Random Forest Resource Allocation for 5G Systems: Performance and Robustness Study

Next generation cellular networks will have to leverage a further cell densification to accomplish the ambitious goals with respect to aggregate multi-user sum rates. It is well known that this requires much more coordination among the transmission points in the system to balance interference with terminal performance. Traditionally, this has been limited by the coordination capabilities in the backbone, nevertheless with the cloud radio access network (CRAN) architecture these limitations are likely to be overcome. This shifts the attention back to applicable resource allocation, which need to be applicable for very short radio frames, large and dense sets of radio heads, and large user populations in the coordination area. So far, mainly channel state information (CSI)-based resource allocation schemes have been proposed for this task. However, they come at a considerable complexity while also incurring a significant price in terms of CSI acquisition overhead on the system. In this paper, we study an alternative approach which promises lower complexity while also having a lower overhead. In particular, we propose to base the resource allocation in multi-antenna CRAN systems on the position information of user terminals only. Based on the user positions, we further propose the application of Random Forests as supervised machine learning approach to determine the multi-user resource al∗Corresponding author Email address: [email protected] (Sahar Imtiaz) Preprint submitted to Elsevier Journal of Computer Communications March 31, 2017 ar X iv :1 70 3. 10 54 9v 1 [ cs .I T ] 3 0 M ar 2 01 7 locations. This likely leads to lower overhead costs, as the acquisition of position information requires less radio resources in comparison to the acquisition of instantaneous CSI. In addition, once a corresponding data structure is learned, the complexity for determining a resource allocation for a given user is low as well. After presenting our design, we extensively benchmark it with the following findings: (I) In general, learning-based RA schemes can achieve comparable spectral efficiency to CSI-based scheme; (II) If taking the system overhead into account, learning-based RA scheme utilizing position information outperform legacy CSI-based scheme by up to 100% ; (III) Despite their dependency on the training data, Random Forests based RA scheme is robust against position inaccuracies and changes in the propagation scenario; (IV) The most important factor influencing the performance of learning-based RA scheme is the antenna orientation, for which we present three approaches that restore most of the original performance when facing random antenna orientations of the user terminal. To the best of our knowledge, these insights are new and indicate a novel as well as promising approach to master the complexity in future cellular networks.