes harvested in greenhouses. The production
of tomatoes should be kept and increased, supported by precision agriculture to meet food and
commercial demand. The wireless sensor network (WSN) has demonstrated to be a tool to provide
farmers with useful information on the state of their plantations due to its practical deployment.
However, in order to measure its deployment within a crop, it is necessary to know the communication
coverage of the nodes that make up the network. The multipath propagation of radio waves between
the transceivers of the WSN nodes inside a greenhouse is degraded and attenuated by the intricate
complex of stems, branches, leaf twigs, and fruits, all randomly oriented, that block the line of
sight, consequently generating a signal power loss as the distance increases. Although the COST235
(European Cooperation in Science and Technology - COST), ITU-R (International Telecommunications
Union—Radiocommunication Sector), FITU-R (Fitted ITU-R), and Weisbberger models provide
an explanation of the radio wave propagation in the presence of vegetation in the 2.4 GHz ICM
band, some significant discrepancies were found when they are applied to field tests with tomato
greenhouses. In this paper, a novel method is proposed for determining an empirical model of
radio wave attenuation for vegetation in the 2.4 GHz band, which includes the vegetation height
as a parameter in addition to the distance between transceivers of WNS nodes. The empirical
attenuation model was obtained applying regularized regressions with a multiparametric equation
using experimental signal RSSI measurements achieved by our own RSSI measurement system for
our field tests in four plantations. The evaluation parameters gave 0.948 for R2, 0.946 for R2 Adj
considering fifth grade polynomial (20 parameters), and 0.942 for R2, and 0.940 for R2 Adj when a
reduction of parameters was applied using the cross validation (15 parameters). These results verify
the rationality and reliability of the empirical model. Finally, the model was validated considering
experimental data from other plantations, reaching similar results to our proposed model.