Towards flow scheduling optimization in multihomed mobile hosts

The problem of network selection in a heterogeneous radio access environment has been extensively studied from the perspective of a single-homed mobile terminal. This work approaches the problem by assuming a multihomed mobile host reachable through multiple global IP addresses that are bound to its various radio interfaces. In the context of a multihomed mobile host, network selection sets further requirements than decision on the best available radio access technology and network provider. In the presence of device, application, user and provider constraints, the decision process involves: 1) selection of a subset of hostpsilas radio interfaces for activation, 2) selection of the best access network to be associated with each active radio interface, 3) assignment of application flows to the most appropriate radio interface in a way that maximizes user utility. For brevity reasons, this work uses the term flow scheduling for referring to the above decision problem. We formulate flow scheduling as an optimization problem and more specifically as a binary integer programming problem. The problem involves minimization of an objective function that combines user utility, economical cost due to network usage and power consumption cost. Problem solutions are presented for various scenarios, drawn from a multi-access network setting, that illustrate how optimized flow scheduling contributes to better utilization of available resources and terminal adaptation to device and network context.