[e2e] (Why) does rate-based AIMD lead to a stable network?

[Panos GEVROS]

Michael Welzl typed :
 
 |paper. It also does not seem to explain why AIMD is still stable in the
 |context of heterogeneous RTT's.

the heterogeneous RTT case can be modelled by a similar system with 
heterogeneous additive increase factors (and the same multiplicative decrease 
factor - although stability holds even for heterogenous MD factors)

more recent work does not use the Chiu&Jain model with the discontinous 
feedback and rate/load adaptation functions, but the cumulative effect that 
ecn marks have on the rate,

 |Which leaves me asking: "Given heterogeneous RTT's, is there ANY proof that
 |a scheme which does not adhere to the 'conservation of packets' principle
 |(e.g. any rate based scheme) and uses AIMD will have the network converge
 |to an equilibrium around the optimal point in terms of fairness and
 |efficiency?"

i study the requirements on aimd(a,b) in the heterogeneous case of the 
Chiu&Jain model as part of my (yet unpublished) thesis - let me know if you 
care for a pointer.

cheers
Panos