Prediction of link blockages and execution of proactive handovers (HO) are essential to establish reliable connectivity between the user equipment (UE) and a base station (BS). In fifth generation (5G) and sixth generation (6G) wireless networks that employ millimeter wave (mmWave) and sub-Terahertz (sub-THz) frequencies. Communication links are easily blocked by obstacles in the outdoor dynamic environment. It is crucial to monitor the link status and identify the future blockages or link status to initiate proactive measures. Multi-base station connectivity to the UE, beam switching, execution of HO between base stations (BS) based on the link status obtained with aid of non-RF sensors, light detection and ranging (LiDAR), and visual information. The reported methodologies of using multi-modal data, multi- BS connectivity, beam selection, deliver sub-optimal utilization of network resources and also impart additional computational overheads at the BS. In this paper, we present a deep-learning based solution for predicting the future line of sight (LOS) link blockage of a mobile UE in dynamic urban environment by using the existing mmWave wireless information. A deep learning based algorithm is trained with blockage instances and used to predict the future time instant of link blockage due to dynamic obstacles, and thereby enabling proactive HO to a nearby BS, to provide reliable communication link in 5G/6G mobile networks. The simulation results shows accuracy close to 90% of blockage prediction using wireless signature in comparison with existing deep learning models.
 

https://ieeexplore.ieee.org/document/10008702