Measuring the perceived quality of VoIP under various wireless network conditions

This paper focuses on the performance of VoIP calls under various situations, namely, during a handover, under different background traffic conditions (normal and heavy traffic load/saturation conditions) at a wireless access point, and in the presence of interference. Using empirical-based measurements, we demonstrate that various network conditions exhibit distinct statistical behaviour in terms of SNR, packet losses and end-toend delays, and thus, impact the VoIP user quality in a different manner. A novel part of this work is the comparative analysis of these conditions with respect to network and user experience benchmarks. It also proposes a Markov-chain based model for the auto-rate fallback (ARF) algorithm of the IEEE802.11 and derives analytically the packet losses and mean delays at the application layer under the presence of constant interference and a single traffic CBR source at the WLAN. The paper shows that only very low SNR values affect the user experience of a VoIP call, since the ARF algorithm conceals the packet losses at the MAC layer and only the losses at the most robust (and lowest bit-rate) modulation scheme at the MAC layer become noticeable at the application layer. Furthermore, it demonstrates situations in which common “rule-of-thumb” metrics cannot reflect the perceived user experience and finer cross-layer data are required for identifying the network conditions and performing an effective adaptation.