Empirical Frequency-Dependent Wall Insertion Loss Model at 3–6 GHz for Future Internet-of-Things Applications

A novel frequency-dependent wall insertion loss model at 3–6 GHz is proposed in this paper. The frequency-dependence of the wall insertion loss is modeled by the Fourier triangular basis neural network. A method to determine the optimal weighted vector and the number of the neurons is introduced. In addition, the impact of the wider continuous spectrum on the wall insertion loss is analyzed and extensive measurements are performed to validate the proposed model. The results obtained with the proposed model match better with the measured results than other models. The proposed model can be used in future indoor Internet-of-Things applications such as service computing.