根據Cisco發佈的「視覺網路指標(Visual Networking Index，VNI)」指出，現今全世界總體行動數據流量到了2018年將會成長超過10倍，使得行動網路將面臨嚴重的阻塞問題，如何有效解決無線接取網路的負荷，是現今相當重要的議題。多數文獻著眼於佔據總流量中最大比例的影音為主，主要針對在無線接取網路中，解決基地台頻寬有限而無法服務過多的使用者的影音需求的問題。
在2012年有學者提出使用無線內容分發網路(wireless CDN)的架構來解決此問題。在單一細胞底下設置數個快取節點(稱為Helpers)，並採取Greedy快取策略，將影片依照熱門程度(被使用者請求的次數)、使用者位置等因素快取到Helpers，使得使用者請求可以儘可能由Helpers來服務，減少基地台負荷。在Greedy策略之下，在影片總數遠多過Helper可以快取的容量時，此策略僅能快取熱門程度高的為主，對於熱門程度低但仍然相當大量的影片並沒有很好的使用者請求命中率(可由Helper服務這筆影片的機率) 。
本論文以Greedy策略中的wireless CDN架構為基礎提出我們的新快取策略，以實際上使用者影片請求的軌跡，分析使用者行為，將熱門程度高與熱門程度低的影片分為兩部分策略，熱門程度高的快取到所有Helpers，熱門程度低的在Helpers仍有快取空間時也平均分配到剩餘的快取空間，增加熱門程度低但請求數也相當大的影片的命中率，並且也引用CDN既有快取策略的Popular(快取最熱門的檔案)、Fuzzy Decision (隨機快取檔案數次並選擇最好的結果)來做比較。
    在模擬評估的結果顯示，當Helpers在快取容量為影片總數的50%以下時，Greedy的策略因為在Helper快取容量有限時，對於大量的非熱門影片請求無法取得有效的命中率，Popular則是只快取了熱門影片，因此命中率更低，Fuzzy Decision的策略則是沒考量影片熱門程度，對於熱門影片的命中率損失大。而我們的新快取策略是以實際有參考性的使用者請求行為來進行快取策略，所以能有更好的請求命中率。並且我們額外也使用了LTE-sim這套近年相當多人使用及討論的LTE模擬環境，模擬出各個策略的結果中，每個使用者下載一筆影片的平均延遲時間，並且與現行沒有使用wireless CDN架構的環境做比較，結果也顯示了wireless CDN確實能有效提升整體效能。

According the Visual Networking Index(VNI) released by Cisco, the total mobile traffic across the world will increase more than ten times until 2018. Mobile network will face to serious network congestion problems. Thus, how to resolve wireless access network loading problem will become rather important issue in the near future. Most references focus on video traffic that occupying the most of total mobile traffic to solve the limited bandwidth problem of base stations.
At 2012, there were some academics proposed that using wireless content delivery network(CDN) architecture to solve the problem. They create several cache nodes called Helpers in single cell, and using Greedy strategy that caching videos to Helpers by video popularities or user locations, and let user requests can served by Helpers as much as possible. By Greedy strategy, when total number of videos far more than cache storage of Helpers, Greedy strategy can only cache the videos with high popularity and get worse user request hit rate, the chance of served by Helpers, with huge number but low popularity videos.
This paper proposes our new cache strategy that also using wireless CDN architecture. By analyzing the real user video request traces, we propose two strategies for high popularity and low popularity videos. Caching high popularity videos to all Helpers, and caching low popularity videos to remaining storage when total storage of Helpers is not full that increasing the request hit rate of huge number of low popularity videos. We also quote existing CDN cache strategy Popular, caching the most popularity videos, and Fuzzy Decision, caching files randomly several times and selecting the best result, to compare.
The simulation result shows that when cache storage of Helpers is less than 50% of total number of videos, Greedy strategy can only get few hit rate about huge number but low popularity videos with limited cache storage of Helpers, Popular can only cache high popularity videos that resulting in far lower request hit rate, and Fuzzy Decision strategy doesn’t consider the video popularity that resulting in huge miss of high popularity video requests. Our new strategy quote real and referential user request traces so we can get better request hit rate. We also use LTE-sim, that be referenced a lot for recent years, to simulate the result of every quoted strategy, and get the average video download delay times for each user about every strategy and current architecture, without wireless CDN architecture, to compare. The results also show that wireless CDN architecture increase the performance indeed.

