Cross-Tier Dual-Connectivity Designs of Three-Tier Hetnets With Decoupled Uplink/Downlink and Global Coverage Performance Evaluation

In the conventionally coupled uplink (UL) and downlink (DL) association (CUDA) scheme, the UL performance is limited greatly by the DL parameters such as the density and power of base stations (BSs). To overcome this issue, this paper focuses on a novel three-tier heterogeneous network (HetNet) as well as global performance evaluation, where the cross-tier dual-connectivity (DC) and decoupled UL and DL association (DUDA) are integrated. On one hand, with the cross-tier DC, in DL or UL, typical user equipment is associated simultaneously with the primary and secondary BSs that are located at different tiers. On the other hand, with DUDA the primary and secondary access points of UL and DL are not necessarily the same. The proposed cross-tier DC scheme meets the need of the next generation networks where control channel comes from Macro BS and data channels from small cell BSs. For such HetNets, the work first presents a comprehensive investigation of the cross-tier DC design of all possible DUDA modes and the corresponding association conditions. Second, by simplifying the association conditions the work gives the comprehensive derivations in terms of the association probabilities and the statistical descriptions of the primary and secondary UL access distances. Third, by using the method from stochastic geometry this paper derives the general form of UL coverage probabilities by creating a common model of DUDA designs so that the global UL average coverage probability (ACP) of the whole network is achieved. The presented numerical and simulated analysis first validates the derivations. Then, the analysis focuses on the global UL ACP comparison of the whole HetNet between the DUDA and CUDA designs. The comparison of global ACP shows that the ACP of DUDA design is higher than the one of the CUDA design. Although the powers and densities of BSs impose the outstanding impact on the ACP of the CUDA designs, they have a very limited effect on one of the DUDA designs due to the utilization of decoupling association.