Error-Correction and Distributed-Storage for Wireless Video Streaming

视频被认为是未来无线通信系统的杀手级应用。视频流量在过去几年中急剧增加，预计未来将继续保持类似的趋势。拟议的研究将为点播和现场视频的存储和通信开发高效和可扩展的机制。将为我们提出的解决方案开发用于身份验证、质量控制和安全的新技术。拟议的研究分为三个主题，每个主题都解决了当今无线网络中的一些直接挑战。本研究的第一个课题将探讨串流纠错码的理论与实务。前向纠错（FEC）码提供了一种自然的机制来保护视频数据包免受通信通道上的错误。与已经研究了几十年的经典FEC相比，低延迟、实时、视频流应用的纠错仍然是一个丰富的研究领域。将研究流码的理论性质，并将其与卷积码的动力系统理论联系起来。在实际应用方面，将根据实际应用开发新的源和通道模型，并在实际环境中评估性能增益。最终，这种流代码可以显著提高高端视频会议系统的效率，并使小型企业主能够负担得起。第二个专题将开发一种可扩展的视频分发机制，利用将在下一代接入点和基站中提供的存储。一个特别的焦点将放在HTTP自适应比特率流的最佳存储上——这是交付点播视频的主要机制。我们提出的技术将在本地测试平台上进行评估，该测试平台由软件定义的无线电和多伦多大学的计算机集群组成，以及由加拿大几所大学的计算节点组成的国家测试平台。拟议的研究将利用未开发的资源——接入点和基站中的缓存——来显著减少由于移动流量爆炸性增长而导致的拥塞，并将为移动运营商提供一种新的尖端技术。第三个主题将集中于视频分发系统中的安全和隐私机制。分布式源代码的原理将用于在这些系统中开发实时认证和质量控制机制。在不同的方向上，我们还将开发在智能电表网络中提供隐私的新方法。智能电表报告实时的用电量测量，这可能会泄露终端用户的敏感信息。为了解决这个问题，将开发一种新技术，通过使用本地储能来掩盖公用事业公司的电力消耗。这种技术可以在诸如插电式混合动力汽车之类的电器中实现，这些汽车已经支持在家庭中内置的能量存储机制。拟议研究中的HQP将获得当今无线行业面临的一些最紧迫挑战的专业知识：(1)高效和可扩展的视频流量分配；(2)安全和隐私。这种培训将使HQP能够为无线和多媒体的标准化活动以及为下一代通信系统开发算法和体系结构做出贡献。提议的主题与PI的长期研究计划密切相关，该计划旨在基于健全的工程原则，创建一个无处不在、可靠和值得信赖的数据网络，以支持我们的经济和国家基础设施。

Video is considered to be the killer application in future wireless communication systems. Video traffic has dramatically increased in the last few years, and is projected to continue a similar trend in the future. The proposed research will develop highly efficient and scalable mechanisms for storage and communication of both on-demand and live-video. Novel techniques for authentication, quality-control, and security will be developed for our proposed solutions. The proposed research is divided into three topics, each of which addresses some immediate challenges in today's wireless networks.The first topic of the proposed research will investigate the theory and practice of streaming error-correction codes. Forward Error Correction (FEC) codes provide a natural mechanism for protecting video packets against errors over communication channels. In contrast to classical FEC, that have been studied for several decades, error correction for low-delay, real-time, video streaming applications remains a fertile area of research. Theoretical properties of streaming-codes will be studied and connections to the dynamical systems theory of convolutional codes will be made. On the practical side, new source and channel models will be developed based upon real-world applications and performance gains will be evaluated in realistic environments. Ultimately such streaming-codes could significantly increase the efficiency of high-end video conferencing systems, and make them affordable to small business owners.The second topic will develop a scalable mechanism for the distribution of video, using storage that will be available in next generation access points and base-stations. A particular focus will be on the optimal storage for HTTP Adaptive Bit-Rate Streaming --- the dominant mechanism for delivering on-demand video. Our proposed techniques will be assessed on a local testbed, consisting of software defined radios and a computer cluster at U of T, as well as a national testbed consisting of computing nodes at several universities across Canada. The proposed research will exploit an untapped resource --- caching in access points and base-stations --- to significantly reduce the congestion due to the explosive growth in mobile traffic and will provide a new cutting edge technology for mobile operators.The third topic will focus on security and privacy mechanisms in video distribution systems. Principles from distributed source coding will be used to develop a real-time authentication and quality-control mechanism in these systems. In a different direction, we will also develop novel approaches for providing privacy in a network of smart-meters. Smart-meters report real-time measurements of electricity usage, which can leak sensitive information about end users. To address this, novel techniques that mask the electricity consumption from utility companies by using local energy storage will be developed. Such techniques may be implemented in appliances such as plug-in hybrid vehicles, which already support built in mechanisms for energy storage at homes. The HQP in the proposed research will obtain expertise in some of the most pressing challenges faced by today's wireless industry: (1) efficient and scalable distribution of video traffic, and (2) security and privacy. Such a training will enable the HQP to contribute to standardization activities in wireless and multimedia, as well as in developing algorithms and architectures for next generation communication systems. The proposed topics closely align with the PI's long-term research program of creating a ubiquitous, reliable, and trustworthy data network, based on sound engineering principles, to support our economy and national infrastructure.