資料流壅塞控制協定(Datagram Congestion Control Protocol, DCCP)具備擁塞控制機制及不可靠傳輸的特性，非常適合具及時性的多媒體應用。然而對某些應用而言，某些關鍵性封包的遺失恐會造成服務品質(Quality of Services, QoS)大幅降低。本論文提出部份可靠性延伸之DCCP (DCCP partial reliability extension, PR-DCCP)，其可傳送需要重傳的遺失封包。由於DCCP所使用的序號會持繼增加，重傳的封包無法使用原始的序號傳送，因此PR-DCCP使用封包序號補償來解決此問題，其概念為重傳之封包會附加一差距值，如此接收端可利用序號及此差距值來得到原始需重傳封包的序號。模擬結果以兩個效能指標評估：可解視框比例(Decodable Frame Ratio, DFR)與無效資料接收比例(Useless Data Received Ratio, UDRR)，前者表示QoS，而後者表示頻寬的浪費程度，來評估不同的傳輸層協定，包括PR-DCCP、DCCP、串流控制傳輸協定(Stream Control Transmission Protocol, SCTP)、傳輸控制協定(Transmission Control Protocol, TCP)與使用者資料流協定(User Datagram Protocol, UDP)的效能。結果顯示PR-DCCP幾乎在所有的模擬中都比其他協定得到較佳的DFR與UDRR。針對各種不同的影片，PR-DCCP的DFR比DCCP增加1.2%~12.4%，而UDRR則可降低約73.2%~85.1%。此外，也以模擬方式研究兩個決定封包是否需要可靠性傳送的策略，最後則是以模擬結果比較PR-DCCP與SCTP之部分可靠性延伸(SCTP partial reliability extension, PR-SCTP)間之差異。
雖然部分可靠性協定可藉由重傳重要的封包以提高接收端收到資料的完整性，然而多媒體應用大多數具時效性，傳送端決定那些封包應該重傳以及重傳的時機，對於提昇多媒體應用的服務品質有很大的影響。因此本論文進一步提出一個調適性重傳方法(Adaptive Retransmission Method, ARM)。此方法可以讓傳送端在判斷封包是否應重傳時不需參考到應用層的資訊，而是根據接收端所估計之來回時間(Round-Trip Time, RTT)與應用層讀取封包的速率，來計算出時效上重傳有效之封包序號臨界值來判斷，將能夠在接收端應用層讀取前及時送達的封包允以重傳。模擬結果顯示，當PR-DCCP採用ARM傳送多媒體應用時，相較於將封包生命時間(lifetime, LT)分別設定為固定值0.2秒(LT=0.2s)與2秒(LT=2s)之重傳的方法，在頻寬較低或傳播延遲較高的環境下，ARM能使多媒體應用獲得最佳的服務品質。模擬結果顯示我們提出的ARM在頻寬較低的情況下可分別提升LT=0.2s與LT=2s 之DFR 17.4%與25.6%，UDRR則降低10.2%與25.5%。而當傳播延遲較高的情況下可分別提升LT=0.2s與LT=2s 之DFR 8.6%與11.3%，UDRR可分別降低12.3%與7.8%。

Datagram congestion control protocol (DCCP), possessing congestion control and unreliable transmission, specially suits real-time multimedia applications. Nevertheless, losses of key packets will cause a substantial decline on quality of services (QoS) in some applications. This dissertation proposes a DCCP partial reliability extension (PR-DCCP) that can retransmit lost packets as needed. Since DCCP uses an incremental sequence number, the retransmitted packets cannot utilize their original sequence number. To solve this problem, PR-DCCP adopts sequence number compensation, which appends an offset to the retransmitted packet; thus the receiver can use the sequence number of this retransmitted packet and the attached offset so as to re-obtain the original sequence number. The simulation uses two performance metrics: decodable frame ratio (DFR) representing QoS, and useless data received ratio (UDRR) representing the bandwidth waste. These are used to evaluate different transport protocols, namely, PR-DCCP, DCCP, SCTP, TCP, and UDP. Simulation results show that PR-DCCP has the better DFR and UDRR than other transport protocols in almost all cases. For various movies, a DFR of PR-DCCP is 1.2–12.4% higher than that of DCCP; while UDRR is lower by 73.2–85.1%. Furthermore, two reliability policies to determine which packets require reliability are investigated. Finally, the comparisons between PR-DCCP and PR-SCTP are examined.
