I came across this question:
At the beginning of transmission t, a TCP connection in congestion avoidance mode has a congestion window w = 60 segments.Packet loss is observed during transmission rounds t, t+10, and t+20 by getting multiple ACKs. What is the congestion window at the end of round t, t + 10, and t + 20? If there's no further packet loss, when will the window of w = 60 segments be reached again?
Congestion window is halved during transmission round t, leading to w = 30. At the beginning of transmission round t + 10, the window has increased to w = 40 but it will be halved again during transmission round t + 10 to w = 20. Similarly, after transmission round t + 20, the window will be w = 15. With no further packet loss, 45 transmission rounds later at t + 65, the window will reach again the original size of w = 60.
I know that due to multiple ACKs, packet loss reduces the window size to half. So at t+10, w = 30 makes sense. However, I don't really get the rest. Why does the window size increase by 10? And at t+20, why doesn't it reduce to 10 instead of 15? If someone could explain the steps, that would be great.