Streaming services has been becoming more and more popular in recent years. The reason can be traced back to the explosive growth of Internet. Along with the prolification of streaming services, many issues showed up, such as the poor quality of service, long transmission delay and frequent disconnection. Many solutions have already been proposed to resolve these issues.
They can be roughly classified into two categories, say, Mesh-pull and Tree-push. Mesh-pull systems divide transmitting video into clips. User nodes send requests to neighbors and download video clips from them if positive responses are received. Overhead and extra delays are incurred
MLDST
SP Tree
MST
ZIGZAG
0 50 100 150 200 250 300 350 400 450
Dela y tim e (m s)
Delay Diameter
42
from large amount of control messages. As a result, Mesh-pull systems seem not appropriate for those services that demand short transmission delay.
Hereby we choose to take Tree-push based approach to model the problem. It avoids the heavy traffic of control messages and thus reduces transmission delay. We model the problem into a MLDST that minimize the worst case packet loss rate while bound delay and node degree. This problem is then proved to be NP-Complete. We propose Heuristic MLDST by modifying the single-source shortest-path algorithm, Dijkstra’s algorithm, by bounding the Delay Diameter and node degree. Proposed solution depicts a spanning tree with minimum virtual Loss Diameter under delay and node degree constraints. Through several experiment evaluations on PlanetLab, we showed that our solution outperforms other tree construction algorithms in video quality.
The issue of peer churn and membership changes is ignored in our research and will be studied in the future.
Reference
[1] Marc Bui, Franck Butelle, and Christian Lavault, "A Distributed Algorithm for Constructing a Minimum Diameter Spanning Tree", Journal of Parallel and Distributed Computing, Vol. 64, No. 5, May, 2004, pp. 571-577.
[2] Miguel Castro, Peter Druschel, Anne-Marie Kermarrec, Animesh Nandi, Antony Rowstron, Atul Singh, "SplitStream: High-Bandwidth Multicast in Cooperative Environments", Proceedings of the 19th ACM Symposium on Operating Systems Principles, 2003, pp. 298-313.
[3] M. R. GAREY and D. S. JOHNSON, W. H. Freeman, "Computers and intractability: A guide to the theory of NP-completeness", W. H. Freeman and Company, 1979.
[4] Yu-Hsuang Guo, John K. Zao, Wen-Hsiao Peng, Lin-Shung Huang, Fang-Po Kuo, Che-Min Lin,
"Trickle: Resilient Real-Time Video Multicasting for Dynamic Peers with Limited or Asymmetric Network Connectivity", Proceedings of 8th IEEE International Symposium on Multimedia,
43
December, 2006.
[5] Jan-Ming Ho, D.T. Lee, Chia-Hsiang Chang, and C.K. Wong, "Bounded-Diameter Minimum Diameter Spanning Trees and Related Problems", Proceedings of the 5th Annual Symposium on Computational Geometry, 1989, pp. 276-282.
[6] Jan-Ming Ho, D. T. Lee, Chia-Hsiang Chang and C. K. Wong, "Minimum Diameter Spanning Trees and Related Problems", SIAM Journal of Computing, Vol. 20, No. 5, 1991, pp. 987-997.
[7] Xiao-Jun Hei, Yong Liu, Keith W. Ross, "IPTV over P2P streaming networks: The mesh-pull approach", IEEE Communications Magazine, Vol. 46, No. 2, February 2008, pp. 86-92.
[8] Hung-Chang Hsiao and Chih-Peng He, "A Tree-Based Peer-to-Peer Network with Quality Guarantees", IEEE Transactions on Parallel and Distributed Systems, Vol. 19, No. 8, August 2008, pp. 1099-1110.
[9] X. Jia, "A Distributed Algorithm of Delay Bounded Multicast Routing for Multimedia Applications in Wide Area Networks", Proceedings of 6th International Conference on Computer Communications and Networks, September, 1997, pp. 208-213.
[10] K. Kalapriya, R. Venkatesh Babu, S. K. Nandy, "Streaming Stored Playback Video Over A Peer-to-Peer Network", Proceedings of IEEE International Conference on Communications, Paris, Vol. 3, June 2004, pp. 1298-1302.
[11] Bo Li, Hao Yin, "Peer-to-Peer Live Video Streaming on the Internet: Issues, Existing Approaches, and Challenges", IEEE Communications Magazine, Vol. 45, No. 6, June, 2007, pp.
94-99.
[12] Jiang-Chuan Liu, Sanjay G. Rao, Bo Li and Hui Zhang, "Opportunities and Challenges of Peer-to-Peer Internet Video Broadcast", Proceedings of the IEEE, Vol. 96, No. 1, January, 2008, pp.
11-24.
