Empirical evaluation of factors affecting network performance in mobile data networks

Abstract

Uniform and affordable Internet is emerging as one of the fundamental civil rights in most of the countries. The mobile Internet traffic has increased by 30 fold over the past decade and it is estimated that there will be a 300\% increase by 2017 compared to 2015. Many users are replacing desktops and laptops with smart phones and tablets as their default computing devices, thereby shifting data traffic from wired to wireless network. It would be every user's dream to use a network of infinite capacity with zero latency on their smart phones. However, throughout history, communication has been limited by various natural and artificial factors like landscapes, distance, communication equipments etc.

With such limitations in mind, it is interesting to know if users are getting optimum QoS that they have paid for. If not, it is necessary to understand the factors that are restricting them from getting that optimal performance. Although there are a wide range of tools and services that are available to gauge and provide a snapshot of network performance, from the user perspective, the lingering question - why did the user get the indicated result at a given time and place? is yet to be answered. Understanding these factors affecting network performance will also help to assess the state of mobile networks and its adoption rate of different mobile standards (2G, 3G, and 4G) for commercial use.

In this thesis, through data analysis and empirical evaluation, some of the major factors and extent to which these factors affect the user's perceived quality of service have been identified. Also, based on empirical evaluation, an algorithm has been developed that explains probable reasons affecting the current performance based on the above mentioned factors. The algorithm is expected to provide information to users to make informed decisions to get a better quality of service.