A distributed globally convergent algorithm for fair, queue-length-based congestion control

We present an easy-to-implement explicit rate congestion control algorithm for connection-oriented data networks. The algorithm is adaptive to changing network conditions, robust to network delays, and is decentralized. Further, it achieves max-min fairness and queue length stability under finite buffer length and minimum rate constraints. The proof of global convergence of the algorithm involves stability analysis of discrete-time linear systems with saturation.