目錄
第一章	緒論	1
1.1前言	1
1.2章節大綱	5
第二章	相關研究背景	6
2.1背景介紹	6
2.2內容分發網路	9
2.3 相關的CDN檔案快取策略	12
2.3.1 Greedy	12
2.3.2 Popular	20
2.3.3 Fuzzy Decision	22
2.3.4 各策略之差異	34
2.3.5 研究目標	35
第三章	我們的新快取策略	36
3.1 新快取策略大綱	36
3.2使用者行為分析	38
3.3詳細流程	40
3.4 動態配置	47
3.5 各策略之差異	49
第四章	模擬結果	50
4.1基地台與通道模型	52
4.2 HELPER模型	53
4.3使用者模型	54
4.4模擬參數	56
4.5比較策略	58
4.6比較參數	59
4.7模擬結果	60
第五章	結論	75
未來工作	77
第六章	參考文獻	80



 
圖目錄
圖1.1 無線接取網路(RAN)示意圖	2
圖2.1 CISCO預估全球行動數據流量成長趨勢圖	6
圖2.2 CISCO預估行動數據類型所佔比例趨勢圖	7
圖2.3 SINGLE CELL場景範例	13
圖2.4 NEGIN GOLREZAEI ET AL.’S GREEDY ALGORITHM	17
圖2.5 GREEDY快取流程圖	18
圖2.6 GREEDY快取演算法	19
圖2.7 POPULAR快取流程圖	21
圖2.8 POPULAR快取演算法	22
圖2.9 隨機配置快取流程圖	23
圖2.10 隨機配置快取演算法	24
圖2.11 FUZZY DECISION計算流程	25
圖2.12 FUZZY DECISION歸屬函數	27
圖3.1 熱門影片快取流程圖	41
圖3.2 熱門影片快取演算法	42
圖3.3 非熱門影片快取流程	44
圖3.4 非熱門影片快取演算法	45
圖4.1 CISCO預估全球行動網路系統分布趨勢圖	51
圖4.1 請求命中次數	60
圖4.2 累計命中次數(WITH HELPER STORAGE 30GB)	63
圖4.3 累計命中次數(WITH HELPER STORAGE 60GB)	64
圖4.4 累計命中次數(WITH HELPER STORAGE 90GB)	65
圖4.5 請求命中率	67
圖4.6 運算複雜度	69
圖4.7 平均命中成本	71
圖4.8 平均延遲時間	73
圖6.1 影片請求路徑	78
圖6.2 影片快取在無線接取網路邊緣	79

 
表目錄
表2.1 歸屬值分類	28
表2.2 模糊規則庫	29
表2.3 模糊決策權重值	32
表2.4 各策略差異	34
表3.1各策略差異	49
表4.1 基地台與通道相關設定	52
表4-2 HELPER相關設定	53
表4-3 YOUTUBE請求紀錄欄位內容	54
表4-4 環境輸入參數	56
表4.5 各策略計算成本類型	69

