同儕式3D串流存取策略(Access Strategies for Peer-to-Peer 3D Streaming, ASPS)計畫探討建立一個以同儕式(Peer-to-Peer, P2P)3D物件場景串流為基礎的網路平台。目地是提供對3D物件串流(3D Streaming)有需求的系統，像是網路虛擬世界(Networked Virtual Environment, NVE) 或者是WEB 3D瀏覽器等，一個可以降低伺服器負擔的P2P網路傳輸平台，同時也提昇原本主從式架構（Client-Server Architecture，C/S）的可擴充性。我們將資料存取分為三個部份分別是：本機存取(Local)，同儕層存取(P2P)以及伺服器存取(Server)，並以階層式架構整合三層存取，藉由提昇Local層與P2P層的資料存取成功率，達到降低Server負擔的目的。我們預計以一年的時間進行下列研究： 使用者行為模式分析與分群：為了提供使用者更多的下載資料來源，除了參考虛擬世界中的使用者位置群聚外，我們將探討如何替過去、現在、或未來會走類似路徑的使用者，建立分群。希望藉由群組這些走類似路徑的使用者，提昇P2P層的下載成功率，降低伺服器的負擔。 建立同儕式網路3D物件資料串流系統：系統設計上我們分三個方向探討：1)分析使用者連線數與系統服務品質的關系，如何保證每一個連線都會有一定的頻寬傳輸資料。2)使用者在選取下載資料來源的機制。3)當使用者同時有多個資料需要下載時，哪些資料會被優先下載。 ; We propose the project Access Strategies for Peer-to-Peer 3D Streaming (ASPS), which aims to build a platform to support peer-to-peer (P2P)-based 3D scene streaming. Our goal is to provide systems that require 3D streaming, such as networked virtual environments or Web3D applications, a delivery platform that may lower the loading of the server, while improving the scalability of existing client-server architectures. We categorize content access in such applications to three hierarchically organized layers: local access, P2P access, and server access, and utilize such hierarchy to facilitate content delivery. Our goal is to lower server load by increasing the success rate of getting content from the local and P2P layers. We plan to investigate the following issues in one year: User Behavior Analysis and Clustering: To provide more potential data sources for content download, besides utilizing user position clustering, we will also investigate how to cluster the past, present or future users who travel similar paths, so that we may obtain a set of peers that has higher probability to provide content, while lowering the server’s load. Peer-to-Peer 3D Object Streaming System Construction: We will explore three potential system designs: 1) to analyze the relation between user connectivity degree and the system’s quality of service (QoS) to make sure that each connection has enough bandwidth for transmission, 2) to design mechanisms for source selection from potential provider peers, and 3) to schedule and prioritize requests for multiple data of necessity. ; 研究期間 9808 ~ 9907
