Dual-camera view-based blockage prediction for indoor environment and comparison with the LIDAR based solution

Abstract

This thesis presents a novel approach for predicting blockages in indoor environments within the context of 5G and beyond communication systems. The proposed solution utilizes a dual-camera view-based method to accurately forecast potential blockages that may affect wireless communication.

The methodology involves detecting the bounding box (BB) coordinates of the user equipment (UE) object from two camera views, followed by locating the object in three-dimensional (3D) space and predicting its next location within the same spatial domain. To assess the threat of blockage, the 3D location is projected onto a single camera view, enabling the calculation of the UE object’s blockage probability.

Simulation results from this study demonstrate the efficacy of the dual-camera-based localization approach, achieving a mean localization error of 0.39m with a deviation of 0.36m. Moreover, the proposed solution exhibits remarkable predictive capabilities, successfully anticipating impending blockages up to 20 frames in advance, which corresponds to approximately one second of video at a frame rate of 21 frames per second (fps).

The findings of this research highlight the potential of the dual-camera-based blockage prediction method for enhancing communication system performance in indoor environments. The ability to accurately estimate blockages in advance enables proactive measures to be taken, such as adjusting transmission parameters or implementing handover strategies, to mitigate the impact on communication quality. Further investigations and developments in this area can lead to substantial improvements in the reliability and efficiency of future wireless networks.