臺灣博碩士論文加值系統

隨著硬體成本的不斷下降，智慧型手機逐漸的普及在我們的生活中，智慧型手機所使用的作業系統也百花齊放，其中又以Android 推動的裝置佔了絕大多數，在當前Android的系統架構設計下，所有Android Framework中應用程式與應用程式間，應用程式與系統服務間，系統服務與系統服務間的IPC ( Inter Process Communication ) [1] 皆是以Open Binder[2]架構為核心設計的Binder Framework運作，此一行為在系統中所佔角色之重要，與其他作業系統迥然不同。本研究基於Binder Framework的運作具有區域性(Locality)特性的假設，實作以伸展樹(Splay-Tree)[3]演算法為主軸的Binder Driver，取代原本紅黑樹(Red-Black Tree)[4]的架構，去驗證是否能改善Binder Framework運作的效能。實驗結果顯示出，Android系統本身的基本System Service便具有相當的區域性，如在特定的使用習慣環境下，以Splay-Tree實作的Binder Driver，改善的幅度更顯著．以目前Android裝置的普及化，若能將此一架構的改變推廣，當可節省巨量的運算能量。

Smart Phones, powered by miscellaneous mobile operating systems, is getting common in our lives. This trend comes with descending of hardware cost. Most of Smart Phone devices are powered by Google Android today. In Android architecture design. The importance of IPC (Inter Process Communication)[1] takes a significant role than any other modern mobile operating system. All IPCs , including application/application, application/system service, system service/system service, happens in Android-Framework are based Binder Framework, which is a descendant of Open Binder[2]. Our assumption is that Binder Framework has locality. If replacing the Red-Black Tree[4] based node managing system by Splay-Tree[3]. The performance should be better. This papper evaluates that the performance of connectivity of IPC node generated by Android System Service has been improved. Consider the popularity of Android devices today. It may save huge amount of computing resources.

目錄
摘要 i
Abstract ii
圖目錄 v
第一章　緒論 1
1-1 Open Binder 簡介 1
1-2 Android Binder Framework簡介 2
1-3 研究目標 3
1-4 研究貢獻 3
1-5 論文文章架構 4
第二章　研究背景 5
3-1 紅黑樹(Red-Black Tree)[4] 5
2-1-1 紅黑樹的插入操作 5
2-1-2 紅黑樹的刪除操作 7
3-2 伸展樹(Splay-Tree)[3] 9
2-2-1 伸展樹的旋轉操作 9
第三章　Splay-Tree Based Android Binder Framework 12
3-1 結構 12
3-2 運作流程 12
3-2-1 New Data 13
3-2-2 Access 14
3-3 實作方法 15
3-4 實驗方式 16
第四章　實驗結果 17
4-1 實驗環境與實驗資料集 17
4-2-1 實驗環境1 17
4-2-2 實驗環境2 17
4-2 實驗環境1之實驗結果 19
4-2-1 Cost - Idle狀態 19
4-3-2 Cost – 8 Random application process bind/unbind 20
4-3-3 Cost – Overall Comparison 22
4-3-4 Response Time 23
4-3-5 Response Time - Improvement 23
4-3 實驗環境2之實驗結果 24
4-3-2 Cost– Idle狀態 24
4-3-3 Cost – 8 Random application processes bind/unbind 25
4-3-4 Cost – Overall comparison 27
4-3-5 Response Time 28
4-3-6 Response Time - Improvement 28
4-3-7 binderAddInts 測試結果 29
4-3-8 資料分析 30
第五章　結論與未來展望 31
參考文獻 32

[1] Jochen Liedtke. “On µ-Kernel Construction”, Proc. 15th ACM Symposium on Operating System Principles (SOSP), December 1995
[2] OpenBinder website: http://www.angryredplanet.com/~hackbod/openbinder/docs/html/
[3] Sleator, Daniel D.; Tarjan, Robert E. (1985), "Self-Adjusting Binary Search Trees", Journal of the ACM (Association for Computing Machinery) 32 (3): 652–686, doi:10.1145/3828.3835.
[4] Rudolf Bayer (1972). "Symmetric binary B-Trees: Data structure and maintenance algorithms". Acta Informatica 1 (4): 290–306. doi:10.1007/BF00289509.
[5] CORBA: http://www.omg.org/spec/CORBA/
[6] Erika Chin, Adrienne Porter Felt, Kate Greenwood, David Wagner . “Analyzing inter-application communication in Android”. MobiSys '11: Proceedings of the 9th international conference on Mobile systems, applications, and services. 2011
[7] https://lkml.org/lkml/2009/6/25/3
[8] CORBA IDL Language Mapping. http://www.omg.org/spec/
[9] Wei Jiang, Chen Ding, Roland Cheng. “Memory access analysis and optimization approaches on splay trees”. LCR '04 Proceedings of the 7th workshop on Workshop on languages, compilers, and run-time support for scalable systems.2004