工作負載分享之協同運算系統：於數位浮水印濾波器評估之演算法分析的研究

Abstract

由於資源分享的能力，點對點通訊已經獲取人們的注意。這些可以被分享的資源有資訊和檔案的交換、執行程序週期、快取記憶空間和硬碟記憶容量。除此之外，點對點系統可以容易且低成本的開發。因此，傳統的主從式架構漸漸地被點對點系統所取代。 本論文所提出的工作負載分享之協同運算系統係利用點對點系統可以分享資源的優點。此協同運算系統是採用Java Servlet來完成實作，因此，點對點系統在安全、動機、實行效能和相容性等方面的缺點也可以被此協同運算系統所克服。透過這個系統，一個擁有繁複計算工作可以更快速且更有效率地被完成。 在本論文中，此協同運算系統應用在分析數位浮水印小波轉換濾波器選擇的方式來探討其強健性。根據浮水印分析範例的結果，此協同運算系統可以利用增加的速率、執行效能、運算時間和所花費的時間等觀點來衡量它的執行成果。

Peer-to-Peer (P2P) communication has attracted peoples’ attention due to its resource sharing capabilities, such as the exchanging of information and content files and sharing processing cycles, cache space, and disk storage. Moreover, P2P system can be easily implemented with low cost. Therefore, the traditional client/server architecture has been gradually replaced by P2P structure. The collaborated computing system for load sharing takes the advantages of P2P system. Furthermore, the disadvantages in P2P system, security, awareness, performance efficiency, and compatibility, have been conquered by the collaborated computing system by implementing the system using Java Servlet. Through this mechanism, a large computing task can be completed with improved efficiency. In this thesis, the collaborated computing system was utilized based on the case of analyzing the robustness of digital watermark by filter bank selection. As the result of this case, the performance can be improved in the aspect of speedup, efficiency, process time and reliability.