[Back]

A. Martinez Castillo:
"Temporal and spatial correlation analysis for mobile network benchmarking";
Supervisor: M. Rupp, P. Svoboda, C. Midoglu; Institute of Telecommunications, TU Wien, 2016; final examination: 06-01-2016.

Crowdsourcing is a new system which nowadays is starting to be used for collecting
information and improving different existing processes. One crowdsourcing application is based on mobile phone datarate measurements for estimating network performance. This Master Thesis is focused on improving mobile network benchmarking through crowdsourcing information analysis. This analysis is focused in two different challenges: one based on reducing resource consumption of the network and the other is based on using an efficient sample size for benchmarking.
The First challenge, resource consumption reduction, can be achieved through reducing
test duration. Different mathematical approximations can help us to reduce the actual
time test. These approximations are three and are based on linear interpolation, exponential functions and Autoregresive Integrated Moving Average (ARIMA) approximations.
The second challenge, efficient sample size, will be addressed using correlation measurement properties in combination with some models for estimating the expected data rate over time, in order to do the same network benchmarking using a reduced number of samples. The publicly available Rundfunk and Telekom Regulierungs-GmbH (RTR) Open Data will be used as data source and Matlab will provide us mathematical tools.

ARIMA; time series; ACF; PACF; RTR Open Data; exponential approximation; linear approximation; JSON.