This PDF file contains the front matter associated with SPIE-IS&T Proceedings Volume 7253, including the Title Page, Copyright information, Table of Contents, Introduction, and Conference Committee listing.

Assessing video content transmitted over networked content infrastructures becomes a fundamental requirement for service providers. Previous research has shown that there is no direct correlation between traditional network QoS and user perceived video quality. This paper presents a study investigating the impact of individual packet loss on four types of H.264 main-profile encoded video streams. Four artifact factors to model the degree of artifacts in video frames are defined. Further, the visibility of artifacts considering the video content characteristics, encoding scheme and error concealment is investigated in conjunction with a user study. The individual and joint impacts of artifact factors are explored on the test video sequences. From the results of user tests, the artifact factor-based assessment method shows superiority over PSNR-based and network QoS based quality assessment.

In this paper, we propose a novel algorithm for constructing an Unequal Error Protection (UEP) FEC code targeted towards video streaming applications. A concatenation of a set of parallel outer block codes followed by a packet interleaver and an inner block code is presented. The algorithm calculates on the fly the optimal allocation of the code rates of the inner and outer codes. When applied to video streaming applications using H.264, the discussed UEP framework achieves gains of up to 5dB in video quality compared to equal error protection (EEP) FEC at the same code rate.

Video streaming over wireless networks is getting very popular because of the high bandwidth and the support of quality of service offered by recent wireless standards, such as IEEE 802.11e. We consider optimizing the quality of video streaming in single-hop wireless networks that are composed of multiple wireless stations. Our optimization problem controls parameters in different layers to optimally allocate the wireless network resources among all stations. We address this problem in two steps. First, we formulate an abstract optimization problem for video streaming in single-hop wireless networks in general. This formulation exposes the important interaction between parameters belonging to different layers in the network stack. Then, we instantiate and solve the general problem for the recent IEEE 802.11e WLANs, which support prioritized traffic classes. We show how the calculated optimal solutions can efficiently be implemented in the distributed mode of the IEEE 802.11e standard. We evaluate our proposed solution using extensive simulations in the OPNET simulator, which captures most features of realistic wireless networks. In addition, to show the practicability of our solution, we have implemented it in the driver of an off-the-shelf wireless adapter that complies with the IEEE 802.11e standard. Our experimental and simulation results show that significant quality improvement in video streams can be achieved using our solution, without incurring any significant communication or computational overhead.

The advent of virtualization and cloud computing technologies necessitates the development of effective mechanisms for the estimation and reservation of resources needed by content providers to deliver large numbers of video-on-demand (VOD) streams through the cloud. Unfortunately, capacity planning for the QoS-constrained delivery of a large number of VOD streams is inherently difficult as VBR encoding schemes exhibit significant bandwidth variability. In this paper, we present a novel resource management scheme to make such allocation decisions using a mixture of per-stream reservations and an aggregate reservation, shared across all streams to accommodate peak demands. The shared reservation provides capacity slack that enables statistical multiplexing of peak rates, while assuring analytically bounded frame-drop probabilities, which can be adjusted by trading off buffer space (and consequently delay) and bandwidth. Our two-tiered bandwidth allocation scheme enables the delivery of any set of streams with less bandwidth (or equivalently with higher link utilization) than state-of-the-art deterministic smoothing approaches. The algorithm underlying our proposed framework uses three per-stream parameters and is linear in the number of servers, making it particularly well suited for use in an on-line setting. We present results from extensive trace-driven simulations, which confirm the efficiency of our scheme especially for small buffer sizes and delay bounds, and which underscore the significant realizable bandwidth savings, typically yielding losses that are an order of magnitude or more below our analytically derived bounds.

