US-Egypt Cooperative Research: Untying the Gordian Knot: Practical and Distributed Optimal Scheduling to Support Multimedia Traffic over Wireless Networks

该项目支持俄亥俄州哥伦布市俄亥俄州立大学的Eylem Ekici博士和埃及亚历山大市埃及-日本科技大学的穆斯塔法·优素福博士的合作研究工作。他们计划研究“实用和分布式的优化调度，以支持无线网络上的多媒体流量。”通信和网络已成为现代生活中密不可分的组成部分。随着移动和联网设备的日益普及，大多数可用的服务在很大程度上依赖于通过联网信息基础设施获得或提供的信息。该项目将为无线网络设计分布式调度算法，以提高效率和“体验质量”，特别是对于需要大量带宽的流媒体应用。该项目基于使用可用的无线资源，但使用更高效的信息传输算法，以确保现有无线网络得到最大限度的利用。随着移动设备的使用继续扩大，这项研究在世界各地都有明显的好处，特别是可能对实时多媒体交付活动，如视频流、视频会议和分布式监控。该项目包括一项明确的计划，将研究工作纳入参与大学学生的教育项目，可能会带来显著的好处。教育计划包括技术转让、学生培训、面向高中生的推广和国际合作。美国PI将利用他所在大学的各种现有项目，从代表性不足的群体中招募合格的女性和学生进行研究。最近，两个重要特征已成为当前以及可能的未来网络环境的显著特征：第一，多媒体流量(例如多媒体播客、广播、交互式实时通信应用程序)在通信网络负载中所占的比例越来越大；第二，连接包含至少一跳的连接，这些连接通过多个用户共享的无线链路传输。无线网络上多媒体服务的广泛使用正在对可用带宽造成压力，并影响用户的体验质量(QOE)，特别是对于流媒体应用。最近对吞吐量最优分布式方案的时延性能的研究普遍表明，它们的时延特性对于服务多媒体业务是不可接受的。这表明需要针对与多媒体流量相关联的QOE度量的替代分布式解决方案。本项目将设计无线网络的分布式调度算法，并在实际平台上实现算法，并评估不同的实际问题对算法性能的影响。研究将是：(1)增加在每个流(而不是聚合)的基础上建立多个优化标准的能力，(2)将算法扩展到拓扑中的多跳，(3)纠正抖动和长期公平性的权衡，(4)考虑流生命周期的有限性和可变性，以及(5)确保所产生的去中心化算法的实际实施。这些目标包括一个具有挑战性的研究议程，而且研究既是渐进的，也是高风险的。美国的PI将专注于设计和理论分析，埃及的PI将专注于实施和原型。该项目由美国-埃及联合科技基金项目资助。这些合作项目的美方支持由美国国务院提供给美国国家科学基金会。埃及政府为埃及方面的合作提供支持。

This project supports a cooperative research effort by Dr. Eylem Ekici of Ohio State University, Columbus, OH, and Dr. Moustafa Youssef of the Egypt-Japan University of Science and Technology, Alexandria, Egypt. They plan to study "practical and distributed optimal scheduling to support multimedia traffic over wireless networks." Communications and networking have become inseparable components of modern life. With the ever-increasing penetration of mobile and networked devices, the majority of available services rely heavily on information obtained or delivered through a networked information infrastructure. This project will design distributed scheduling algorithms for wireless networks that will increase efficiency and "quality of experience", especially for streaming multimedia applications that demand significant bandwidth. The project is based on using available wireless resources but with more efficient information transfer algorithms to ensure existing wireless networks are used to their fullest extent. This research has clear benefits around the world as the use of mobile devices continues to expand, with specific benefits likely for real-time multimedia delivery activities such as video streaming, video conferencing and distributed surveillance. The project includes an explicit plan to incorporate the research work into educational programs for students at the participating universities, with signicant benefits likely. The education plan includes technology transfer, student training, outreach to high school students, and international collaboration. The US PI will leverage various established programs at his university to recruit qualified women and students from underrepresented groups for the research.Recently, two important properties have emerged as prominent features of current, and likely future, networking environments: First, multimedia traffic (e.g. multimedia podcasts, broadcasts, interactive real time communication applications) constitute an increasing fraction of the communication network load; second, there is a significant shift to connections that contain at least one hop delivered over wireless links that are shared by multiple users. The widespread use of multimedia services over wireless networks is causing stress on the available bandwidth and is affecting the quality of experience (QoE) for users, especially with streaming multimedia applications. Recent studies of the delay performance of throughput-optimal distributed schemes have generally revealed that their delay characteristics are unacceptable for serving multimedia traffic. This indicates a need for alternative distributed solutions geared towards the QoE metrics associated with multimedia traffic. This project will design distributed scheduling algorithms for wireless networks, and will implement the algorithms on real platforms and evaluate the effect of different practical issues on the performance of the algorithms. The research will be: (1) to add the ability to establish multiple optimization criteria on a per-flow (as opposed to an aggregate) basis, (2) to extend the algorithm to multiple hops in a topology, (3) to rectify tradeoffs in jitter and long-term fairness, (4) to account for finiteness and variability in flow lifetimes, and (5) to ensure practical implementation of the resulting decentralized algorithm. The objectives encompass a challenging research agenda, and the research is both incremental and high risk. The US PI will focus on design and theoretical analysis, and the Egyptian PI to focus on implementation and prototyping. This project is funded through the US-Egypt Joint Science and Technology Fund Program. Support for the U.S. side of these cooperative projects is provided to the National Science Foundation by the U.S. Department of State. The Egyptian Government provides support for the Egyptian side of the collaboration.