Feedback Modeling in Internet Congestion Control
Sergey Gorinsky · 2003
Abstract This paper examines the impact of feedback modeling on analytical conclusions about Internet congestion control. In particular, we focus on the problem of convergence of binary adjustment algorithms to fairness. First, we review results for the traditional model that assumes uniform feedback to all users. Then, we demonstrate that under a more realistic feedback model, a larger family of binary adjustment algorithms converges to fairness. We also discuss practical implications of our analytical ndings for enhancing Internet congestion control.