1A Linear Optimization Approach for Achieving Flow Fairness in Unicast and Multicast Networks

Abstract—In this paper, we present an analytical solution to the general problem of flow control for both unicast and multicast IP networks. Relying on the so-called max-min flow fairness metric, we formulate a pair of centralized and decentralized convex opti-mization problems that can be analytically solved with quadratic and linear complexities respectively. Utilizing the solution to the decentralized optimization problem, we then propose a flow con-trol algorithm requiring no per flow state information. The pro-posed algorithm can be implemented by merely making use of a simple Explicit Congestion Notification (ECN) marking scheme to convey minimum per link flow fairness information to the end nodes. Relying on the results of our current work, we also address the flow control issue of distributing real-time multimedia traffic across multicast IP networks