An Optimized Handover Management Scheme Tailored for Heavy Hitters in a Disaggregated 5G O-RAN Architecture

In this paper we propose a tailored handover scheme for Heavy Hitters (HHs) in Next Generation Self-Organizing Network (NG SON). The conventional handover scheme takes into consideration only the signal strength of source and target cell on handover decision, without considering the traffic specifics of the User Equipments (UEs). We believe that customizing the handover decision on individual UEs' needs gives higher Quality of Experience (QoE). Particularly, in this work we focus on optimizing the QoE of HH UEs in the 5th generation (5G) network by controlling and customizing the handover mechanism specifically for these users. We test our approach in Network Simulator 3 (ns-3) and use the emerging Open RAN (O-RAN) framework as a supporting architecture to proactively monitor the presence of HHs in the network and react upon their appearance. The contribution in this paper is twofold: We present an offline detection scheme for HHs based on Deep-Neural Network (DNN) with an accuracy of 94% and a Greedy Handover algorithm that improves the overall average throughput of HHs.