K-best branch and bound technique for the MINLP resource allocation in multi-user OFDM systems

In this paper, we propose a simplified approach for the Mixed Integer Non-linear Programming (MINLP) problem in the context of dynamic resource allocation for multiuser downlink Orthogonal Frequency Division Multiplexing (OFDM) systems. Our solution is based on the appropriate selection of the K-best branches of the branch and bound technique. The proposed algorithm significantly decreases the computational complexity of the conventional branch and bound technique, yet with a slight reduction in the average user throughput. Simulations results show that a trade-off between complexity decrease and performance reduction could be maintained with the proposed algorithm.