Hierarchical MAB Framework for Energy-Aware Beam Training for Near-Field Communications.

Area of Science:

• Wireless Communications
• Signal Processing
• Telecommunications Engineering

Background:

• XL-MIMO systems require efficient beam training for multi-user communication.
• Balancing communication quality with device energy efficiency is crucial for system longevity.

Purpose of the Study:

• To propose a novel near-field beam training scheme for XL-MIMO multi-user frequency division duplex systems.
• To enhance system energy efficiency and extend device lifespan by integrating energy-aware user scheduling and hierarchical beam training.

Main Methods:

• A two-phase combinatorial multi-armed bandit (MAB) framework is utilized.
• An energy-aware upper confidence bound (UCB) algorithm is designed for user scheduling, prioritizing users with high rates and sufficient battery.
• Hierarchical beam training employs two UCB algorithms: one for initial angle acquisition using DFT codebooks, and another for optimal polar-domain codeword selection.

Main Results:

• Simulations confirm the effectiveness of the proposed scheme.
• Notable achievable rate gains are demonstrated for multi-user communications.
• The scheme successfully balances communication quality and device battery levels.

Conclusions:

• The proposed MAB-based beam training scheme significantly improves energy efficiency in XL-MIMO systems.
• The integration of energy-aware scheduling and hierarchical beam training enhances system performance and device lifespan.
• This approach offers a viable solution for optimizing future wireless communication systems.