Data Science Institute | Sense, Collect & Move Data Center Seminar
Speaker: Jonathan Ostrometzky, Tel Aviv University
This research deals with the case where parameter estimation is required, but only observations of extreme values (i.e., the minimum observed value and/or the maximum observed value per interval) are available. Such estimation problem characterizes rain estimation from measurements logged by network managements systems, implemented on commercial microwave links in the backhaul of cellular networks. Based on a novel analytical approach which we developed in order to approximate the cumbersome expressions of the relevant Fisher information matrices and to produce simple, practical, and solvable forms, a fully applicable workflow of rain estimation will be presented. This new rain estimation workflow uses the available quantized version of the minimum and the maximum measured signal levels (reported at 15-minute intervals) which are produced by commercial microwave links of cellular networks. A demonstration of the established workflow using actual cellular data is presented, and is shown to produce accurate rain estimates, potentially in real time, with no need for training, prior, or side information.