Selection of Component Carriers Using Centralized Baseband Pooling for LTE-Advanced Heterogeneous Networks

Hiroyuki SEKI  Takaharu KOBAYASHI  Dai KIMURA  

IEICE TRANSACTIONS on Communications   Vol.E96-B   No.6   pp.1288-1296
Publication Date: 2013/06/01
Online ISSN: 1745-1345
DOI: 10.1587/transcom.E96.B.1288
Print ISSN: 0916-8516
Type of Manuscript: Special Section PAPER (Special Section on Heterogeneous Networks for Future Cellular Systems)
LTE-advanced,  carrier aggregation,  component carrier,  heterogeneous networks,  baseband pooling,  

Full Text: PDF(2.2MB)>>
Buy this Article

Bandwidth expansion in Long Term Evolution (LTE)-Advanced is supported via carrier aggregation (CA), which aggregates multiple component carriers (CCs) to accomplish very high data rate communications. Heterogeneous networks (HetNets), which set pico-base stations in macrocells are also a key feature of LTE-Advanced to achieve substantial gains in coverage and capacity compared to macro-only cells. When CA is applied in HetNets, transmission on all CCs may not always be the best solution due to the extremely high levels of inter-cell interference experienced by HetNets. Activated CCs that are used for transmission should be selected depending on inter-cell interference conditions and the traffic offered in the cells. This paper presents a scheme to select CCs through centralized control assuming a centralized baseband unit (C-BBU) configuration. A C-BBU involves pooling tens or hundreds of baseband resources where one baseband resource can be connected to any CC installed in remote radio heads (RRHs) via optical fibers. Fewer baseband resources can be prepared in a C-BBU than those of CCs in RRHs to reduce the cost of equipment. Our proposed scheme selects the activated CCs by considering the user equipment (UE) assigned to CCs under the criterion of maximizing the proportional fairness (PF) utility function. Convex optimization using the Karush-Kuhn-Tucker (KKT) conditions is applied to solve the resource allocation ratio that enables user throughput to be estimated. We present results from system level simulations of the downlink to demonstrate that the proposed algorithm to select CCs can outperform the conventional one that selects activated CCs based on the received signal strength. We also demonstrate that our proposed algorithm to select CCs can provide a good balance in traffic load between CCs and achieve better user throughput with fewer baseband resources.