Data-driven handover optimization in small cell networks

Since the advent of 1G through 5G networks, telecommunication industry has gone through phenomenal transformation in the way we communicate, we work, and we socialize. In dense or ultra-dense mobile communication networks, the users are very frequently handed over to other cells making seamless mobility a challenging and complex problem. Therefore, robust connectivity in such networks becomes a very critical issue. In this paper, we present a data-driven handover optimization approach aiming to mitigate the mobility problems including handover delay, early handover, wrong selection of target cell and frequent handover. The proposal is based on collecting the information from the network and developing a model to determine the relationship between the features drawn from the collected dataset and key performance indicator (KPI) expressed as the weighted average of mobility problem ratios. Handover design parameters- time to trigger and handover margin are optimized to improve KPI. The KPI estimation drawn on time to trigger and hysteresis margin design parameters is estimated through neural network multilayer perception method. It is established through simulation results that the proposed approach yields significantly improved handover performance mitigating mobility problem in ultra-dense cellular networks to notable extent.