基于多层增量式可扩展扇形兴趣区域的大规模DVE场景对等渐进式传输机制  

Multi-Layered Incremental &Scalable Sector of Interest(MISSOI)Based Efficient Progressive Transmission of Large-Scale DVE Scenes

作　　者：贾金原[1] 王伟[2] 王明飞[1] 范辰[1] 张晨曦[1] 俞阳[1] 

机构地区：[1]同济大学软件学院,上海201804 [2]中国科学院云计算产业技术创新与育成中心,广东东莞523808

出　　处：《计算机学报》2014年第6期1324-1334,共11页Chinese Journal of Computers

基　　金：国家自然科学基金面上项目(61272276);国家"九七三"重点基础研究发展规划项目基金(2010CB328106);国家十二五计划重大科技支撑项目(2012BAC11B00-04-03);高等学校博士学科点专项科研基金项目(20130072110035);吉林省重点科技攻关课题(20140204088GX);长春高新区长白慧谷英才计划(3-2013006)资助~~

摘　　要：为了解决在有限的互联网带宽上高效下载海量虚拟场景的"瓶颈"问题,文中针对基于对等网络(P2P)的虚拟场景渐进式传输的增量判断和场景下载源节点最优选择这两个关键点,首先将传统的兴趣域(Area of Interests,AOI)细化为兴趣扇区域(Sector of Interest,SOI),又将其提升为多层增量式可扩展兴趣扇形区域(Multi-Layered&Incrementally Scalable SOI,MISSOI),给出了MISSOI的分划扩展算法,并基于MISSOI提出了一个高效的整数级SOI场景下载的拾取算法;然后设计了一个基于对等网的海量DVE场景渐进式下载框架,并给出了在P2P-DVE中快速搜索场景下载源节点以及选择最优场景下载源节点的高效算法.仿真实验结果表明作者提出的方法在可视场景下载域的拾取、下载源查询成功率和下载请求失败率等性能指标上明显好于目前主流海量虚拟场景P2PDVE传输机制FLoD,更是远优于传统的Client/Server虚拟场景传输模式.This paper proposes a new P2P based progressive transmission strategy for downloading huge DVE scenes interactively via limited bandwidth on Internet,aiming at solving these two key points：the virtual scene incremental judgments of progressive transmission and optimal choice of downloading source peers base on P2P network.Firstly,subdivides traditional AOI into SOI （Sector of Interest）,then upgrades it to MISSOI （Multi Layered ＆ Incrementally Scalable SOI）,the algorithms of partition MISSOI adaptively and pickup integer level downloading SOI scene efficiently are proposed subsequently.Secondly,designs a progressive download framework of massive DVE scene based on P2P network,simultaneously,proposes an efficient algorithm that search downloading source peers quickly and chooses downloading source peers optimally.Experimental results show that our method outperforms FLoD and outclasses C/S mode in terms of downloading source search success ratio,downloading request failure ratio and server downloading ratio.

关 键 词：分布式虚拟现实 对等网络 渐进式传输 下载增量 多层增量式可扩展兴趣扇形区域 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]