Talk about the new CCA: Elastic-TCP

[Tao Tse]

Hi,

I'm not sure it is suitable to discuss another CCA here.

The paper: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8642512

Cause the elastic-tcp is simple enough from my comprehension, so I spent some time  to implement one: https://github.com/vertextao/tcp-elastic , and did a simple test:

But, as the paper described, the cwnd increasing function curve should be square root like, the test result looks not. What I missed?

Note: I observe the RTT value to reflex the cwnd value (because of the bufferbloat, the more inflght, the bigger RTT, I confirmed this on Reno and Cubic).

Thanks for any discussion!