User-association and resource allocation in future TDD access and backhaul networks

The success of LTE heterogeneous networks heavily depends on the efficient interplay between user association and uplink/downlink resource allocation. Much work has been devoted to investigate this interaction and to capture the involved trade-offs. Nevertheless, different schemes are based on different assumptions and optimization objectives (e.g., improve throughput, delay, load balancing, etc.) So, it remains rather unclear how these two problems should interact. Additionally, most of the literature in this context ignore the backhaul network which is expected to be a key performance bottleneck in future networks and thus it threatens their performance. In this work, we develop a framework that efficiently tackles the interplay of (i) user-association, (ii) access radio, and (iii) backhaul resource allocation for a family of different objective functions. We propose an algorithm that reduces the complexity of this multi-variable problem by decomposing it in three optimization levels and guarantees convergence to a global optimal point. Simulation results validate the performance benefits of the proposed joint optimization approach compared to traditional techniques. Index Terms user association, backhaul, queueing theory, uplink, downlink, hetnets, resource allocation, dynamic TDD.