We address the problem of the proper choice of the thickness of pre-encoded video layers in congestion-controlled streaming applications. While congestion control permits to distribute the network resources in a fair manner among the different video sessions, it generally imposes an adaptation of the streaming rate when the playback delay is constrained. This can be achieved by adding or dropping layers in scalable video along with efficient smoothing of the video streams. The size of the video layers directly drives the convergence of the congestion control to the stable state. In this paper, we derive bounds on both the encoding rates of the video layers that depend on the prefetch delay that can be used for stream smoothing. We then discuss the practical scheduling aspects related to the transmission of layered video when delays are constrained. We finally describe an implementation of the proposed scheduler and we analyze its performance in NS-2 simulations. We show that it is possible to derive a media-friendly rate allocation for layered video in different transmission scenarios, and that the proper choice of the layer thickness improves the average video quality when the prefetch delay is constrained.

Distributed interactive applications tend to have stringent latency requirements and some may have high bandwidth demands. Many of them have also very dynamic user groups for which all-to-all communication is needed. In online multiplayer games, for example, such groups are determined through region-of-interest management in the application. We have investigated a variety of group management approaches for overlay networks in earlier work and shown that several useful tree heuristics exist. However, these heuristics require full knowledge of all overlay link latencies. Since this is not scalable, we investigate the effects that latency estimation techqniues have ton the quality of overlay tree constructions. We do this by evaluating one example of our group management approaches in Planetlab and examing how latency estimation techqniues influence their quality. Specifically, we investigate how two well-known latency estimation techniques, Vivaldi and Netvigator, affect the quality of tree building.

The quality of service for latency dependent content, such as video streaming, largely depends on the distance and available bandwidth between the consumer and the content. Poor provision of these qualities results in reduced user experience and increased overhead. To alleviate this, many systems operate caching and replication, utilising dedicated resources to move the content closer to the consumer. Latency-dependent content creates particular issues for community networks, which often display the property of strong internal connectivity yet poor external connectivity. However, unlike traditional networks, communities often cannot deploy dedicated infrastructure for both monetary and practical reasons. To address these issues, this paper proposes Corelli, a peer-to-peer replication infrastructure designed for use in community networks. In Corelli, high capacity peers in communities autonomously build a distributed cache to dynamically pre-fetch content early on in its popularity lifecycle. By exploiting the natural proximity of peers in the community, users can gain extremely low latency access to content whilst reducing egress utilisation. Through simulation, it is shown that Corelli considerably increases accessibility and improves performance for latency dependent content. Further, Corelli is shown to offer adaptive and resilient mechanisms that ensure that it can respond to variations in churn, demand and popularity.

In this paper, we use experimental measurements to study the performance of multimedia applications over a commercial IEEE 802.16 WiMAX network. Voice-over-IP (VoIP) and video streaming applications are tested. We observe that the WiMAX-based network solidly supports VoIP. The voice quality degradation compared to high-speed Ethernet is only moderate, despite higher packet loss and network delays. Despite different characteristics of the uplink and the downlink, call quality is comparable for both directions. On-demand video streaming performs well using UDP. Smooth playback of high-quality video/audio clips at aggregate rates exceeding 700 Kbps is achieved about 63% of the time, with low-quality playback periods observed only 7% of the time. Our results show that WiMAX networks can adequately support currently popular multimedia Internet applications.

Since mobile devices are battery powered, several mobile TV standards dictate using energy saving schemes to increase the viewing time on mobile devices. The most common scheme for saving energy is to make the base station broadcast the video data of a TV channel in bursts with a bit rate much higher than the encoding rate of the video stream, which enables mobile devices to turn off their radio frequency circuits when not receiving bursts. While broadcasting bursts saves energy, it increases the channel switching delay. The switching delay is an important performance metric, because long and variable switching delays are annoying to users and may turn them away from the mobile TV service. In this paper, we first analyze the burst broadcasting scheme currently used in many deployed mobile TV networks, and we show that it is not efficient in terms of controlling the channel switching delay. We then propose new schemes to guarantee that a given maximum switching delay is not exceeded and that the energy consumption of mobile devices is minimized. We prove the correctness of the proposed schemes and derive closed-form equations for the achieved energy saving. We also implement the proposed schemes in a mobile TV testbed to show their practicability and to validate our theoretical analysis.

