Universal Beam Prediction without Beam-Codebook Variation Information at User-Equipment
Journal
Proceedings - IEEE Global Communications Conference, GLOBECOM
Start Page
5880
End Page
5885
ISSN
2334-0983
ISBN (of the container)
979-833157781-0
ISBN
[9798331577810]
Date Issued
2025-12-08
Author(s)
Abstract
Spatial domain (SD) beam prediction to derive the received powers of all beams from a subset of beam measurements can significantly reduce the reference signal overhead and power consumption, and is currently under 3GPP standardization for 6G. In this paper, we identify the beam codebook variation problem, which is crucial for the field deployment of commercial beam prediction algorithm. To be more specific, note that the codebook for 5G mmWave beamforming has not been standardized, and the user-equipment (UE) beam prediction must accommodate different beam codebooks from different base stations (BSs), which are typically unknown to UE and lead to uncertain SD correlation across beams. To solve this problem, we develop a new downlink beam prediction approach universally applicable to all the BSs' beam codebooks without the knowledge of the codebooks at the UE side by leveraging the post-processing of the prediction report in the BS. We first develop a mathematical model aligned with the ongoing 3GPP standardization and then derive the low-complexity universal beam prediction scheme. Our simulation based on a 3GPP channel model shows that the proposed universal approach achieves a prediction error probability 100 times lower than that of the conventional and machine learning model based algorithms, and matches the prediction performance achieved with the full beam codebook knowledge on the UE side.
Event(s)
2025 IEEE Global Communications Conference, GLOBECOM 2025
Subjects
beam management
Beam prediction
codebook variations
reference signal overhead reduction
Publisher
Institute of Electrical and Electronics Engineers(IEEE)
Type
conference paper