一個程式被切割成數個小程式並被分配到不同的處理單元上以節省程式執行的時間，許多的研究成果已證明利用平行計算方法可提高處理效能。然而無線網路特殊的性質，像是電腦的移動能力、不定時的斷線以及電池的使用時間，影響平行處理應有的效能。因此這篇論文架構了一個分散式的無線網路環境，並在此環境上對分散式平行計算做效能評估。此論文利用了兩種網路傳輸機制並且模擬了無線網路上所常碰到的斷線問題。從實驗結果可以證明使用預先傳送機制、避免無線網路上頻寬的競爭以及動態調整發配的工作量，的確可以減少整體系統執行時間。

Parallel computing is a common technique to reduce execution time in distributed systems. An application is divided into several subtasks that can be executed on a set of computers. Numerous protocols and experiments have been measured on wired network environments. However, little attention was devoted to wireless networks. There exist characteristics that are unusual in wired environments, such as limited bandwidth, frequent disconnection,low power, and mobility. Due to the unique attributes, the evaluation for distributed applications on wireless environment becomes necessary. This thesis demonstrates the performance measurements for distributed applications in the wireless environment. It compares two transmitting mechanisms and simulates the disconnection that is frequent occurred in the wireless network. The results show that avoiding network contention and dispatching subtasks dynamically improve the system performance.

1. Introduction....................................................1
1.1 Distributed Systems........................................2
1.2 Mobile Computing and Networks..............................3
1.3 Parallel Computing.........................................3
1.4 IEEE 802.11 Wireless LAN...................................4
1.5 Contribution...............................................5
1.6 Organization of This Thesis................................5

2. Related Work....................................................7
2.1 Resource Management........................................7
2.2 Adaptive Methodology for Wireless Environment..............7
2.3 Parallel Languages and Compilers...........................8

3. Adaptive Transmission Techniques and Disconnection Simulation...9
3.1 Token-based Transmission Technique.........................10
3.2 Pre-sending Technique......................................11
3.3 Disconnection Simulation...................................13

4. Performance Evaluation..........................................15
4.1 Experimental Environment...................................15
4.2 Experimental Results.......................................18

5. Conclusion......................................................26

Reference..........................................................27

[1] George H. Forman and John Zahorjan, "The Challenges of Mobile Computing,"
Computer, pp. 38-47, Apr. 1994.
[2] Kuo-Feng Ssu and Bin Yao and W. Kent Fuchs and Nuno Ferreira Neves, "Adaptive Checkpointing with Storage Management for Mobile Environments," IEEE Trans on Reliability, vol. 48, no. 4, pp. 315-324, Dec. 1999.
[3] A. S. Tanenbaum and M. V. Steen, Distributed Systems Principles and Paradigms. Prentice Hall, 2002.
[4] G. Coulouris and J. Dollimore and T. Kindberg, Distributed Systems Concepts and Design. Addison-Wesley, 2001.
[5] G. S. Almasi and A. Gottlieb, Highly Parallel Computing. The Benjamin/Cummings Publishing Company, 1994.
[6] National Center for High-Performance Computing, http://www.nchc.gov.tw/.
[7] D. L. Calli, Distributed Operating System. Prentice Hall. 2000.
[8] A. Hills, "Large-scale Wireless LAN Design," IEEE Communications Maganize,
vol. 39, pp. 98-107, Nov. 2001
[9] B. P. Crow and I. Widjaja and L. G. Kim and P. T. Sakai, "IEEE 802.11 Wireless Local Area Networks," IEEE Communications Magazine, vol.35, pp.116-126, Sept. 1997.
[10] D. E. Comer, Internetworking with TCP/IP. Prentice Hall, 2000.
[11] S. Pope, "Application Migration for Mobile Computers," Proceedings of Third International Workshop on Services in Distributed and Networked Environments, pp.20-26, Jun. 1996.
[12] The D System, http://www.cs.rice.edu/~dsystem/.
[13] Phone Lin and Y. B. Lin and J. Y. Jeng, "Improving GSM Call Completion by Call Reestablishment," IEEE Journal on Selected Areas in Communications, vol. 17, pp. 1305-1317, July 1999.