WiFi Performance Estimation for Voice Services

Abstract

The urge of having high data rates for mobile services puts a lot of pressure on existing non-cellular wireless networks such as WiFi, which have to adapt to the increasing demand of data. Carrying mobile services over WiFi networks is not straightforward, as they were not designed to carry conversational services like voice. Before suggesting any modification on the old, yet successful, WiFi networks, their performance must be re-evaluated for conversational services. In this paper, we develop a novel model to estimate the performance of a Voice over-WiFi (VoWiFi) system. Performance is described in terms of packet loss probability and average packet delay. Packet loss due to call blocking is differentiated from the loss due to reaching the maximum number of transmission retrials. Markov chain modeling is used to model each station, while the interaction between all stations is modelled as a closed queueing system. The model does not have a huge computational burden even when the number of nodes is large. Numerical results shows that WiFi networks still show a good performance for voice services but can be severely affected when transmission errors are present.