利用虛擬機器進行M2M系統之效能分析

Performance Evaluation of Machine-to-Machine System with Virtual Machines

陳君翰(Chun-Han Chen)

指導教授 ： 洪士灝

國立臺灣大學/電機資訊學院/資訊網路與多媒體研究所/碩士(2012年)

https://doi.org/10.6342/NTU.2012.01945

摘要

M2M系統被廣泛的運用到許多不同的應用領域, 例如智慧型手機,監試
系統,遠端操控,遠端醫療,消費型產品等等。 M2M系統結合了許多不同
種類的系統硬體元件,並且透過網路的方式將其串連在一起。為了讓應用
程式能夠即時的完成工作,對於M2M系統來說, 效能是關鍵因素之一。對於
部份的M2M系統來說,能量消耗則是另外需要考量到的部份, 然而開發者去
建立M2M系統開發環境以及效能評估是相當複雜的。
在此論文中, 我們提出一個架構,利用虛擬機器來模擬不同的機器上執
行M2M軟體,並且進行效能分析。透過虛擬網路裝置VND來傳遞訊息在虛
擬機器之間,這樣的架構能夠幫助使用者在不需修改軟體的情況下進行模
擬,並且利用我們所提供的效能工具來分析執行時間,能量消耗以及網路傳
輸進而最佳化整個系統效能。

關鍵字

M2M通訊 ； 效能評估 ； 虛擬機器

並列摘要

Recently, machine-to-machine (M2M) systems provide services in a variety of
application domains, such as smart home, surveillance, remote control, healthcare,
consumer devices, etc. The architecture of an M2M system is usually composed
of different types of hardware components interconnected with a network. The
performance of an M2M system is a critical factor, especially for those applications
which have real-time requirements. For some M2M systems which are powered
by batteries, energy consumption is also a concern. However, the complexity of
setting up a development and performance evaluation environment for an M2M
system is quite challenging to the developersss.
In this thesis, we propose to evaluate the performance of an M2M system by
running the M2M software over the virtual machines which collectively simulate
the M2M system. The virtual machines are connected with our virtual network
devices (VND) which model the network in the M2M system. This approach al-
lows the developer to deploy unmodified software onto the simulation environment
and use our tools to analyze the execution time, energy consumption, and network
transactions on each virtual machinesss.

並列關鍵字

M2M communication ； performance estimation ； virtual mahcine

參考文獻

framework of simulation environment for wireless sensor networks with co-simulation

[3] J. Polley, D. Blazakis, J. McGee, D. Rusk, and J. Baras, “Atemu: a fine-grained sensor

[4] B. Titzer, D. Lee, and J. Palsberg, “Avrora: scalable sensor network simulation with pre-

[5] P. Boyko and A. Mazo, “Qemunet: an approach to an automated virtualized testbed,” in

T. Baumgartner, “Flexible experimentation in wireless sensor networks,” Commun. ACM,