Internet performance profiling of countries and destinations

The thesis presents research into data capturing techniques and data analysis in order to discern the Internet performance profiles of countries, both spatially and temporally. The methodology may be described as implementing flexible yet tractable models, which simultaneously account for long-term traffic dependence, latency, dynamic random routing, and spatial dependence. Active Internet probing and traffic monitoring strategies are used in collaboration with the European RIPE Atlas project to capture raw Quality of Service (QoS) measurements. The research goal is to contribute towards developing large-scale network testing methods for superior performance detection, e.g. bottleneck detection as an application of reconstructed delay or loss estimation and build Internet QoS profiles for the countries of interest, as well as obtain indicators of the current performance state of Ireland's access to the global Internet. This thesis is organised into six chapters. Following an introduction and literature review, two major global Internet performance measurement systems CAIDA and RIPE are reviewed and compared. This is followed by two chapters detailing research methodology used in this project and research results and analysis. These chapters include details how and when the measurements were conducted, an example of the results obtained, together with an explanation of how these results can be used in order to acquire the needed information. The final chapter presents conclusions with a summary of the findings and ideas for future work. From the results collected, Internet QoS performance profiles for 37 countries are constructed. The worst case scenarios are explored and a reason for the low performance is given. The final results may be analysed and serve as a benchmark for mapping evolving and future performance profiling over longer periods of time, different paths or more destinations within a specific country, depending on the purpose of the experiment conducted. So for some of these outcomes and results have been published in 2013 at an Irish inter-University Engineering, Informatics and Science Research Day [95].