Prediction of annual joint rain fade on EHF networks by weighted rain field selection

K S Paulson,I D Chinda

doi:10.1002/2015rs005695

Abstract

AbstractWe present a computationally efficient method to predict joint rain fade on arbitrary networks of microwave links. Methods based on synthetic rain fields composed of a superposition of rain cells have been shown to produce useful predictions of joint fade, with low computational overhead. Other methods using rain fields derived from radar systems have much higher computational overhead but provide better predictions. The proposed method combines the best features of both methods by using a small number of measured rain fields to produce annual fade predictions. Rain fields are grouped into heavy rain and light rain groups by maximum rain rate. A small selection of rain fields from each group are downscaled and fade predictions generated by pseudointegration of specific attenuation. This paper presents a method to optimize the weights used to combine the heavy rain and light rain fade predictions to yield an estimate of the average annual distribution. The algorithm presented yields estimates of average annual fade distributions with an error small compared to year‐to‐year variation, using only 0.2% of the annual data set of rain fields.

Highlights

The International Telecommunication Union-Radiocommunication Sector (ITU-R) maintains a set of internationally recognized propagation models
This paper presents a method to optimize the weights used to combine the heavy rain and light rain fade predictions to yield an estimate of the average annual distribution
The Global Integrated Network Simulator (GINSIM)-WeightedSelect system aims to produce a substantial decrease in the number of Nimrod rain maps required, and the associated computational effort, by using a weighted sum of the fades produced by a small selection of rain fields

Summary

Introduction

We propose a new method known as GINSIM-WeightedSelect This method uses only a small selection of measured rain fields over the period of interest to estimate the long-term distributions of rain fade. The very large reduction in computational effort, compared to GINSIM, comes from the following: (1) no temporal interpolation is required as time series are not produced, (2) relatively small numbers of Nimrod rain maps containing rain are required, and (3) optimizing the weighted average of distributions yields better estimates using less data. The GINSIM-WeightedSelect system aims to produce a substantial decrease in the number of Nimrod rain maps required, and the associated computational effort, by using a weighted sum of the fades produced by a small selection of rain fields

Downscaling Data

GINSIM-WeightedSelect

GINSIM-WeightedSelect Testing

Findings

Conclusions