[13] Vinay Pai, Kapil Kumar, Tamilmani, Vinay Sambamurthy, Alexander E. Mohr, "Chainsaw:
Eliminating Trees from Overlay Multicast", Proceedings of 4th International Workshop on Peer-to-Peer Systems, February, 2005.
44
[14] Mohit Singh, Lap Chi Lau, "Approximating Minimum Bounded Degree Spanning Trees to within One of Optimal", Proceedings of the 39th Annual ACM Symposium on Theory of Computing, San Diego, California, USA, June, 2007, pp.661-670.
[15] Duc A. Tran, Kien A. Hua, Tai Do, "ZIGZAG: An efficient peer-to-peer scheme for media streaming", Proceedings of 22nd IEEE INFOCOM Conference, Vol. 2, 2003, pp. 1283-1292.
[16] DEB Veeravalli, JB Weissman, "A Robust Spanning Tree Topology for Data Collection and Dissemination in Distributed Environments", IEEE Transactions on Parallel and Distributed Systems, Vol. 18, No. 5, May, 2007, pp. 608-620.
[17] Vidhyashankar Venkataraman, Kaouru Yoshida, Paul Francis, "Chunkyspread: Heterogeneous Unstructured Tree-Based Peer-to-Peer Multicast", Proceedings of the 14th IEEE International Conference on Network Protocols, 2006, pp. 2-11.
[18] Aggelos Vlavianos, Marios Iliofotous and Michalis Faloutsos, "BiTos: Enhancing BitTorrent for Supporting Streaming Applications", Proceedings of 25th IEEE International Conference on Computer Communications, 2006, pp. 1-6.
45
附錄: 衍生的相關研究
A、 適用於 MANET 之選擇性保證封包送達之通訊協定設計(Partial Reliable TCP for MANET) 許多網路上的資訊服務,如影音資訊, 在網路上傳輸時並未要求封包都送達,不同的封包有 不同的重要性, 例如 MPEG 的關鍵畫面(I -Frame)就比其他畫面重要。 常用的傳輸層通訊 協定中,UDP 與 TCP 都對所有的封包一視同仁,前者不做任何保證,而後者雖可保證所有封 包的送達,但效率較差。本研究提出一個新的 TCP,”Partial-Reliable TCP”,使用選擇 性重傳機制,配合應用程式的需求,對指定的封包提供遞送保護。當封包遺失時,只重傳保 護的封包,可減少額外的網路資源消耗,並提升服務的品質。此外,我們提出 Single-Side 的版本,接收端可以使用一般的 TCP,在封包傳送時,讓接收端以為封包都是無誤傳達的,
在 Client-Server 架構的網路服務中,只有伺服器端必頇使用我們的 Partial-Reliable TCP,
大幅提高本通訊協定的可行性。最後我們利用網路模擬工具 NS-2 來模擬實際網路環境,將我 們的方法與現行的通訊協定在可解畫面封包數、PSNR 值及額外耗用的網路資源三個參數做比 較。我們使用兩個 Video 影像作為傳輸標的,在高遺失率的有線與無線網路的環境中進行實
驗。當傳輸時間限制很短時,(例如影像會議的應用),無論在有線或無線的通訊環境中,都
可以有效的提升影音傳輸的品質。本研究之成果發表在 ICS2008 上[8]。
B、適用於無線隨意式網路及感測 網路之逐節點 TCP 傳輸協定
行動隨意式網路(MANET)是一種具有高度動態拓撲結構的網路。 每一個行動隨意式網路由一 組移動節點(Node)組成,彼此之間互相支援轉送封包可以不依靠基地台建構成 Intranet。此 種網路中,因節點移動之緣故,連線不穩定、頻寬較窄,錯誤率亦較高。傳統的 TCP 傳輸協 定在行動隨意式網路上的效能不免遭受重創。傳統的 TCP 在封包遺失時,只能從傳送端進行 重傳,而行動隨意式網路傳輸品質極不穩定,常常重送多次才可到達目的地,導致要耗費極 長的時間才能將封包送達目的地,然而,在行動隨意式網路中,大量傳輸資料的需求並不大,
