CAREER: Internet Resource Management to Deliver High Quality Live and On-demand Streaming for Wireless Clients

NSF CNS-0746649

2008 - 2013 (extended to 2015)

Introduction
Today the traffic of streaming services is dominant on the Internet. For example, YouTube gets more than 100 million video accesses per day. Many P2P-based systems, such as PPLive and PPStream, are delivering streaming content to millions of users daily. In addition to entertainment, Internet streaming is widely used in many other applications and services, such as video tele-conferencing, tele-nursing, and tele-medicine. In contrast to the rich and fast growing resources available on the Internet, such as bandwidth, CPU cycles, and streaming content, wireless users still have difficulties in receiving high quality streaming services on their wireless devices.

In this CAREER project, we have identified three related critical problems and laid out a detailed research and education plan for five years to fundamentally address them for both typical server-based on-demand streaming and P2P-based live streaming through efficient Internet resource management. The three issues to be considered are power saving , device heterogeneity, and security protection for wireless clients. The main objective of the project is to effectively utilize the abundant under-utilized or idle resources on the Internet to improve the streaming performance on resource-poor wireless devices. The intellectual challenges of the proposed project are as follows.

First, the battery power is a major and critical constraint for a wireless device. With the best existing power saving schemes, the limited battery capacity can only support a short period of high quality streaming services. Studies have shown that the wireless network interface card for data transmission consumes a significant portion of energy. However, existing power saving schemes on the network interface card commonly trade off the application's performance for power saving, which is detrimental to QoS of Internet streaming. In this project, we propose to leverage the under-utilized Internet resources (e.g., bandwidth) and application communication patterns to minimize the power consumption for data transmissions without degrading the application's performance.

Second, due to multi-dimensional heterogeneity of wireless devices, such as screen size, bandwidth capacity, and color depth, most existing Internet streaming content cannot be directly delivered to various wireless devices. Efficiently addressing multi-dimensional heterogeneity issues often requires runtime content adaptation on a dedicated server that can afford a significant amount of CPU cycles. Existing studies have focused on developing caching strategies to trade storage for CPU cycles on the content adaptation server for on-demand streaming. However, in a P2P streaming system, such dedicated infrastructure support is practically impossible. In this project, we propose to effectively leverage idle peer computing resources and novel distributed content adaptation schemes to address the multi-dimensional heterogeneity problem.

Finally, due to the nature of wireless communications and the resource constraint on wireless devices, various and fatal attacks can be easily mounted. With the increasing security demand from users and Internet streaming applications, directly applying existing security mechanisms have high overhead and little consideration about the resource constraint on wireless devices. Aiming to reduce the power consumption while providing robust and power-efficient wireless Internet streaming services, we propose to leverage the unused covert communication channel and application characteristics to design a set of new authentication and encryption schemes to satisfy the demand of future wireless Internet streaming services.

Project Personnel
PI:Songqing Chen
Yao Liu: Research Assistant. (Defended dissertation April 2013. Now an Assistant Professor at SUNY-Binghamton.)
Mohammed Hassan: Research Asssitant. (Senior Ph.D. student.)
Mengbai Xiao: Research Assistant. (Junior Ph.D. student.)

Relevant Publications

Journal Publications

Conference/Workshop Publications