Partially reliable transport protocol is able to improve the amount of the valid data received by the receiver through retransmitting key packets. Nevertheless, since most multimedia applications are time-sensitive, it has significant impact on the QoS of multimedia applications that the sender determines which packets need to be retransmitted and when these packets are retransmitted. Therefore, this dissertation further proposes Adaptive Retransmission Method (ARM). This method allows the sender to retransmit packets without referring the information in the application layer. It is according to the packet fetching rate and the Round-Trip Time (RTT) to obtain a threshold of sequence number. The sender uses this threshold to determine which packets can arrive in time and only retransmits these packets. From simulation results, in comparison with the other retransmission methods, which the lifetime (LT) of packets is set as a fixed 0.2s (LT=0.2s) or a fixed 2s (LT=2s), PR-DCCP adopting ARM can achieve the best QoS under the environments with the low bandwidth or the high propagation delay. When the environment with the low bandwidth, the DFRs of LT=0.2s and LT=2s are improved by 17.4% and 25.6%, while UDRRs are reduced by 10.2% and 25.5%, respectively. Under the condition with the long propagation delay, the DFRs of LT=0.2s and LT=2s are improved by 11.3% and 8.6%, while UDRRs are reduced by 12.3% and 7.8%, respectively.

[1].E. Kohler, M. Handley, S. Floyd, and J. Padhye, “Datagram Congestion Control Protocol (DCCP),” IETF RFC 4340, March 2006.
[2].E. Kohler, M. Handley, and S. Floyd, “Designing DCCP: Congestion Control Without Reliability,” SIGCOMM’06, September 2006.
[3].H. Wang, Y. Jin, W. Wang, J. Ma, and D. Zhang, “The Performance Comparison of PRSCTP, TCP and UDP for MPEG-4 Multimedia Traffic in Mobile Network,” ICCT 2003, April 2003.
[4].R. Stewart et al., “Stream Control Transmission Protocol (SCTP),” IETF RFC 2960, October 2000.
[5].R. Stewart, M. Ramalho, Q. Xie, M. Tuexen, and P. Conrad, “Stream Control Transmission Protocol (SCTP) Partial Reliability Extension (PR-SCTP),” IETF RFC 3758, May 2004.
[6].L. Wang, K. Kawanishi, and Y. Onozato, “Simulation-Based Optimization on MPEG-4 over SCTP Multi-streaming with Differentiated Retransmission Policy in Lossy Link,” APSCC 2007, pp. 164-171, December 2007.
[7].M. Molteni and M Villari, “Using SCTP with Partial Reliability for MPEG-4 Multimedia Streaming,” BSDC on Europe 2002, October 2002.
[8].X. Lei and C.M. Leung, “Applying PR-SCTP to Transport SIP Traffic,” IEEE Globecom 2005, November 2005.
[9].H. Hagino, Y. Miyazaki, Y. Onoe, Y. Atsumi, H. Komaki, M. Taniguchi, and N. Yamanouchi, “A Playout Time Oriented Retransmission Scheme for Multimedia Streaming Systems,” HSNMC 2003, July 2003.
[10].J. Ott, S. Wenger, N. Sato, C. Burmeister, and J. Rey: Extended RTP Profile for RTCP-based Feedback (RTP/AVPF), draft-ietf-avt-rtcp-feedback-05.txt, IETF (2003)
[11].A. Huszak, and S. Imre, “Source Controlled and Delay Sensitive Selective Retransmission Scheme for Multimedia Streaming,” ISCC 2007, July 2007.
[12].A. B. Johnston, “SIP Understanding the Session Initiation Protocol”, Third Edition, ARMech House, 2001.
[13].G. Camarillo, R. Kantola, and H. Schulzrinne, “Evaluation of Transport Protocols for the Session Initiation Protocol,” IEEE Communications Society, pp. 40-46, vol. 17, issue 5, September-October 2003.
[14].G. Marco, D. Vito, and M. Longo, “SCTP as a transport for SIP: a case study,” in Proceedings of SCI 2003, pp. 284-289, vol. 11, 2003.
[15].The Network Simulator - ns-2. http://www.isi.edu/nsnam/ns
[16].http://www.tkn.tu-berlin.de/research/trace/ltvt.html
[17].Z. Lifen, S. Yanlei, and L. Ju, “Performance Study of Transmitting MPEG4 over SCTP,” IEEE International Conference on Neural Networks & Signal Processing, December 2003.
[18].A. Ziviani, B. E. Wolfinger, J. F. Rezende, O. C. M. B. DuARMe, and S. Fdida, “Joint Adoption of Qos Schemes for MPEG Streams,” Multimedia Tools and Applications, vol. 26, no. 1, pp. 59-80, May 2005.