Extraction of input parameters for the theory of radiative energy transfer using deconvolution

• Main Author: Cui, Huajian
• Other Authors: Richter, Jurgen
• Published: University of South Wales 2009
• Subjects: 621.382; Radio wave propagation
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.496064

The ever growing application of wireless communication systems requires accurate models for characterising radiowave propagation when affected by the presence of a variety of obstacles. In particular if the obstacles take the shape of vegetation volumes, like single trees or groups of trees and are present in the radio path, they give rise to absorption and scattering of radio signals. This thesis presents a literature review of common models for radiowave propagation through vegetation, the theory of Radiative Energy Transfer (RET) is one of these models and provides an accurate analysis of radiowave propagation through a vegetation media. Extensive measurements have been designed and conducted in a controlled indoor environment to provide valuable measurement data for later development of deconvolution approaches. It can be shown that the measured directional spectra are convolution products of the phase function pattern and the receiver antenna radiation patterns, which impacts determination of the RET input parameters. Consequently, in order to achieve more accurate determination of the RET input parameters, the adverse influence caused by receiver antenna radiation patterns have to be removed from measured directional spectra by implementing a process of deconvolution. This thesis provides successful implementation of two iterative based deconvolution techniques on the measurement directional spectra. To the author's knowledge, this is its first kind of application to eliminate distortion caused by the receiver antenna radiation pattern during measurements. This thesis reports a number of novel approaches. These include the further development and extension of deconvolution techniques such as combining the Bennia-Riad criterion and an error function to determine optimal parameters, as well as using pre-filtering techniques to improve the deconvolution results. Development of clearly defined criteria based on the knowledge of the central-limit theorem and discussion of loss of information avoidance during convolution is another novel contribution. Further novelty lies in the modification of the two methods to suit implementation on the measurement data from radiowaves impacting on vegetation volumes. As a result of these refinements, extracted RET input parameters from the restored patterns after applying the deconvolution processes show evident improvements compared to those extracted from directly measured patterns. Early stage results of this project are published in the IEEE Proceedings on Next Generation Applications, Services and Technologies.