電腦及網路系統之新分析方法及最佳化

Abstract

本計畫的目標在於建立基礎效能分析之理論，以開發出新的分析方法及數學推導模 式來評估各種電腦及網路系統的效能。為了?成這些目標，我們利用各種新的數學方法 來進行計算及最佳化之研究。本計畫與俄羅斯之Dr. Gerasimov 進行合作，我方提供計畫的 規劃，機制之設計及模擬模式之建立，而對方則提供分析模式之建立，數學之推導及效能最佳 化之研究。本計畫將針對複雜無線網路系統進行研究，首先將以在智慧型傳輸系統 (Intelligent Transportation Systems; ITS)所執行的Push-to-Talk 機制進行分析及模擬研 究，並分別建立其分析及模擬模型以探討系統之效能。在後續的研究中，將再深入推廣 此分析及模擬模型，以應付各種複雜系統 (例：無線隨意網路及各類無線系統) 之需求。

The objective of this project is to conduct fundamental theoretical studies, to develop and construct new analytical approaches, and to derive mathematical formulae for the determination of the basic performance characteristics of computer systems and networks. In order to meet these objectives we will make use of certain new mathematical methods for the calculation, optimization and sensitivity of the performance characteristics of such systems and networks. Dr. Gerasimov has carried out preliminary investigations into the development of equations (rather than algorithms) for carrying out such analysis. Other approaches, including entropy maximization and extensions and variations of the theorems of Kovalenko and Guseinov and also the method of Marie and the shadow server method of Sevcik and Kaufmann for priority networks, will be examined. Among other it is supposed to carry out research on the following scientific directions: investigations (both in technical (estimating of characteristics and performance measures (such, for example, as throughput), estimating of reliability, estimating of performability and estimating of survivability)), and in economic planes) on new technology of wireless systems with respect to practical (that is to say from point of view it』s practical realizability), financial and market prospects of it』s employment in the Internet; research on a new original analytical queuing network mathematical model, developed as a result of research carried out within the framework of the present project, and on a proposed new original analytical approach to the analysis of microprocessor computer systems with Itanium Architecture (also called IA-64); the new architecture has 「stop bits」, so that the actual instruction word length is variable; that way, you don』t waste too much space with no-ops; developing a methodology for the optimal assignment of wavelengths in an optical network that carries voice and video connections. We will consider two versions of the problem: static and dynamic; goal is to derive fast algorithms that networks can deploy and use in real time and to minimize call blocking probabilities by means of statically or dynamically assigning paths to wavelengths based on the number of ongoing calls along the various network paths; studying the impact of service-level dynamics in bandwidth-sharing networks such as the Internet; developing approaches for quick selecting the best possible route in communication networks; the best 「route」 or 「path」 can be defined as the path with minimum delay or the path with minimum point to point time, or the path with lowest cost. We will make investigations on ad hoc wireless networks. Wireless systems can』t provide the bandwidth for individual video feeds that a fiber network can. Current satellite technology delivers a great broadcast video signal, but satellites have a problem delivering data from the home (from personal computer) back into the network (the back channel (or the feedback channel)) to provide true interactivity. A solution is to use the wireless communication networks for the back channel (or the feedback channel). With a special add-in circuit, direct-broadcast satellites can send data to personal computers as well as to television sets. Over long distances low-orbit satellites can provide transmission speeds comparable to those available on fiber. Also, the significant transmission delay associated with satellites is also overcome by means of low-orbit satellites. This issue will be under study in the project.