分散式無線多輸出入多媒體通訊系統---子計畫五：分散式無線多輸出入環境之無線資源管理及行動計算

Abstract

Wimax 的寬頻及行動管理，在多媒體通訊的時代，提供了即時的通訊能力。而主從架構 client/server，一直是一個廣泛應用的計算平台。因此本計畫探討如何利用Wimax 來 建構以主從架構為基礎的即時行動計算。在client/server 架構上的多媒體通訊、應用 軟體有很多種，包括media player,IE, office 等等。這些不同的應用同時會產生不同 特性的訊號要傳送、包括圖形、文字、影像，等等。因此在4G 等級的通訊系統也開始 對不同的訊號提供不同QoS 的服務流。本研究計畫的目標就是要來探討如何建構無線通 訊的資源管理及資料排程以提供不同的訊號所要的品質跟QoS。 輕型的行動計算是以終端機連線為主。終端機連線是把整個server 的桌面訊息傳送出 去、讓client 端可以看到這個server 上的所有應用程式，再把client 端的按鍵傳回 server 來啟動各種應用軟體。通常桌面訊息是由很多不同的應用軟體所產生，它們產出 的訊息會顯現在桌面上各個角落，而遠端桌面只把最後合成的畫面以一張影像傳送出 去，為了因應桌面上各個角落所要的不同QoS 的服務，我們需要把桌面上各個位置所對 應的應用軟體確定出來，並把它們所產生的資料切割出來，以提供順暢的行動計算。目 前大部分的client / server 連線軟體，像是RDP、VNC、ICA 並沒有這種功能，因此從 桌面上切割訊息來改進這些軟體的功能，也是本計畫的重點，尤其是切割出video 來。 最近雲端計算的觀念開始萌芽，為了使行動端愈省電、愈方便，應用遠端桌面這種架構 的行動計算應該會愈來愈普遍。這種應用只傳送整個桌面的訊息、所需要的計算都在 server 上進行。這種client/server 的通訊方式有保密的優點、在一般辦公室系統已經 很廣泛使用。它的另一個優點是client 可以很簡單、只負責訊息的顯現。目前很多嵌 入式系統都朝這個方向在發展。配合4G 及Wimax 的出現、行動式thin client 也開始 引人重視。利用行動式thin client 透過Wimax 連線到辦公室或是家中的電腦，不管何 時何地都能處理事情。Client 只需負責兩件事情，第一、接收處理後的圖形結果並顯示 在螢幕上，第二、將輸入傳遞給遠端的Server 以控制server 的多媒體應用軟體。因此 管理好終端機連線傳送的訊息、就能提升行動計算的效率。而這個效率決定於兩個因 素，一是訊號處理，另一個是無線傳輸的即時性規劃。 現在的遠端桌面連線系統，是透過space domain 的彩度壓縮方式來傳送訊息，這種方 式是針對圖形、文字來設計。也就是說傳送過來的圖形、文字或許彩度不高，但是其文 字必須保持清晰。但是有影像訊號出現時、它的效率就變差了。目前影像的壓縮幾乎是 利用frequency domain 的MPEG 方式。因此如何整合兩個不同空間的壓縮方式於桌面連 線也是本研究計畫的一個主題。

The bandwidth and mobility management of the Wimax system have provided a feasible platform for mobile computing. Client/server architecture has been a widely used computing platform. In this project, we investigate the application of the Wimax system for real time client/server based mobile computing. There are many different kinds of application softwares that are used in the client/server based multimedia communication such as media player, IE, office etc. These different applications will generate many different kinds of information for transmission simultaneously such as graphics, text, video. Therefore in the 4G communication system, different service flows with different QOS are begun to be considered for various kinds of signals. Our goals are to investigate how to construct the scheduling algorithms of the wireless communication system in order to provide different QOS for the various kinds of signals transmission during mobile computing. Thin type of mobile computing is based on terminal service. Terminal service transfers the desktop of the server to the client such that the client can see all the application softwares of the server. The key stroke and mouse click of the client is then sent back to the server to activate the application softwares. The desktop is composed of many different kinds of signals generated by different kinds of applications. To meet the different QOS requirements, segmentation of the desktop image is required. In most terminal service such as RDP, VNC, ICA etc, the desktop is transmitted as a whole picture. Therefore to detect the output area of the different applications such that the whole picture can be segmented is essential to provide smooth mobile computing. Recently, the concept of cloud computing has become popular. This will make thin mobile computing more feasible, since only the desktop is transmitted to the client, all the computing is done in the server. Only the desktop image is transmitted, not the actual signal code, this will fulfill the purpose of information security. Since only graphics and video rendering is needed, embedded system has been developed for this purpose. With the introduction of 4G and Wimax communication system, mobile thin client has begun to attract the attention. Therefore efficient terminal service over wireless communication system becomes essential for the mobile computing to be feasible. Most of the terminal service is aimed at graphics and text transmission. In order to maintain the sharpness and clarity of the signal, space domain compression is used. This is different from the methods used for video compression. In video compression, high frequency components are discarded. Since desktop image contains video information in many multimedia applications, space domain compression is not efficient for video signal. Thus to detect video signal and combine the two different compression methods is an important issue for thin mobile computing.