基於省電叢集式管理架構下的多媒體傳輸快取置換策略

隨著網路的普及，改變了人們的日常生活習慣，而網路應用也與日俱增，像是Google、Bing 以及Yahoo!等搜尋引擎，FTP、P2P 分享檔案應用，MSN、Yahoo 即時通、Skype 和Face book 等廣受歡迎的社交工具以及多樣的線上遊戲和網路購物等等，換言之，我們的生活已經逐漸與網路密不可分。為了滿足人們對於網路便利性之需求，傳統的有線網路(不包含光纖網路)已經不敷使用，且慢慢被無線網路所取代，而無線網路技術也因此積極向前，因此現今無線網路的運用以及技術也被人們大大的重視；如今任意的一台筆記型電腦或是手機都能夠輕易的連上網際網路，在未來，可以預見的無線網路的影音串流運用以及點對點協定將會大行其道。然而，在這些行動裝置上仍然有些許先天上之缺點困擾著人們。這些限制就是行動裝置的電力以及記憶容量有限，這些問題使得行動電話無法承受大量的資料傳輸或造成較差的使用品質。因此，如何提供好的電力及快取記憶體管控向來是一個重要的議題。在這篇論文中，我們提供了一個新的叢集架構，此架構不只提供電力上的管控並且也有支援快取的資料置換策略；對於此叢集，我們採用了由下而上(agglomerative)的建構方式，並且藉由計算能力值的方式確立各成員在叢集中的位置與關係；在快取策略部分，我們提出了Conditioned‐LRU策略，此策略是依據H.264/SVC 特性所設計，適用於不同播放設備以及多變的網路環境所使用，其概念是將每一frame 視為一路徑上的節點，並依據資料取得成本和畫面品質計算出成本最小路徑再從此路徑中挑選適當的被刪除者。而另外一個部分是電力管控，我們結合了叢集架構的特性以及IEEE 802.11 PSM 來控制更精確的耗電量，叢集中的每一位成員都會根據自己在叢集中的位置來決定睡眠時間，通常處於管理者的成員，電力和其它外在條件都較好，因此建議可以醒久一點來為大家服務，而下面的員工則睡久一點以節省自身電量，根據此方法，我們可以獲得不錯的省電效果。

關鍵字

叢集；電源管理；快取管理；無線網路

並列摘要

Accompanying the popularization of Internet is varying the daily life of human. The application of Internet is growing strongly day by day such as the search engine – Google, Bing and Yahoo!, the application of data sharing – FTP and P2P, the popular social tool – MSN, Yahoo message, Skype and Face book, varying on‐line game and e‐shopping…etc. In other words, we already cannot live without Internet gradually. In order to satisfy the convenience that people want, the traditional wire network (not includes fiber) is superseded by the wireless network gradually and forwarding the wireless technique positively. Hence, the application and technique of wireless network are respected by people nowadays. Now, a laptop or a mobile phone can connect to Internet easily and it is bringing huge amount of different and interested applications and technique. In the future, it is foreseeable that the video streaming and P2P application will be more and more popular. However, there are still some restrictions to torment users, that is the memory and power of mobile phone is limited. These problems cause the mobile phone cannot bear to transmit and store a considerable quantity of data and make a worse quality of usage. Therefore, it is an important issue to manage power consumption and cache well. In our paper, we address novel cluster architecture. The cluster architecture not only has power management but also cache management to decide which data item has to replace. We adopt the mechanism of button‐up (agglomerative) to establish the cluster, and ensure the relationship of each member of this cluster by the capability value that every member owns. In the cache management, we design a cache replacement strategy called Conditioned‐LRU (C‐LRU). In C‐LRU, it is designed that bases on the property of H.264/SVC, suits different playbacks and varying network environments. The strategy considers the link cost and the images’ quality to replace data item. In our paper, C‐LRU uses the concept of minimal cost path to find a better candidate to replace. The other issue is the power management. We combine the cluster and IEEE 802.11 PSM to reduce more power consumptions. Each member has different sleeping time that is based on the virtual location of management. In other words, different height of management stands for different sleeping time. Usually, the manager has a better capability than employees. Hence, we suggest the manager can wake up longer to serve others and employees sleep longer to save the power itself. By this mechanism, we can obtain a not bad performance to save more power.

並列關鍵字

Cluster ； Power Management ； Cache Management ； Wireless Network

參考文獻

[25] 黃柏昌、李維聰, “應用於H.264/SVC之適應性正向錯誤修正與交錯式順序機制於無線網路之研究,” 淡江大學電機工程學系碩士班 (積體電路與計算機系統組) 碩士論文民98.

[1] J.F. Lu, J.B. Tang, Z.M. Tang, J.Y. Yang, “Hierarchical initialization approach for K-Means clustering,” Pattern Recognition Letters 29 April 2008, Pages 787-795.

[2] Chung-Horng Lung and Chenjuan Zhou , “Using hierarchical agglomerative clustering in wireless sensor networks: An energy-efficient and flexible approach,” Ad Hoc Networks 8, May 2010, Pages 328–344.

[3] H. M. N. Dilum Bandara and Anura P. Jayasumana , “An Enhanced Top-Down Cluster and Cluster Tree Formation Algorithm for Wireless Sensor Networks,” Second International Conference on Industrial and Information Systems, ICIIS 2007, Sri Lanka, 8 – 11 August 2007, Page(s): 565 - 570.

[4] Rui Xu, and Donald Wunsch II, “Survey of Clustering Algorithms,” IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL. 16, NO. 3, MAY 2005.

國際替代計量

基於省電叢集式管理架構下的多媒體傳輸快取置換策略

全文下載

主題瀏覽