The Research On The Key Technologies For Streaming Media Delivery In Wireless Mesh Network

Wireless Mesh Network(WMN) usually consists of the wireless backbone infrastructure andwireless end users. The resources in the wired internet can be extended by the wireless backboneinfrastructure into the wireless domain. Thanks to its low cost and quick deployment, it is quitepromising to be a prototype of the future Open Wireless Architecture. Born in the 1990's, thestreaming media delivery technology is one of the most important applications provided by theglobal information infrastructure, which gains a lot of attentions. Currently, the wireless meshnetwork has gone through several commercial trials as in Tai Wan and Philadelphia. Motivated bythe challenges from the industrial area, the academic community starts to focus on the problemsarising from the streaming media distribution in wireless mesh networks.The delivery latency is one of the most important issues.In wireless networks, due to the broadcast nature of the wireless medium (a single transmit-ting node can reach multiple one-hop neighboring nodes with a single transmission), the broad-cast becomes an efficient approach to delivery streaming media in the wireless backbone infras-tructure. The delivery latency involved mainly includes start-up delay (the construction time fora root to compute the broadcast tree, which re?ects the interval between the time it decides tobroadcast and the time the first packet is actually sent out) and broadcast latency (maximum delaybetween the transmission of a packet by a source node and its eventual reception (over multi-hoppaths) by all the intended receivers).Caching at the edge of the wireless mesh network is an efficient approach to reduce theuser's perceived delay. The user's perceived delay is the interval between the time when the userlaunches its request and the time when the user receives the response.This thesis focuses on reducing the streaming media distribution latency, regarding of thenew features of the wireless mesh network. It looks into the 3 problems: 1) low start-up delaymultirate broadcast tree construction; 2) low broadcast latency multirate broadcast tree construc-tion; 3) streaming media delivery at the edge of the network.The thesis is financial supported from the 2006 Huazhong University of Science and Tech-nology Outstanding PhD Thesis Foundation, 2006-2007 Open Foundation in the Key Lab ofthe smart internet technology in Hubei Province,"Key Technology of Cognitive Radio Networkin Ad Hoc Network"(HSIT-200603), 2005-2006"The research on the streaming media cachingnetwork based the quality and time 2 dimension property"Natural Science Foundation of Hu Bei Province (No. 2005ABA264) and 2004"The research on the interactive streaming mediadistribution technology over the internet"Natual Science Foundation of China(No.60302004).This thesis contains 4 contributions.1 A Centralized Low Start-up Latency Multi-rate Broadcast Tree ConstructionFor a given transmission power and efficiency, there exists a relationship between the trans-mission rate and its coverage. Using this relationship, based on the existing centralized construc-tion algorithm, a fast and centralized multi-rate broadcast algorithm (Rate First Algorithm, RF)isproposed, which reduces the start-up delay of the clients'requests. This algorithm can be usedin building a multi-rate broadcast tree in the case where each node in the backbone knows globaltopology information.2 A Distributed Low Broadcast Latency Multi-rate Broadcast Tree ConstructionA minimized multi-rate broadcast latency problem is proposed. To strike the balance be-tween the accuracy and efficiency. A distributed construction algorithm (Distributed Rate Firstalgorithm, DRF) is proposed. This algorithm reduces the distribution latency significantly, com-paring with the current distributed construction algorithm. This algorithm can be used to in thecase where each node in the backbone knows local topology information.3 Modeling on the Streaming Media Content Popularity DistributionDue to the differences in each individual's behavior when accessing the streaming media,the streaming media content popularity distribution is not even. By studying a real case of thestreaming media on demand web site, a K-parameter based Zipf-like distribution model is pro-posed. By analyzing the streaming media segment popularity distribution from the log files, thismodel is verified.4 Low User Perceived Latency Streaming Media Caching LayoutThe thesis takes the different demands on the playback quality from the heterogeneousclients into consideration. The quality hit ratio is proposed. The dynamic programming is usedin the proposed streaming media proxy caching layout algorithm (Layout for Quality Hit Ratio,LQHR). The simulation demonstrates that it does not only maximize the quality hit ratio, it alsoreduces the perceived latency of the clients significantly.