Improving Downlink Coordinated Multipoint Performance in Heterogeneous Networks.
We propose a novel method for practical Joint Processing downlink coordinated multipoint (DL CoMP) implementation in LTE/LTE-A systems using supervised machine learning. DL CoMP has not been thoroughly studied in previous work although cluster formation and interference mitigation have been studied extensively. In this paper, we attempt to improve the cell edge data rate served by a heterogeneous network cluster by means of dynamically changing the DL SINR threshold at which the DL CoMP feature is triggered. We do so by using a support vector machine (SVM) classifier. The simulation results show a cell edge user throughput improvement of 33.3% for pico cells and more than four-fold improvement in user throughput in the cluster. This has resulted from a reduction in the downlink block error rate (DL BLER) and an improvement in the spectral efficiency due to the informed triggering of the multiple radio streams as part of DL CoMP.
Publisher URL: http://arxiv.org/abs/1608.08306