SemQNet: Semantic-Aware Quantised Network for mmWave Beam Prediction

TY - GEN

T1 - SemQNet: Semantic-Aware Quantised Network for mmWave Beam Prediction

AU - Khan, Ahsan Raza

AU - Yadav, Poonam

N1 - This is an author-produced version of the published paper. Uploaded in accordance with the University’s Research Publications and Open Access policy.

PY - 2025/3/27

Y1 - 2025/3/27

N2 - Millimetre-wave (mmWave) communication systems use large antenna arrays and narrow beams to achieve strong signal power. However, this approach requires extensive beam training, which leads to high overhead. Recently proposed vision-aided beam prediction methods show promising results, reducing this overhead. However, these techniques have considerable computational complexity, hindering practical deployment. To address this issue, we propose a Semantic-Aware Quantised Network (SemQNet) framework that leverages image compression and a lightweight computer vision model to extract semantic information used for training a fully connected neural network (FCNN). Additionally, the proposed SemQNet also uses quantisation-aware training (QAT), which enables low-precision arithmetic operation, reducing the model size in the training process. Our tests on the DeepSense 6G dataset show that SemQNet achieves almost the same top-1 accuracy as existing vision-based methods while reducing the model size by 74.21\%. This smaller model size reduces the communication overhead, making SemQNet a practical and efficient solution for energy-constrained mmWave communication systems.

AB - Millimetre-wave (mmWave) communication systems use large antenna arrays and narrow beams to achieve strong signal power. However, this approach requires extensive beam training, which leads to high overhead. Recently proposed vision-aided beam prediction methods show promising results, reducing this overhead. However, these techniques have considerable computational complexity, hindering practical deployment. To address this issue, we propose a Semantic-Aware Quantised Network (SemQNet) framework that leverages image compression and a lightweight computer vision model to extract semantic information used for training a fully connected neural network (FCNN). Additionally, the proposed SemQNet also uses quantisation-aware training (QAT), which enables low-precision arithmetic operation, reducing the model size in the training process. Our tests on the DeepSense 6G dataset show that SemQNet achieves almost the same top-1 accuracy as existing vision-based methods while reducing the model size by 74.21\%. This smaller model size reduces the communication overhead, making SemQNet a practical and efficient solution for energy-constrained mmWave communication systems.

U2 - 10.1109/WCNC61545.2025.10978576

DO - 10.1109/WCNC61545.2025.10978576

M3 - Conference contribution

SN - 979-8-3503-6837-6

T3 - IEEE Wireless Communications and Networking Conference

BT - Workshop on Sustainable and Intelligent Green Internet of Things for 6G and Beyond (6GSIoT)

PB - IEEE Communications Society

T2 - 2025 IEEE Wireless Communications and Networking Conference

Y2 - 24 March 2025 through 27 March 2025

ER -