Congestion-aware network selection and data offloading

Wi-Fi offloading is a cost-effective approach to provide an immediate capacity relief to the congested areas in a cellular network. However, previously proposed schemes mainly focus on alleviating the cellular congestion by offloading the data traffic to Wi-Fi as much as possible, but without systematic considerations of the network congestion, switching penalty, and pricing in the networks. In this paper, we consider the network selection and data offloading problem in an integrated cellular Wi-Fi system by incorporating the practical considerations of (i) user mobility, (ii) location, user, and time dependent WiFi availabilities, (iii) network dependent switching time and switching cost for changing network connections, and (iv) usage-based pricing into our modeling. We formulate the interactions of the users' congestion-aware network selection decisions across multiple time slots as a non-cooperative network selection game (NSG), where the strategy of each user corresponds to a route on a graph. We prove that the NSG is equivalent to a congestion game, which implies that the game has the finite improvement property. As a result, when the players repeatedly perform better response updates, the system is guaranteed to converge to a pure Nash equilibrium. Simulation results show that our proposed NSG scheme achieves a better load balancing than two static heuristic schemes.