Joint load-driven frequency allocation and user association in dense cellular networks

Fuelled by the proliferation of smartphones, wireless traffic has experienced huge growth, which will continue and exacerbate the capacity strain in cellular networks. Network densification has emerged as a powerful paradigm to boost spectral efficiency and accommodate the continual rise in demand for wireless capacity. These dense networks also make resource allocation more challenging though, as they result in more irregular cells with possibly overlapping coverage areas and greater variability in traffic loads. To deal with these load imbalances, there are typically two methods to dynamically match capacity and demand: "bring users to capacity" (user association) and "bring capacity to the users" (frequency/spectrum allocation). In this paper we study the joint operation of load-aware dynamic user association and frequency allocation algorithms in dense cellular networks. Motivated by a joint load-balancing optimization problem, we consider load-aware algorithms that operate using load measurements at the access points (APs) and can react to changing load conditions without knowledge of difficult-to-obtain system parameters like arrival patterns of users. We present extensive simulation results for various parameter settings, allowing for the user association and frequency allocation algorithms to operate on different time scales. The simulation results show that the joint operation of these algorithms leads to excellent performance without the need to provide an initialized and optimized frequency allocation