Why is the threshold set to one half of the current congestion window size when the timer expires at the sender?

link|improve this question
1  
Do you have an issue that pertains to this attribute of the TCP specification? As is, this question feels a lot like homework.. – Shane Madden Jan 4 at 22:58
No but I am researching for my dissertation at the moment.. and got a little confused about congestion control algorithms... – StefanGrech Jan 4 at 23:06
@StefanGrech For the future, this is why we have reference books (any school with a decent CS program should have TCP/IP Illustrated in the library - definitely grab a copy while you're working on your thesis), and if you're working on a dissertation you may want to have a copy handy for when you get out into the job market and need to layeth the holy smack down upon someone. – voretaq7 Jan 4 at 23:08
You may want to refer to the RFCs. This may be a result of the example alogrithm. – BillThor Jan 5 at 0:12
feedback

1 Answer

up vote 2 down vote accepted

1) It's not (always).
That is one step in one of many possible congestion control algorithms.

2) If it is, it's because TCP falls back and attempts another "slow start" of the connection, looking for a point where there is no packet loss.


Absent any other information may I suggest a copy of TCP/IP Illustrated and perhaps a read through the TCP Congestion Control RFCs(follow references too!) or the short version on Wikipedia which describe some (common) congestion control algorithms?

link|improve this answer
1  
Thanks, I will try to get hold of this book as soon as possible – StefanGrech Jan 4 at 23:06
feedback

Your Answer

 
or
required, but never shown

Not the answer you're looking for? Browse other questions tagged or ask your own question.