Traffic offloading/onloading in multi-RAT cellular networks

We analyze next generation cellular networks, offering connectivity to mobile users through multiple radio access technologies (RATs), namely LTE and WiFi. We develop a framework based on the Markovian agent formalism, which can model several aspects of the system, including user traffic dynamics and radio resource allocation. In particular, through a mean-field solution, we show the ability of our framework to capture the system behavior in flash-crowd scenarios, i.e., when a burst of traffic requests takes place in some parts of the network service area. We consider a distributed strategy for the user RAT selection, which aims at ensuring high user throughput, and investigate its performance under different resource allocation schemes.