[Back]

O. Trindade:
"Supporting high quality cellular and Internet services inside railway vehicles: a statistical evaluation of wireless repeater performance on Railjet and Cityjet trains";
Supervisor: L. C. Kretly, C. Mecklenbräuker; University of Campinas (UNICAMP), Brasil, 2018; final examination: 05-18-2018.

The present work aims to support improvement on Internet and telephony services on-board railway vehicles by providing theoretical methodological apparatus to evaluate technological solutions in this context. The proposed method is used to benchmark the impacts of an Amplify-and-Forward (AF) wireless repeater on Radio Frequency signal quality (downlink) inside High Speed Train (HST)s, with focus on statistics of Radio Frequency (RF) received signal. To do so, mobile phones, under quasi-real usage conditions (voice and data services usages), are deployed inside an operational High Speed Train (HST) equipped with AF repeaters. Field measurements took place on railway tracks in Austria and covered five relevant bands: Global System for Mobile Communications (GSM) (bands 900MHz and 1800MHz), Universal Mobile Terrestrial Services (UMTS) (band 2100MHz) and Long Term Evolution (LTE) (bands 800MHz and 2600 MHz). The work combines bootstrap with parametric and non-parametric statistical hypothesis tests to characterize distributional behavior of Radio Frequency signal power level on thousand-point-large data samples. Comparison of the Student´s t-test and Wilcoxon test has shown that parametric t-test does not correctly quantify the change, yielding an unreliable statistical significance, this test is efficient to point the direction of improvement; however. It was found that the combination of this test with bootstrapping and non-parametric test is efficient in measuring the stochastic change in signal levels inside HSTs resulting repeater influence. Data analysis have shown that repeater effect changes the stochastic features of signal levels on HSTs, it produces a left-tailed and right-bounded distribution with large departure from the normal distribution, an indication that a measure of performance cannot be given solely via change in the mean. Moreover, data investigation strongly support downlink improvement under repeater condition in rural and urban areas, except for cases in which signal strength inside the HST averages above -80dBm most of the time. In addition, analytical results also support that repeater produces different improvements under diverse Radio Access Technology (RAT).

Cellular communication inside trains; Cellular network repeater on trains; vehicular communications; Inferential Statistics; Hypothesis test.

https://publik.tuwien.ac.at/files/publik_272055.pdf