隨著智慧型個人裝置在硬體上的進步，智慧型手機在市場的佔有率有了顯著的提升。然而，在硬體資源有限的嵌入式裝置上，記憶體的管理成為一項重要的議題。在本篇論文中，我們將討論現有架構上，記憶體管理機制的各項優缺點，藉由實測來驗證這些優缺點，並闡述可用記憶體與應用程式啟用時間之間的關聯性及如何互相影響。最後，以手機應用程式為運作平台，實作一個以類神經網路為核心機制的記憶體回收方式。經過我們的實作以及驗證，該機制可以類神經網路為基礎，在限定的資料範圍內，預測使用者近期可能開啟的應用程式，將其保留於記憶體空間內，並關閉短期內不會被開啟的應用程式，藉此優化記憶體的可支配空間，並且達到縮短程式啟用時間的目的。在這項實驗中，我們的預測機制，可以在數個時間單位內達到76.3%的預測準確率，並縮短16.04%的程式啟用時間。

Memory management is an important issue nowadays on embedded systems. In the thesis, first, we discuss the benefit and weakness of the conventional memory reclaim mechanism. Second, we measure the memory behavior to observe how it affects the application launch time. Finally, we propose a memory reclaim mechanism which is implemented as an application. The application maintains a service to reclaim the memory by applying a neural network deciding mechanism. The proposed mechanism is able to predict the upcoming application and reclaim the application which will not resume in a short time. In our experiment, the accuracy of prediction is 76.3% approximately. By adopting the neural network, the proposed mechanism is able to speed up user-experienced application launch time by 16.04% in average.

[1] Android. Google Inc., [Online]. Available: https://developer.android.com/index.html
[2] Android Open Source Project. Google Inc., [Online]. Available: https://source.android.com
[3] Google Play Store. Google Inc., [Online]. Available: https://play.google.com/
[4] Clean Master. Cheetah Mobile. Inc., Available: https://www.cmcm.com/zh-tw/clean-master/
[5] Android Run Time and Dalvik. Google Inc., [Online]. Available: https://source.android.com/devices/tech/dalvik/index.html
[6] Hossain Shahriar, Sarah North and Edward Mawangi, “Testing of Memory Leak in Android Applications,” IEEE 15th International Symposium on High-Assurance Systems Engineering, pp. 176 - 183, Jan. 2014.
[7] Android Interfaces and Architecture. Google Inc. [Online]. Available: https://source.android.com/devices/
[8] Application Program Interface Guide. Google Inc., [Online]. Available: https://developer.android.com/guide
[9] Hyun-Joo Yoo, Seong-jeen Kim and Min-soo Jung. “Study of Garbage Collection Performance on Dalvik VM Heap Considering Real-Time Response,” International Conference on IT Convergence and Security (ICITCS), pp. 1 - 3, Dec. 2013.
[10] Low Memory Killer, Google Inc., [Online]. Available: https://android.googlesource.com/kernel/common.git/+/android-3.4/drivers/staging/android/lowmemorykiller.c
[11] Geunsik Lim, Changwoo Min and Young Ik Eom, “Enhancing application performance by memory partitioning in Android platforms,” IEEE International Conference on Consumer Electronics (ICCE), pp. 649 - 650, Jan. 2013.
[12] Clayton Shepard, Ahmad Rahmati, Chad Tossell, Lin Zhong and Phillip Kortum “LiveLab: measuring wireless networks and smartphone users in the field,” ACM SIGMETRICS Performance Evaluation Review, vol. 38, pp. 15 - 20, Dec. 2010.
[13] Balakrishnan Jayavel, Subbaramaiah Mandava and Jyoti Johri, “Enhanced page Reclaim for Android devices,” Eighth International Conference on Contemporary Computing (IC3), 459 - 462, Aug. 2015.
[14] Kumar Vimal and Aditya Trivedi, “A memory management scheme for enhancing performance of applications on Android,” IEEE Recent Advances in Intelligent Computational Systems (RAICS), pp. 162 - 166, Dec. 2015.
[15] Hyojong Kim, Hongyeol Lim, Dilan Manatunga, Hyesoon Kim and Gi-Ho Park, “Accelerating Application Start-up with Nonvolatile Memory in Android Systems,” IEEE Micro, Vol. 35, pp. 15 - 25, Jan. 2015.
[16] Processes and Application Life Cycle. Google Inc., [Online], Available: https://developer.android.com/guide/topics/processes/process-lifecycle.html
[17] Kyosuke Nagata, Yuta Nakamura, Shun Nomura and Saneyasu Yamaguchi, “Measuring and Improving Application Launching Performance on Android Devices,” First International Symposium on Computing and Networking, pp. 636 - 638, Dec. 2013.
[18] Sang-Hoon Kim, Jin-Soo Kim and Seungryoul Maeng, “SmartLMK: A Memory Reclamation Scheme for Improving User-Perceived App Launch Time,” ACM Transactions on Embedded Computing Systems (TECS),Vol. 15.
[19] ARM DS-5 Streamline Performance Analyzer. ARM Inc., [Online]. Available: https://developer.arm.com/products/software-development-tools/ds-5-development-studio/streamline
[20] Android Debug Bridge. Google Inc., [Online]. Available: https://developer.android.com/studio/command-line/adb.html
[21] Donghee Lee, Jongmoo Choi, Jong-Hun Kim and S. H. Noh, Sang Lyul Min, Yookun Cho and Chong Sang Kim, “LRFU: a spectrum of policies that subsumes the least recently used and least frequently used policies,” IEEE Transactions on Computers, Vol. 50, pp. 1352 - 1361, Aug. 2001
[22] W. Stallings et al., Operating Systems: Internals and Design Principles, 7th Edition, 2012, ISBN: 978-0-1338-0591-8
[23] Hossein Falaki, Ratul Mahajan, Dimitrios Lymberopoulos, Ramesh Govindan and Deborah Estrin, “Diversity in Smartphone Usage”, International Conference on Mobile Systems, Applications, and Services (MobSys ‘10), pp. 179 – 194, 2010.
[24] Nathan Eagle and Alex (Sandy) Pentland, “Reality Mining: Sensing Complex Social System,” Personal and Ubiquitous Computing, Vol. 10, no. 4, pp. 255 – 268, Mar. 2006.
[25] Jiangchuan Zheng and Lionel M. Ni, “An Unsupervised Framework for Sensing Individual and Cluster Behavior Patterns from Human Mobile Data,” ACM Conference on Ubiquitous Computing (UbiComp ‘12), pp. 153 – 162, 2012