We characterize the performance of both video and network layer properties of Skype, the most popular video telephony application. The performance in both mobile and stationary scenarios is investigated; considering network characteristics such as packet loss, propagation delay, available bandwidth and their effects on the perceptual video quality, measured using spatial and temporal no-reference video metrics. Based on 200+ live traces, we study the performance of this mobile video telephony application. We model video quality as a function of input network parameters and derive a feed-forward Artificial-Neural-Network that accurately predicts video quality given network conditions (0.0206 ≤ MSE ≤ 0.570). The accuracy of this model improves significantly by incorporating end-user mobility as an input to the model.

Video quality assessment in mobile devices, for instances smart phones and tablets, raises unique challenges such as unavailability of original videos, the limited computation power of mobile devices and inherent characteristics of wireless networks (packet loss and delay). In this paper, we present a metric, Temporal Variation Metric (TVM), to measure the temporal information of videos. Despite its simplicity, it shows a high correlation coefficient of 0.875 to optical flow which captures all motion information in a video. We use the TVM values to derive a reduced-reference temporal quality assessment metric, Temporal Variation Index (TVI), which quantifies the quality degradation incurred in network transmission. Subjective assessments demonstrate that TVI is a very good predictor of users' Quality of Experience (QoE). Its prediction shows a 92.5% of correlation to subjective Mean Opinion Score (MOS) ratings. Through video streaming experiments, we show that TVI can also estimate the network conditions such as packet loss and delay. It depicts an accuracy of almost 95% in extensive tests on 183 video traces.

The IEEE 802.11n standard supports very high physical layer data rates using Multiple Input Multiple Output (MIMO) antenna technologies. Despite such high rates, High Definition (HD) video streaming is still challenging in WLAN deployments. In this paper, we show that the wireless channel probing overhead of existing 802.11n data rate adaptation mechanisms can be detrimental to HD video performance. We propose VARA, a Video-Aware Rate Adaptation protocol that addresses this problem by adapting the frequency and timing of wireless probing to both video encoding rate variations and wireless channel variations. In addition, VARA employs novel strategies that multiplex several Variable Bit Rate (VBR) HD video streams by minimizing their aggregate peak rate requirement. Our experimental evaluations for static and mobile scenarios in a MIMO 802.11n wireless testbed demonstrate the practical benefits of VARA over state-of-the-art 802.11n rate adaptation protocols.

Peak Signal-to-Noise Ratio (PSNR) is the simplest and the most widely used video quality evaluation methodology. However, traditional PSNR calculations do not take the packet loss into account. This shortcoming, which is amplified in wireless networks, contributes to the inaccuracy in evaluating video streaming quality in wireless communications. Such inaccuracy in PSNR calculations adversely affects the development of video communications in wireless networks. This paper proposes a novel video quality evaluation methodology. As it not only considers the PSNR of a video, but also with modifications to handle the packet loss issue, we name this evaluation method MPSNR. MPSNR rectifies the inaccuracies in traditional PSNR computation, and helps us to approximate subjective video quality, Mean Opinion Score (MOS), more accurately. Using PSNR values calculated from MPSNR and simple network measurements, we apply linear regression techniques to derive two specific objective video quality metrics, PSNR-based Objective MOS (POMOS) and Rates-based Objective MOS (ROMOS). Through extensive experiments and human subjective tests, we show that the two metrics demonstrate high correlation with MOS. POMOS takes the averaged PSNR value of a video calculated from MPSNR as the only input. Despite its simplicity, it has a Pearson correlation of 0.8664 with the MOS. By adding a few other simple network measurements, such as the proportion of distorted frames in a video, ROMOS achieves an even higher Pearson correlation (0.9350) with the MOS. Compared with the PSNR metric from the traditional PSNR calculations, our metrics evaluate video streaming quality in wireless networks with a much higher accuracy while retaining the simplicity of PSNR calculation.

Generating a secret key between two parties by extracting the shared randomness in the wireless fading channel is an emerging area of research. Previous works focus mainly on single-antenna systems. Multiple-antenna devices have the potential to provide more randomness for key generation than single-antenna ones. However, the performance of key generation using multiple-antenna devices in a real environment remains unknown. Different from the previous theoretical work on multiple-antenna key generation, we propose and implement a shared secret key generation protocol, Multiple-Antenna KEy generator (MAKE) using off-the-shelf 802.11n multiple-antenna devices. We also conduct extensive experiments and analysis in real indoor and outdoor mobile environments. Using the shared randomness extracted from measured Received Signal Strength Indicator (RSSI) to generate keys, our experimental results show that using laptops with three antennas, MAKE can increase the bit generation rate by more than four times over single-antenna systems. Our experiments validate the effectiveness of using multi-level quantization when there is enough mutual information in the channel. Our results also show the trade-off between bit generation rate and bit agreement ratio when using multi-level quantization. We further find that even if an eavesdropper has multiple antennas, she cannot gain much more information about the legitimate channel.

Wireless simulators provide full control to researchers in investigating traffic flow behavior, but do not always reflect real-world scenarios. Although previous work pointed out such shortages are due to the limitation of radio propagation models in the simulators, it is still unclear how these imperfect models affect network behavior and to what degree. In this paper, we quantify network behavioral differences between simulations and real-world testbed experiments. We compare and analyze the experimental results from indoor and outdoor experiments with the results from NS-2 and Qualnet simulations. We find that in the PHY layer, the distribution of received signal strength in experiments is usually different from simulation due to the antenna diversity. However, path loss, which is regarded as a dominating factor in simulator channel modeling, can be configured to match the real-world behavior. For the MAC layer, increasing traffic load on a flow may cause significant performance degradation in experiments, but it is not the case in simulations. Interference is inadequately captured in simulations and cannot show the flow level unfairness phenomenon. In the network layer, a few dominant routes exist in testbed experiments while routes in simulations are less stable. These findings give the wireless research community an improved overview about the differences between simulations and testbed experiments. They will help researchers to choose between simulations and experiments according to their particular research requirements.