A study of mobile VoIP performance in wireless broadband networks

Voice service is to date still the killer mobile service and the main source for operator revenue for years to come. Additionally, voice service will evolve from circuit switched technologies towards packet based Voice over IP (VoIP). However, using VoIP over wireless networks different from 3GPP cellular technologies makes it also a disruptive technology in the traditional telecommunication sector. The focus of this dissertation is on determining mobile VoIP performance in different wireless broadband systems with current state of the art networks, as well as the potential disruption to cellular operators when mobile VoIP is deployed over different access networks. The research method is based on an empirical model. The model and experiments are well documented and based on industry standards for voice quality evaluation. The evaluation provides results from both experiments in a controlled laboratory setup as well as from live scenarios. The research scope is first, evaluate each network technology independently; second, investigate vertical handover mobility cases; third, determine other aspects directly affecting end user experience (e.g., call setup delay and battery lifetime). The main contribution of this work is a systematic examination of mobile VoIP performance and end user experience. The research results point out the main challenges for achieving call toll quality, and how derive the required changes and technological performance roadmap for improved VoIP service. That is, investigate how the performance and usability of mobile VoIP can eventually be improved to be a suitable substitute for circuit switched voice. In addition, we evaluate the potential disruption to cellular operators that mobile VoIP brings when deployed over other access networks. This research extends the available knowledge from simulations and provides an insight into actual end user experience, as well as the challenges of using embedded clients in handheld devices. In addition, we find several issues that are not visible or accounted for in simulations in regard to network parameters, required retransmissions and decreased battery lifetime. The conclusion is that although the network performance of several wireless networks is good enough for near toll quality voice in static scenarios, there are still a number of problems which make it currently unfeasible to use as a primary voice service. Moreover, under mobility scenarios performance is degraded. Finally, there are other issues apart from network performance such as energy consumption, hardware limitations and lack of supporting business models (e.g., for WiFi mesh) that further limit the possibility of rolling out mobile VoIP services