Patterned Beam Training: A Novel Low-Complexity and Low-Overhead Scheme for ELAA

TL;DR

This paper introduces a low-overhead, low-complexity patterned beam training scheme for extremely large antenna arrays, significantly reducing training time while maintaining high beam alignment accuracy.

Contribution

The paper proposes a novel patterned beam training scheme that simplifies beam selection to a single linear operation, reducing overhead and complexity compared to traditional methods.

Findings

Outperforms traditional exhaustive search in accuracy

Reduces training overhead by at least 50%

Balances SNR conditions with training efficiency

Abstract

Extremely large antenna arrays (ELAAs) can provide higher spectral efficiency. However, the use of narrower beams for data transmission significantly increases the overhead associated with beam training. In this letter, we propose a novel patterned beam training (PBT) scheme characterized by its low overhead and complexity. This scheme requires only a single linear operation by both the base station and the user equipment to determine the optimal beam, reducing the training overhead to half or even less compared to traditional exhaustive search methods. Furthermore, We discuss the pattern design principles in detail and provide specific forms. Simulation results demonstrate that the proposed scheme outperforms the compared methods in terms of beam alignment accuracy and achieves a balance between signal-to-noise ratio (SNR) conditions and training overhead, making it a promising alternative.