Incentivizing Efficient Load Repartition in Heterogeneous Wireless Networks with Selfish Delay-Sensitive Users

International audienceAlmost all modern mobile devices are equipped with a number of various wireless interfaces simultaneously, so that each user is free to select between several types of wireless networks. This opportunity raises a number of challenges, since in general selfish choices do not lead to a globally efficient repartition of users over networks. The most popular approach in this context is to charge an extra tax for connecting to overloaded networks, thus incentivizing users to choose less congested alternatives. In this paper we apply that idea to a system where several networks with a common coverage area coexist. Moreover we assume that users -or the applications they use- are heterogeneous in their sensitivity to the congestion-varying Quality-of-Service (QoS). We show the technical and computational feasibility of computing taxes leading to a globally optimal outcome for any number of networks and application types (QoS-sensitivities), hence generalizing the results from previous works