Traffic profile based clustering for dynamic TDD in dense mobile networks

The traffic profile of future mobile networks is foreseen as becoming more variable both in time and among base stations (BSs) due to the widespread heterogeneity in applications and services. Dynamic time division duplex (TDD) has been recognized as an important enabler to cope with this traffic variability, especially in dense networks with many BSs and a small number of user equipments (UEs) served by each BS. In this context, one of the main issues is to develop an incisive scheme to manage the BS-to-BS and UE-to-UE interference that arises. In this work, we propose a novel long-term BS clustering scheme that groups BSs that have a similar traffic profile and would be characterized, without clustering, by strong BS-to-BS interference. Due to the complexity of the optimal solution, we propose a heuristic algorithm which solves the optimization problem efficiently. Numerical evaluations in a dense homogeneous pico BS network show that the proposed scheme strongly outperforms even the baseline dynamic TDD without clustering by ensuring a reduction in packet delay of up to 55% and a gain in the UE packet throughput of about 75%.