The objective of this work is to determine a close-to-optimal user-to-base station (BS) association that maximizes the number of users served by the downlink of a heterogeneous network (HetNet). Such an association must not only ensure that the number of accommodated users is maximized but also that the network resources are efficiently utilized and the users' quality of service (QoS) demands are met. In its simplest form, the optimization problem that underlies this association is combinatorial NP-hard and the difficulty of solving it is aggravated in HetNets by the disparity of transmit powers and BS computational capabilities. To find close-to-optimal user-to-BS associations, we develop a two-phase method based on semidefinite relaxation (SDR) with randomization, which is a powerful method for solving a class of combinatorial problems. Unlike the majority of other polynomial complexity techniques, SDR has a provable approximation accuracy. Numerical examples show that, in comparison with other user-to-BS association approaches, the one proposed herein enables more efficient utilization of resources and a significantly higher number of users to be accommodated.

Additional Metadata
Keywords Convex optimization, Gaussian randomization, Heterogeneous cellular networks
Persistent URL dx.doi.org/10.1109/VTCFall.2015.7391104
Conference 82nd IEEE Vehicular Technology Conference, VTC Fall 2015
Citation
Sokun, H.U. (Hamza Umit), Gohary, R, & Yanikömeroǧlu, H. (2016). QoS-guaranteed user association in HetNets via semidefinite relaxation. In 2015 IEEE 82nd Vehicular Technology Conference, VTC Fall 2015 - Proceedings. doi:10.1109/VTCFall.2015.7391104