Joint spectrum partition and user association in multi-tier heterogeneous networks

The joint spectrum partition and user association problem for multi-tier heterogeneous networks is studied in this paper, where disjoint spectrums are allocated among tiers and users are associated with each tier with a biased received power. The random placement of base-stations (BSs) of different tiers are modeled using stochastic geometry, which accounts for their practical deployment and also makes analysis tractable. We derive an upper bound of the average user proportional fair utility based on the user coverage rate, from which we formulate a network utility maximization problem. The optimization of the proposed utility bound shows that the optimal spectrum allocation for each BS tier matches the average proportion of users associated with that tier. The solution to the optimization problem also provides closed-form expressions for the optimal user associated bias factors. Compared to system-level optimization solutions based on specific network topology and channel realization, our offline analytical approach offers deployment insights. Simulation results demonstrates the effectiveness of the proposed approach.