反而是封包的快速送達更為重要,因此加速封包的送達成為比增大傳送量更為重要的目標。
46
為了使封包較快送達目的地端,我們提出了 Hop-by-Hop TCP 的方法,使每個節點使用當地 重傳以保證封包成功的傳到 下一個節點,遺失的封包不必重新由傳送端重傳,能更快反應封 包遺失,並且提昇傳輸可靠度,使封包在高遺失率的情形之下能順利且較為快速的送達目的 地端。 而在感測網路中,這種情況更為明顯,我們將這個協定延伸到感測網路中。我們利用 NS-2 網路模擬器進行實驗,驗證我們的機制,實驗在不同的拓樸及負載等參數下進行,觀察 傳輸成功率及封包傳輸時間,以及公平性。實驗結果指出,本方法無論在 MANET 或感測網路 中均有不錯的表現。此項研究成果已經發表在兩個學術會議,以及一個期刊中[2,4,9]。
C. 有助於提高服務品質的前瞻式封包排程機制
受到封包網路原本忽略時效性特性之影響,對時效性要求極高的多媒體網路服務,如 Voice over IP (VoIP)以及 Video on Demand (VoD)在 All-IP 整合的核心網路上提供時,其服務品 質低於傳統之電路交換網路。 封包在網路的傳遞過程中受到各種因素之影響,於到達目的地 時,可能會造成 long delay time,high jitter 或 packet loss,而在目的地端幾乎已經沒 有補救機會,故而如果能在傳遞的過程中,依封包的時效性及重要性做適度的次序調動 (rescheduling) 而不要依序傳遞(FIFO),讓過遲的封包提前送出,而將有時間餘裕的封包稍 緩送出,如此截長補短,可提高網路效能及整體 QoS 滿意度。我們在 BBQ (Budget-Base QoS) 的架構下發展一套簡單而有效的方法,在單佇列 (Single Preemptive Queue)及多佇列 (Multiple FIFO Queue) 的 router 架構下,根據封包時效性及重要性賦予合適的 profit function, 並參考封包在後續路程上各 router 的負載狀態以便能更精確預估 封包是否能及 時到達目的地並調整 profit function 參數以調整封包的送出順序,如此能提高排程的效能。
我們先對單一服務等級的封包排程進行研究,獲得參數調整之技巧,再根據其結果設計多服 務等級的封包排程方法,其重點在於如何調配 profit function 給不同的服務等級。我們藉 由 NS-2 模擬模擬器進行實驗,評估本方法的效能, 實驗結果顯示我們的方法可以較每個 router 僅根據自身所知的資訊進行排程更可以有效提高網路效能,且能對不同的服務等級做 差異化處理以提高整體 QoS 滿意度。最後在多等級服務的實驗環境及評估指標下,網路高負 載的情況,本方法與 Simulated Priority Queue 排程演算法比較可以提升 34%的整體滿意度。
47
本研究之成果發表在 NBiS08 上[8]。
D、在 P2Pnet 下選擇性保證串流資料傳播
網路技術發展的日新月異帶領了眾多新網路服務的崛起,例如即時影音串流這類的多媒體服 務。但即時影音串流服務所產生的龐大資料流和傳輸延遲時間的嚴格限制也為網路環境帶來 許多挑戰,在這些條件下,傳統的 Clien-Server 拓樸架構將 client 要求的影音資料以單一 鏈結傳輸時,常會因為頻寬不足而面臨嚴重的封包遺失,或是資料流擁擠造成的額外傳輸延 遲使得封包無法達到即時性的需求。P2P 網路擁有 Clien-Server 架構所難以達到的規模伸縮 性,且對於節點、鏈結失效所引起的傳輸錯誤也較能容忍,更重要的是,它有效的分散了原 本負載在少數 link 上的龐大資料流。因此 P2P 架構近年來風行於即時影音串流服務。
目前 P2P (Peer-to-Peer)網路的拓樸多是隨意的,當網路成員規模龐大時,由傳送端出 發到遠方的接收端,途中可能經過無數的鏈結,每一個鏈結都會由於頻寬的不足使得資料流 遭受某程度的品質損害,且在即時影音服務中若資料流的累積延遲時間超出可容忍範圍時,
無法為使用者接受。
本研究嘗試找出一個較好的拓樸用以傳輸多媒體資料流,使得位於最遠端節點的延遲亦 能為使用者接受,且資料品質的損害程度最小。本問題是 NP-Complete 問題,我們修改 Dijkstra 演算法加上 degree 限制,作為本問題的次佳解演算法,我們以 PlanetLab 在實際 的網路上進行實驗,證實我們的演算法比傳統的 Minimum-Spanning Tree 有較高的效能。
E、基於群組模式下之大型網路語音會談系統
目前利用 VoIP 進行網路會談(VoIP Conference)已有越來越流行的趨勢,尤其在遠距離 語音會議的應用方面,不但能夠節省費用,且還能夠同時允許多人上線會談。但隨著上線人 數的增加,網路的流量負載與連線數量也相對的倍增,此時會議的品質(QoS)往往大打折扣。
尤其在說話聲音延遲時間(Delay Time)方面,更是大型網路語音會談的最大致命傷。使用者 在進行語音通話時,最多能容忍 350 至 400ms 的延遲時間,然而多人網路會議在傳統的傳輸
48
架構上,往往因人數稍微增加就使得延遲時間超過此限制。
針對此問題,過去有一些研究提出解決方案:例如選擇網路會議中處理能力較強的使用
針對此問題,過去有一些研究提出解決方案:例如選擇網路會議中處理能力較強的使用