[1] Cisco Visual Networking Index: Forecast and Methodology, 2013-2018, [Online]. Available: http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/ns705/ns827/white_paper_c11-481360_ns827_Networking_Solutions_White_Paper.html
[2] S. Rudd, “Global Backhaul Investment Gap of $9.2 Billion could increase Customer Churn,” [Online]. Available:
http://www.strategyanalytics.com/default.aspx?mod=pressreleaseviewer&a0=5332
[3] I. Lazar and W. Terrill, “Exploring content delivering networking,” in IT Professional, Vol. 3(4), PP. 47-49, Aug. 2001.
[4] G. Pallis and A. Vakali, “Insight and Perspectives for Content Delivery Networks,” in Communications of the ACM - Personal information management, Vol. 49(1), PP. 101-106, Jan. 2006.
[5] M. Pathan and R. Buyy, “A Taxonomy of CDNs, Content Delivery Networks,” in Germany: Springer-Verlag, 2008.
[6] Y. Chen, L. Qiu, W. Chen, N. Luan and R. H. Katz, “Clustering Web content for efficient replication,” in IEEE International Conference on Network Protocols, PP. 165-174, Nov. 2002.
[7] N. Golrezaei, K. Shanmugam, A. Dimakis, A. Molisch, and G. Caire, ”Femtocaching: Wireless video content delivery through distributed caching helpers,” in INFOCOM, 2012 Proceedings IEEE, PP. 1107-1115, Mar. 2012.
[8] N. Golrezaei, A.G. Dimakis and A.F. Molisch, “Wireless device-to-device communications with distributed caching,” in IEEE International Symposium on Information Theory Proceedings (ISIT), PP. 2781-2785, Jul. 2012.
[9] C. Bouras, N. Kanakis, V. Kokkinos, and A. Papazois, “Application layer forward error correction for multicast streaming over LTE networks,” in International Journal of Communication Systems, Feb. 2012.
[10] C. Xu, F. Zhao, J. Guan, H. Zhang, and G.-M. Muntean, “QoE-driven user-centric VoD services in urban multi-homed P2P-based vehicular networks,” in IEEE Transactions on Vehicular Technology, Vol. 62(5), PP. 2273-2289, Nov. 2012.
[11] M. Iturralde, T. Ali Yahiya, A. Wei, A. Beylot, "Resource allocation using shapley value in LTE networks," in IEEE International Symposium on Personal Indoor and Mobile Radio Communications Conference, PP. 31-35, Sept. 2011.
[12] S. Sivasubramanian, M. Szymaniak, G. Pierre, and M. van Steen, “Replication for web hosting systems,” in ACM Computing Surveys, Vol. 36(3), PP. 291-334, Sept. 2004. 
[13] L. Qiu, V. Padmanabhan, and G. Voelker, “On the placement of web server replicas,” in INFOCOM 2001. Twentieth Annual Joint Conference of the IEEE, PP. 1587-1596, Apr. 2001. 
[14] N. Golrezaei, A.G. Dimakis, A.F. Molisch, and G. Caire, “Femtocaching and Device-to-Device Collaboration: A New Architecture for Wireless Video Distribution,” in IEEE Communications Magazine, vol. 51(4), PP. 142–49, 2013. 
[15] M. Esteve, G. Fortino, C. Mastroianni, C. E. Palau, and W. Russo, “CDN-supported Collaborative Media Streaming Control,” in IEEE Multimedia, PP. 60-71, Apr.-Jun. 2007.
[16] J. B. Chen, “Efficient Content Placement on Multimedia CDN Using Fuzzy Decision Algorithm,” in Applied Mathematics & Information Sciences, Vol. 6(2), PP. 471-477, Apr. 2012.
[17] S. Borst , V. Gupta , A. Walid, “Distributed caching algorithms for content distribution networks,” in Information communications, 29th conference, PP. 1478-1486, Mar. 14-19. 2010.
[18] G. Piro, L. Grieco, G. Boggia, F. Capozzi, P. Camarda, “Simulating LTE Cellular Systems: An Open-Source Framework,” in IEEE Transactions on Vehicular Technology, Vol. 60(2), PP. 498-513, Feb. 2011.
[19] YouTube Traces From the Campus Network, 2008, [Online] Available: http://traces.cs.umass.edu/index.php/Network/Network
[20] M. Zink, K. Suh, Y. Gu, and J. Kurose, "Watch Global Cache Local: YouTube Network Traces at a Campus Network - Measurements and Implications,” in IEEE MMCN, Jan. 2008.
[21] E.Perahia and R.Stacey, “Next generation wireless LANs:throughput, robustness and reliability in 802.11n,“ in Cambridge University Press, 2008.
[22] B. R. Tan, and L. Massouli ́e, “Optimal content placement for peer-to-peer video-on-demand systems,” in IEEE/ACM Transactions on Networking, Vol.21(2), PP. 566-579, Apr. 2013.
[23] S. Borst , V. Gupta , and A. Walid, “Distributed caching algorithms for content distribution networks,” in Information communications, 29th conference, PP. 1478-1486, Mar. 2010.
[24] H. Huang, P. Xia, S.G Chan, G. Shi, and H. Zhang, “Joint optimization of content replication and server selection for video-on-demand,” in IEEE International Conference on Communications, PP. 2065-2069, Jun. 2012.
[25] W. Jiang, S. Ioannidis, L. Massoulíe, and F. Picconi, “Orchestrating massively distributed CDNs,” in ACM CoNEXT, 2012.
[26] J. P. Muñoz-Gea, J. Malgosa-Sanahuja, and P. Manzanares-Lopez, “Optimizing content placement in a peer-assisted VoD architecture,” in Peer-to-Peer Networking and Applications, Vol.6(3), PP. 340-360, Sept. 2013.
[27] H. Ahlehagh, and S. Dey, “Video-Aware Scheduling and Caching in the Radio Access Network,” in IEEE/ACM Transactions on Networking, Jan. 2014.