Multidimensional big spatial data modeling through a case study: LTE rf subsystem power consumption modeling

Abstract

This paper presents a case study for comparing different multidimensional mathematical modeling methodologies used in multidimensional spatial big data modeling and proposing a new technique. An analysis of multidimensional modeling approaches (neural networks, polynomial interpolation and homotopy continuation) was conducted for finding an approach with the highest accuracy for obtaining reliable information about a cell phone consumed power and emitted radiation from streams of measurements of different physical quantities and the uncertainty ranges of these measure ments. The homotopy continuation numerical approach proved to have the highest accuracy (97%). This approach was validated against another device with a different RF subsystem design. The approach modelled the power consumption of the validation device with an accuracy of 98%.