目前網路環境中, 有許多不同種類的應用程式利用UDP傳輸協定來傳送資料, 如多媒體的網路應用程式。其大多需要平順傳輸速率來維持使用者的舒適度, 因此, 傳輸速率變化程度太大的TCP傳輸協定便不適合, 所以才發展出TCP友善性控制協定, 目的為希望在長時間的競爭環境下能與TCP資料流獲得相同的頻寬佔有率且得到較為平順的傳輸速率。

GAIMD為TCP友善性控制協定之一, 做法是在壅塞避免的狀態中, 若一個往返時間內沒有任何封包遺失, 傳送端的窗口大小便會增加一個的大小, 但如果接收到三個重複回應封包, 窗口大小就會變為原來的倍。利用公式的運算, 求得適當的與來達成TCP友善性。

然而模擬的結果顯示GAIMD並沒有達到良好的TCP友善性。故在此篇論文中提出靜態加權GAIMD, 靜態內插GAIMD, 動態加權GAIMD以及動態內插GAIMD四種方法來計算所需要的值, 以來達成TCP友善性, 所設定的皆為0.875。從模擬實驗結果顯示在經過這四種演算法皆能較原先的GAIMD壅塞控制法計算出更正確的值。靜態加權GAIMD與靜態內插GAIMD皆比原始的GAIMD有較好的TCP友善性, 但其缺點為無法對於網路上突發的壅塞情況作反應, 且不易得到精確的封包遺失率。因此, 我們針對此缺點提出動態加權GAIMD與動態內插GAIMD, 根據每個往返時間內所計算封包遺失率來, 得到符合網路環境的最佳值, 以求得到最佳的TCP友善性。實驗結果顯示動態GAIMD的TCP友善性比靜態GAIMD更佳。

In the current network, many kinds of applications, using the UDP protocol, such as streaming multimedia, need the smooth sending rate to maintain the user-perceived quality. However, the abrupt change of rate in using TCP congestion control is a significant impediment to the deployment of TCP by these emerging applications such as streaming multimedia. Therefore, the TCP-friendly protocol has been developed to suit to these applications. A TCP-friendly flow has the similar sending rate with a TCP flow under the same condition.

The GAIMD mechanism wants to achieve TCP-friendliness by changing the mechanism, increased by one and decreased to half, used in the TCP Reno for window adjustment. That is, in the congestion avoidance state, window size is increased by  per window of packets acknowledged and it is decreased to  of current window size when there is a packet loss. However, previous papers indicated that GAIMD flows (= 0.31, = 0.875) can not achieve the TCP-friendliness.

In the paper, we extended the original GAIMD to propose four new mechanisms, such as static weighted GAIMD, static interpolated GAIMD, dynamic weighted GAIMD, and dynamic interpolated GAIMD, to calculate the accurate  to be exact TCP- friendliness. We conducted some experiments to validate the new mechanisms to be more TCP-friendly. The static weighted GAIMD and static interpolated GAIMD are both more TCP-friendly than original GAIMD. However, they have two fatal drawbacks they can not immediately response to congestion and can not get the accurate loss rate. Therefore, we proposed dynamic weighted GAIMD and dynamic interpolated GAIMD to conquer these problems. According to the loss rate calculated in each rtt, the sender calculates the more appropriate  to be more TCP-friendly. The simulations results show that the dynamic GAIMD is more TCP-friendly than the static GAIMD.

[1] V. Jacobson. Congestion avoidance and control. In ACM SIGCOMM’ 88, 1988.
[2] K. Thompson, G. Miller, and M. Wilder. Wide-area Internet traffic patterns and characteristics. IEEE Network Magazine, 11(6): Nov. 1997.
[3] D. Clark and J. Wroclawski. An approach to service allocation in the Internet, internet draft, 1997.
[4] D. Tan and A . Zakhor. Real-time Internet video using error resilient scalable compression and TCP-friendly transport protocol. IEEE/ACM Transactions on Multimedia, May 1999.
[5] D.-M. Chiu and R. Jain. Analysis of the increase and decrease algorithm for congestion avoidance in computer networks. Computer Networks and ISDN Systems, 17(1):1-14, 1989.
[6] K. Thompson, G. Miller, and M. Wilder. Wide-area Internet traffic patterns and characteristics. IEEE Network Magazine, 11(6): Nov. 1997.
[7] D. Clark, S. Shenker, and L. Zhang. Supporting realtime applications in an integrated services packets network: Architecture and mechanism. In ACM SIGCOMM’90, July 1992.
[8] S. Floyd and K. Fall. Promoting the use of end-to-end congestion control in the Internet. IEEE/ACM Transations Review, 26(3):5-21, July 1996.
[9] V. Jacobs and A. Eleftheriadis. Providing video services over network without quality of service quarantees. In World Wide Web Consortium Workshop on Real-time Multimedia and the Web, 1996.
[10] J. Mahdavi and S. Floyd. TCP-friendly unicast rate-based flow control. Technical note sent to the end2end-interest mailing list, Jan. 1997.
[11] T. Turletti, S. F. Parisis, and J. Bolot. Experiments with a layered transmission scheme over the Internet. In IEEE INFOCOM‘ 98, San Francisco, California, U.S.A., Mar. 1998.
[12] D. Sisalem and H. Schulzrinne. The loss-delay based adjustment algorithm: A TCP-friendly adaption scheme. In NOSSDAV ’98, 1998.
[13] S. Cen, C. Pu, and J. Walpole. Flow and congestion control for Internet streaming applications. In Multimedia Computing and Networking, Jan. 1998.
[14] V. Ozdemir and I. Rhee. TCP emulation at the receivers(TEAR), presentation at the rm meeting, Nov. 1999.
[15] R. Rejaie, M. Handly, and D. Estrin. An end-to-end rate-based congestion control mechanism for realtime streams in the Internet. In IEEE INFOCOM ’99, Mar. 1999.
[16] J. Padhye, J. Kurose, D. Towsley, and R. Koodli. A model based TCP-friendly rate control protocol. In NOSSDAV’ 99, 1999.
[17] L. Vicisano, L. Rizzo, and J. Crowcroft. TCP-like congestion control for layered multicast data transfer. In IEEE INFOCOM’ 99, New York, New York, U.S.A., Mar. 1999.
[18] S. Floyd, M. Handly, J. Padhye, and J. Widmer. Equation-based congestion control for unicast applications. In ACM SIGCOMM 2000, Aug. 2000.
[19] M. Mathis, J. Semke, J. Madhavi, and T. Ott. The macroscopic behavior of the TCP congestion avoidance algorithm. ACM Computer Communication Review, 27(3): 67-82, July 1997.
[20] J. Padhye, V. Firoiu, D. Towsley, and J. Kurise. Modeling TCP throughput: A simple model and its empirical validation. In of ACM SIGCOMM’ 98’ Vancouver, Canada, 1998.
[21] Y. R. Yang and S. S. Lam. General AIMD Congestion Control. Technical Report TR-200009, Department of Computer Science, University of Texas at Austin, May 2000.