Allocation-pricing game for multipath routing in virtual networks

Abstract

In this paper we consider the problem of routing data packets in virtual network. It is assumed that each virtual link may consist of arbitrary collection of paths composed of physical links. Traffic originating from a source node may reach its destination using many paths simultaneously. In order to maximize the utility in a network where a large number of such transmissions occur, it is necessary to allocate bandwidth of physical links appropriately among all component flows, as well as to decide on the division of capacity of virtual links. We apply gametheoretic analysis to the formulated multipath routing problem in order to develop a distributed mechanism for solving network utility maximization problem in virtual network. We show that the obtained allocations form a strategy profile which is a pure Nash equilibrium. Moreover, the proposed mechanism allows to allocate the bandwidth in a way which maximizes the total utility.