Meng-Hsun Wu

Meng-Hsun Wu

Auto-generated profile, based on publicly available information

Meng-Hsun Wu

Department of Computer Science and Information Engineering, National Taiwan University, Taipei, Taiwan

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 August 2023 in IEEE Communications Letters
Verified
Journal Article
Closed access
HCHong-Yunn Chen
Hong-Yunn Chen
MWMeng-Hsun Wu
Meng-Hsun Wu
TYTa-Wei Yang
Ta-Wei Yang
JLJia-Wei Liao
Jia-Wei Liao
CHChih-Wei Huang
Chih-Wei Huang
CCCheng-Fu Chou
Cheng-Fu Chou
Upcoming 6G wireless networks promise faster speeds, lower latency, and increased capacity. A key innovation is the intelligent reflecting surface (IRS), which enhances coverage, capacity, and energy efficiency. However, the complex training and computational costs associated with the IRS’s passive components pose challenges for channel prediction. We address this by applying the denoising method on raw data as well as multihead attention for discovery of hidden patterns in complex data, and then using sparse encoding in latent space to retain important information for capturing cross-domain features in the space, time, and frequency domain. Numerical results demonstrate significant performance improvements in channel prediction for IRS-assisted millimeter-wave MIMO OFDM systems.

5 February 2025 in IEEE Transactions on Cognitive Communications and Networking
Verified
Journal Article
Closed access
MWMeng-Hsun Wu
Meng-Hsun Wu
HCHong-Yunn Chen
Hong-Yunn Chen
TYTa-Wei Yang
Ta-Wei Yang
CHChih-Chuan Hsu
Chih-Chuan Hsu
CHChih-Wei Huang
Chih-Wei Huang
CCCheng-Fu Chou
Cheng-Fu Chou
Sixth-generation technology, which surpasses 5G with faster speeds, lower latency, and increased capacity, highlights the pivotal intelligent reflecting surface (IRS) and optimizes wireless communication by precisely handling incoming radio waves, resulting in enhanced network coverage, capacity, and energy efficiency. In real-world scenarios, however, hybrid precoding in the IRS is affected by noise and interference. To address this, we propose DVAE-CATT-Precoding (variational denoising autoencoder model with cross attention precoding), in which cross attention enhances channel precoding by reevaluating the significance of the time domain. A variational autoencoder (VAE) is trained with reconstruction and regularization losses for parameter optimization, and its learned latent space facilitates feature capture for intra-channel precoding in both the frequency and spatial domains. We use an autoencoder to reconstruct the latent space, improving our analog precoding module’s ability to capture global domain features. Additionally, a CNN models frequency and spatial domain intricacies for efficient channel precoding, reducing interference in the proposed digital precoding module. Numerical results show remarkable improvements in MSE, achievable rate, generalizability, and robustness compared to previous works.

3 March 2025 in IEEE Transactions on Machine Learning in Communications and Networking
Verified
Journal Article
Open access
MWMeng-Hsun Wu
Meng-Hsun Wu
HCHong-Yunn Chen
Hong-Yunn Chen
TYTa-Wei Yang
Ta-Wei Yang
CHChih-Chuan Hsu
Chih-Chuan Hsu
CHChih-Wei Huang
Chih-Wei Huang
CCCheng-Fu Chou
Cheng-Fu Chou
In massive MIMO systems, achieving optimal end-to-end transmission encompasses various aspects such as power control, modulation schemes, path selection, and accurate channel estimation. Nonetheless, optimizing resource allocation remains a significant challenge. In path selection, the direct link is a straightforward link between the transmitter and the receiver. On the other hand, the indirect link involves reflections, diffraction, or scattering, often due to interactions with objects or obstacles. Relying exclusively on one type of link can lead to suboptimal and limited performance. Link management (LM) is emerging as a viable solution, and accurate channel estimation provides essential information to make informed decisions about transmission parameters. In this paper, we study LM and channel estimation that flexibly adjust the transmission ratio of direct and indirect links to improve generalization, using a denoising variational autoencoder with attention modules (DVAE-ATT) to enhance sum rate. Our experiments show significant improvements in IRS-assisted millimeter-wave MIMO systems. Incorporating LM increased the sum rate and reduced MSE by approximately 9%. Variational autoencoders (VAE) outperformed traditional autoencoders in the spatial domain, as confirmed by heatmap analysis. Additionally, our investigation of DVAE-ATT reveals notable differences in the temporal domain with and without attention mechanisms. Finally, we analyze performance across varying numbers of users and ranges. Across various distances—5m, 15m, 25m, and 35m—performance improvements averaged 6%, 11%, 16%, and 22%, respectively.