A distributed camera network allows for many compelling applications such as large-scale tracking or event detection. In most practical systems, resources are constrained. Although one would like to probe every camera at every time instant and store every frame, this is simply not feasible. Constraints arise from network bandwidth restrictions, I/O and disk usage from writing images, and CPU usage needed to extract features from the images. Assume that, due to resource constraints, only a subset of sensors can be probed at any given time unit. This paper examines the problem of selecting the "best" subset of sensors to probe under some user-specified objective - e.g., detecting as much motion as possible. With this objective, we would like to probe a camera when we expect motion, but would not like to waste resources on a non-active camera. The main idea behind our approach is the use of sensor semantics to guide the scheduling of resources. We learn a dynamic probabilistic model of motion correlations between cameras, and use the model to guide resource allocation for our sensor network. Although previous work has leveraged probabilistic models for sensor-scheduling, our work is distinct in its focus on real-time building-monitoring using a camera network. We validate our approach on a sensor network of a dozen cameras spread throughout a university building, recording measurements of unscripted human activity over a two week period. We automatically learnt a semantic model of typical behaviors, and show that one can significantly improve effciency of resource allocation by exploiting this model.

lThe Web is such a rich architecture that it is giving birth to new applications that were unconceivable only few years ago in the past. Developing these applications being different from developing traditional applications, generalist programming languages are not well suited. To help face this problem, we have conceived the Hop programming language whose syntax and semantics are specially crafted for programming Web applications. In order to demonstrate that Hop, and its SDK, can be used for implementing realistic applications, we have started to develop new innovative applications that extensively relies on the infrastructure offered by Web and that use Hop unique features. We have initiated this effort with a focus on multimedia applications. Using Hop we have implemented a distributed audio system. It supports a flexible architecture that allows new devices to catch up with the application any time: a cell phone can be used to pump up the volume, a PDA can be used to browse over the available musical resources, a laptop can be used to select the output speakers, etc. This application is intrinsically complex to program because, i) it is distributed (several different devices access and control shared resources such a music repositories and sound card controllers), ii) it is dynamic (new devices may join or quit the application at any time), and iii) it involves different heterogeneous devices with different hardware architectures and different capabilities. In this paper, we present the two main Hop programming forms that allow programmers to develop multimedia applications more easily and we sketch the parts of the implementation of our distributed sound system that illustrate when and why Hop helps programming Web multimedia applications.

Software-based reactive multimedia computation systems are pervasive today in desktops but also in mobile and ultra-portable devices. Most such systems offer a callback-based architecture to incorporate specific stream processing. The Synchronous Data flow model (SDF) and its variants are appropriate for many continuous stream processing problems such as the ones involving video and audio. SDF allows for static scheduling of multi-rate processing graphs therefore enabling optimal run-time efficiency. But the SDF abstraction does not adapt well to real-time processing because it lacks the notion of time: executing non-trivial schedules of multi-rate data flows in a time-triggered callback architecture, though possible through buffering, causes jitter, excessive latency and run-time inefficiencies. In this paper we formally describe a new Time-Triggered SDF (TTSDF) model with a static scheduling algorithm that produces periodic schedules than can be split among several callback activations, solving the above-mentioned problems. The model has the same expressiveness than SDF, in the sense that any graph computable by one model will also be computable by the other. Additionally, it enables parallelization and run time load balancing between callback activations.

The popularity of social media has grown dramatically over the World Wide Web. In this paper, we analyze the video popularity distribution of well-known social video websites (YouTube, Google Video, and the AOL Truveo Video Search engine) and characterize their workload. We identify trends in the categories, lengths, and formats of those videos, as well as characterize the evolution of those videos over time. We further provide an extensive analysis and comparison of video content amongst the main regions of the world.