I am currently studying the Additive Increase Multiplicative Decrease method, used in TCP as congestion avoidance technique. If we have K TCP sessions sharing a common link of bandwidth R, it is said that this technique guarantees fairness for all the sessions, i.e, each session will have a throughput of R/K.
Now, I'd like to prove this fairness mathematically (reaching the conclusion that, regardless of the initial values of the throughput of each session, they will all eventually tend to R/K).
Thanks !