Ensemble Techniques for Scheduling in Heterogeneous Wireless CommunicationscNetworks
Operators can increase the capacity of their wireless networks by installing Small Cells in high traffic regions. However, User Equipments (UEs) at the edges of Small Cells suffer severe interference from neighbouring high-powered Macro Cells. It is desirable to tradeoff the rates of cell-centre UEs, so that cell-edge UEs do not experience dropped calls or unacceptably slow rates. Fairness can be achieved by intelligently scheduling Small Cell attached UEs. Grammar-based Genetic Programming is employed to automatically evolve models which map measurement reports to schedules on a millisecond timescale. The evolved models operate as ensembles. The proposed system significantly outperforms a state of the art benchmark algorithm and is within 7.5% of the estimated optimum.