Heterogeneous cellular networks under diverse coupling and association criteria

Resumen

Current networks are moving towards Heterogeneous Cellular Networks (HCN) arising from the combination of small cells with existing macrocells. The aim of this thesis is to analyze various performance indicators of heterogeneous cellular networks under diverse coupling and association criteria. We considered a two-tier heterogeneous cellular network with macro and pico BSs and UEs uniformly distributed. Realistic path loss models given by 3GPP have been taken into account for both macro and pico tiers. In this work, three association criteria were used to associate users to macro or a pico tier which include the coupled and decoupled association criteria. The coupled association criteria encompass nearest BS and maximum downlink average power whereas in decoupled association criteria, users were associated in DL by maximum average receive power and in UL by minimum path loss. Cell Range Expansion (CRE) and Fractional Power Control (FPC) techniques have been considered. The results showed a remarkable lack of independence and correlation between uplink and downlink coverage has been guessed even under independent Rayleigh fading. Simulation results showed that taking into account limits on the maximum spectral efficiency and on the number of simultaneous active users within a cell strongly modify the results on joint binary rate. It has been investigated that deploying a denser infrastructure by increasing the total number of BSs, not only improves coverage and average rates but also the energy efficiency and fairness. Limiting the number of simultaneous active users does not affect the coverage and average symmetric binary rate but energy efficiency improves for its lower values, since the resources of the cell are distributed among less users, at the expense of an unfairer treatment as more users are kept inactive. It has been inspected that including more picocells has no worth for average rate and fairness under average criteria mainly due to the high difference of power between both tiers but energy efficiency slightly improves as more users get inactive. Cell range expansion bias reduces coverage but fairness is maximum for its mid-range values and the energy efficiency as well as the binary rate has improved much beyond those mid-range values. Finally, slight increment in fractional power control improves coverage and joint rate and provides better user fairness treatment. Our results show that for realistic path loss models, the decoupled DL/UL association does not improve the results sufficiently to compensate for the implementation difficulties it represents.