Hierarchical optimization of user association and flexible TDD allocation for access and backhaul networks

The success of future heterogeneous networks (HetNets) heavily depends on the interplay between user association and resource allocation on both the access and backhaul network. While user association is key to improve both the user and network performance, it is becoming a multi-objective optimization problem that should consider the number and type of base station in range. Furthermore, the increasing spatiotemporal heterogeneity in downlink(DL) and uplink(UL) traffic suggests that DL/UL resources can be tuned to optimally serve the respective workload. Split DL/UL association and flexible TDD offer such an opportunity. While much literature exists on these problems, the majority consider them separately. In this work, we develop a framework that tackles the optimal interplay of (i) user-association, (ii) radio resource allocation, and (iii) backhaul resource allocation of TDD resources, for a family of objective functions. We propose an algorithm that reduces the complexity of this problem by decomposing it into three optimization subproblems, each potentially solved by a different network element and at different timescales. We prove convergence to the

global optimum, and provide simulation results that demonstrate the performance benefits of our approach.