Optimal flow control and capacity allocation in multi-service networks

The optimal flow control of noncooperative users in multiservice networks is investigated in a game-theoretic way. We adopt a model that differs from previously considered ones in that a user's strategy (viz., flow or average rate) is constrained between two numbers: those are specified by the user's quality of service requirements or by the user's technology. Each user tries to maximize its own performance measure, defined by a utility function in a more or less standard fashion. We prove existence and uniqueness of the Nash equilibrium. There is a simple algorithm that determines this Nash equilibrium point (NEP). We also comment on the convergence properties of distributed algorithms towards this NEP. Finally, we give a suggestion on how noncooperative users may distribute their flows among several links, by imposing an appropriate pricing scheme that encourages load balancing.