Christian Moldovan

Modeling the YouTube stack

YouTube is one of the most popular and volume-dominant services in todays Internet, and has changed the web for ever. Consequently, network operators are forced to consider it in the design, deployment, and optimization of their networks. Taming YouTube requires a good understanding of the complete YouTube stack, from the network streaming service to the application itself. Understanding the interplays between individual YouTube functionalities and their implications for traffic and user Quality of Experience (QoE) becomes paramount nowadays. In this paper we characterize and model the YouTube stack at different layers, going from the generated network traffic to the QoE perceived by the users watching YouTube videos. Firstly, we present a network traffic model for the YouTube flow control mechanism, which permits to understand how YouTube provisions video traffic flows to users. Secondly, we investigate how traffic is consumed at the client side, deriving a simple model for the YouTube application. Thirdly, we analyze the implications for the end user, and present a model for the quality as perceived by them. This model is finally integrated into a system for real time QoE-based YouTube monitoring, highly useful to operators to assess the performance of their networks for provisioning YouTube videos. The central parameter for all the presented models is the buffer level at the YouTube application layer. This paper provides an extensive compendium of objective tools and models for network operators to better understand the YouTube traffic in their networks, to predict the playback behavior of the video player, and to assess how well they are doing in practice in delivering YouTube videos to their customers.

To each according to his needs: Dimensioning video buffer for specific user profiles and behavior

Todays video streaming platforms offer videos in a variety of quality settings in order to attract as many users as possible. But even though a sufficiently dimensioned network can not always be provided for the best experience, users are asking for high QoE. Users consume the content of a video streaming platform in different ways, while video delivery platforms currently do not account for these scenarios and thus ensure at best mediocre QoE. In this paper, we develop a queuing model and provide a mean-value analysis to investigate the impact of user profiles on the QoE of HTTP Video Streaming for typical user scenarios. Our results show that the user profile and particularly the scenario have to be respected when dimensioning the buffer. Further, we present recommendations on how to adapt player parameters in order to optimize the QoE for individual users profiles and viewing habits. The provided model leads to relevant insights that are required to build a system that guarantees each user the best attainable QoE.

TCP video streaming and mobile networks

The rise in popularity of TCP-based video streaming in recent years is unbroken. These streaming services not just operate on wired access lines but more and more specifically target users of mobile networks as well. Yet it still remains difficult to evaluate the performance of such streaming approaches in mobile networks. This is especially critical as mobile networks exhibit much more potential for undesirable interactions between the network protocol layers and control plane properties on one side and the protocols and strategies of the application layer on the other side, ultimately resulting in scenarios with bad QoE for video streaming.This paper aims to rectify this lack of knowledge and understanding in a multi-pronged approach as follows: The first contribution provides an easy means to investigate such interactions with a streaming simulation framework built on ns-3. In a second contribution three exemplary scenarios within this framework are investigated in order to uncover the nature of such interactions. The final undertaking attempts to unravel the issues of the mobility scenario using context information. Such information can be collected through crowdsensing and collaboratively processed in a Big Data approach. This results in a tailor-made analytical solution through the formulation of a Mixed Integer Linear Program (MILP) that can prevent video stalling in this particular scenario.

Enriching HTTP adaptive streaming with context awareness: A tunnel case study

Video streaming provided by Over-The-Top (OTT) service providers through a cellular network is a common usage scenario for many people today. While video streaming will work reasonably well in a stationary scenario, several issues arise for mobile users. For instance, travelling through short areas without cellular coverage, such as an automobile tunnel, will often result in quality degradation or video stalling. To combat this, this paper provides an analytical investigation of the video quality degradation problem as it is experienced by mobile users and suggests a context-aware HTTP Adaptive Streaming (HAS) strategy to prevent stalling and minimize the impact on Quality of Experience (QoE). This provides a solution that can completely prevent stalling when appropriate context information (such as positional information from satellite navigation) is present. The collected evaluation results encourage further research on how context-awareness can be exploited to further enhance video service provisioning by OTT service providers.

YouTube Can Do Better: Getting the Most Out of Video Adaptation

YouTube, as one of the major HTTP Adaptive Streaming video services, accounts for a large fraction of todays Internet traffic. Therefore, it is important to understand how efficiently YouTube uses available network resources. Previous work observed that the YouTube player replaces previously buffered segments with higher quality segments. This is good for the user as it increases the average quality level. However, the lower quality level segments are discarded and their traffic is redundant and therefore wasted. In this paper, we use two independent approaches to evaluate the efficiency of YouTubes quality adaptation algorithm. The first approach performs regression based on previously collected video views from a large experimental data set. In the second approach we formulate a mixed integer linear program and calculate the optimal video quality adaptation. The results show that the simplistic regression approach gives an accurate estimation of the optimal adaptation. Furthermore, the optimization shows that the Quality of Experience (QoE) can be significantly improved compared to the actual average quality level observed in the real-world experiments, demanding for better video quality adaptation mechanisms by YouTube.

