Towards Measuring the Deployment Information of Different TCP Congestion Control Algorithms: The Multiplicative Decrease Parameter

The Internet has recently been evolving from homogeneous congestion control to heterogeneous congestion control. Five years ago, Internet traffic was mainly controlled by the standard TCP AIMD algorithm, whereas Internet traffic is now controlled by many different TCP congestion control algorithms, such as AIMD, CUBIC, CTCP, and FAST. However, there is very little work on the performance and stability study of the Internet with heterogeneous congestion control. One fundamental reason is the lack of the deployment information of different TCP congestion control algorithms. In this paper, we present our initial work towards measuring such important information in the Internet. Specifically, considering that web traffic comprises a significant portion of the total Internet traffic, we propose a method that can actively infer the TCP multiplicative decrease parameter of a remote web server. The multiplicative decrease parameter, together with other TCP characteristics (such as the window growth function to be measured in our future work), can be used to uniquely identify the TCP congestion control algorithm of a web server. Our extensive lab test-bed experiments show that our method can infer the TCP multiplicative decrease parameter with reasonably good accuracy.