PhD Defence – Jason Ernst

PhD candidate Jason Ernst will defend his thesis "Connection Point Selection in HWNs to Improve QoS/QoE, Reduce Cost and Increase Profits" on December 3, 2015, at 9:30am in Reynolds 219.

Title

Connection Point Selection in HWNs to Improve QoS/QoE, Reduce Cost and Increase Profits

Abstract

A heterogeneous wireless network (HWN) is made up from multiple radio access technologies (RATs) such as Bluetooth, Wi-Fi, and Long Term Evolution (LTE). The use of multiple RATs may allow for increased capacity, coverage, flexibility, reliability and increased spectral and energy efficiency. One of the most important challenges for the users and operators of HWNs is matching the quality of service (QoS) and quality of experience (QoE) requirements of the user with the particular RATs the operators offer. Furthermore, typical cellular operators have complicated service agreements between the users and other operators. The users wish to select the best network for the lowest cost, while operators wish to select users which maximize their profits. Traditionally, in homogeneous wireless networks such as Wi-Fi, the best access point (AP) is selected based on the strongest signal. This dissertation makes several contributions. First, an in depth general survey of the state of the art in HWNs is presented, along with many of the current challenges and open research topics, with a particular focus on those which affect QoS and QoE. Next, focus is drawn to how networks are currently measured and optimized, which is typically with only a single metric. The limitations of this methodology are outlined. Since many potential networks involved in a HWN may potentially interfere with one another, such as Wi- Fi and Bluetooth, an experiment was performed to examine what the performance degradation may be in the case where a naïve approach is taken and interference is not avoided between RATs operating in the same frequency range. Next, a module for simulating this behaviour in the NS3 simulation tool is presented. This is followed by an experiment which demonstrates the limitation of choosing an access point (AP) in a homogeneous network, which is a simple case of the more complex problem of RAT selection in heterogeneous networks. Finally, the challenges of selecting a RAT in a HWN are examined with two more simulation experiments which take into account selection criteria other than signal strength, along with the complex relationships between operators and the pricing schemes between operators and users. The key contributions of this work are threefold, first - providing an overview and critical evaluation of previous works in this area and to provide a roadmap of open areas of research in HWNs (Chapter 1 and 2). Second, to identify the limitations in the manner in which networks are currently measured, evaluated and ranked (Chapter 3, 4, 5, 6). Finally, to propose and implement demonstrations for solutions in which this situation may be improved, along with a framework to build on for future work (Chapters 4, 5, 6). The framework provided is in the form of a simulation tool that allows for modelling of complex financial relationships between users and operators, and between operators and other operators that is lacking from all of the popular simulation tools that currently exist. It can easily be integrated into popular tools such as NS3 and NS2.

Advisor: Stefan Kremer