Impact of Variances on the QoE in Video Streaming

In the current Internet video streaming is dominating fixed and mobile consumer traffic. However, wireless and mobile networks may lack bandwidth or suffer from severe bandwidth fluctuations. The question arises: What is the impact of bandwidth fluctuations on the Quality of Service (QoS) and the user perceived quality? The key contribution of this paper is the analysis of QoS in terms of application-level parameters (stalling duration and frequency) for reliable video-on-demand streaming. The QoS is mapped to Quality of Experience (QoE) based on an existing QoS-QoE model for HTTP streaming. Different network conditions, in particular average and variances of network bandwidths, and video bit rates are taken into account in the analysis. We approach this problem by modeling the video player as a queueing system and evaluate the transient phase of the non-stationary system by means of discrete-event simulations. The results of our study show that the characteristics of the network have a high impact on the QoS and a surprisingly low impact on the QoE under certain conditions.

Practical QoE Evaluation of Adaptive Video Streaming

Video streaming is an increasingly popular service on the Internet. In HTTP adaptive video streaming (HAS), the video is played while being downloaded, and the quality is selected according to the available bandwidth. Due to this, variations in the transmission affect the playback. The quality of the playback can be rated by technical parameters, which can be grouped by the term ‘Quality of Service’ (QoS), like the video quality, the number and duration of stallings or the time until the video starts playing. These metrics differently influence the user experience.

The Value of Context-Awareness in Bandwidth-Challenging HTTP Adaptive Streaming Scenarios

Video streaming has become an indispensable technology in people’s lives, while its usage keeps constantly increasing. The variability, instability and unpredictability of network conditions pose one of the biggest challenges to video streaming. In this chapter, we analyze HTTP Adaptive Streaming, a technology that relieves these issues by adapting the video reproduction to the current network conditions. Particularly, we study how context awareness can be combined with the adaptive streaming logic to design a proactive client-based video streaming strategy. Our results show that such a context-aware strategy manages to successfully mitigate stallings in light of network connectivity problems, such as an outage. Moreover, we analyze the performance of this strategy by comparing it to the optimal case, as well as by considering situations where the awareness of the context lacks reliability.

Keep Calm and Don’t Switch: About the Relationship Between Switches and Quality in HAS

Video streaming has established itself as the main method to consume multimedia content on the Internet. The most widespread technique is HTTP Adaptive Streaming which is used by large video service platforms such as YouTube and Netflix. In order to guarantee a high QoE, different adaptation strategies have been developed that lead to different video playout patterns. While some strategies may be more aggressive than others and often adapt the video quality, it is not clear which strategy is the best.,,In this paper, we want to identify the trade-off between the average video quality and switches in the quality during playout. We do this with a user-centric view and try to optimize the adaptation depending on the user preference with a quadratic program. This work allows to put existing and future video adaptation algorithms in perspective with respect to user preferences. Our results show that the video quality can already be increased greatly by allowing few switches while more switches lead to diminishing gains. This is a novel discovery that is important for user-centric QoE-management which is of high interest for ISPs and video service providers.

Viability of Wi-Fi caches in an era of HTTPS prevalence

Caches are often employed to better manage the ever-growing amount of global Internet traffic. Particularly, reverse proxies are often used to free up congested peering links and reduce inter-domain traffic. In addition, caches place content close to the edge, leading to lower latency which can be beneficial for many applications. In an effort to bring content even closer to end devices, home routers and public access points are now in the spotlight as a novel location for caches. Compared to CDNs, a Wi-Fi router only offers scarce storage and processing capabilities, putting an increased emphasis on resource management. In addition, a large share of traffic is transmitted through secured connections preventing the use of forward caches and other middle boxes. This paper strives to analyze how effective caching can be in public Wi-Fi routers. For this purpose, we conducted a field study with caches at public Wi-Fi hotspots in a field study. In addition, we present a queueing model for caches that we use to perform a mean-value analysis. We compare results from both methods in a performance analysis. Our results provide insights into difficulties of caching systems that are caused by an increase in the ratio of secured connections in the Internet.