1 vote

i have a self doubt that till now i solve the TCP question where the sender has $1\hspace{0.1cm}mss$ at starting so during time out generally reduced the senders window size to $1 \hspace{0.1cm}mss$ but now there is a question that supposes that at starting the sender has $2 \hspace{0.1cm}mss$ to send and so at time out what will be the sender size i.e we will put it down to $1 \hspace{0.1cm}mss$ or to $2\hspace{0.1cm}mss$ .........

0 votes

If initially sender has window size equal to cwnd = 2MSS then sender will continue with it until congestion detected by -

1) time out occur, then start with slow start phase.

2) 3 ACKs then start with congestion avoidance phase.

In case 1 Sender's threshold will be reduced to half of the current window size $SSthreshold = \frac{1}{2}cwnd$ and again congestion window size will be set to 1 ie $cwnd = 1$.

In case 2 Sender's threshold will be reduced to half of the current window size $SSthreshold = \frac{1}{2}cwnd$ and congestion window size will be set to sender's threshold ie $cwnd = SSthreshold$. It then start congestion avoidance phase.

For more reference refer

https://gateoverflow.in/2156/gate2012-45

PS :

http://inst.eecs.berkeley.edu/~ee122/fa06/notes/18-Congestion-2up.pdf