Scholarlink

<p>In this study, a dynamic resource allocation framework for fifth-generation (5G) and beyond wireless networks is presented. The work combines neural networks with adaptive modulation and Low-Density Parity-Check (LDPC) coded transmission. The proposed model predicts per-subchannel traffic demand using historical time series and spatial correlations, enabling real-time bandwidth adaptation across multiple subchannels. Modulation and coding schemes are adjusted based on predicted demand and the observed Signal-to-Noise Ratio (SNR), thereby improving transmission efficiency while maintaining reliable performance under Rayleigh fading. Experimental results demonstrate strong prediction accuracy, with a mean absolute error of 0.036 and an R<sup>2</sup> score above 0.97 across all subchannels. The adaptive modulation and coding strategy yields substantial improvements in bit error rates (BER), with values consistently between 1.2 × 10⁻<sup>2</sup> and 5.9 × 10⁻<sup>2</sup>, a significant reduction compared to fixed-modulation baselines. Visualization of bandwidth shifts reveals effective reallocation aligned with traffic bursts. These findings underscore the practical value of combining the proposed forecasting model with adaptive physical-layer strategies to enhance link robustness and spectral efficiency in next-generation networks.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Intelligent bandwidth allocation in 5G/6G subchannels based on predictive traffic models

  • 期刊论文

  • 发表日期：

    2026年7月10日

查看全文
  • Bülent Bilgehan,
  • Özlem Sabuncu

摘要

In this study, a dynamic resource allocation framework for fifth-generation (5G) and beyond wireless networks is presented. The work combines neural networks with adaptive modulation and Low-Density Parity-Check (LDPC) coded transmission. The proposed model predicts per-subchannel traffic demand using historical time series and spatial correlations, enabling real-time bandwidth adaptation across multiple subchannels. Modulation and coding schemes are adjusted based on predicted demand and the observed Signal-to-Noise Ratio (SNR), thereby improving transmission efficiency while maintaining reliable performance under Rayleigh fading. Experimental results demonstrate strong prediction accuracy, with a mean absolute error of 0.036 and an R2 score above 0.97 across all subchannels. The adaptive modulation and coding strategy yields substantial improvements in bit error rates (BER), with values consistently between 1.2 × 10⁻2 and 5.9 × 10⁻2, a significant reduction compared to fixed-modulation baselines. Visualization of bandwidth shifts reveals effective reallocation aligned with traffic bursts. These findings underscore the practical value of combining the proposed forecasting model with adaptive physical-layer strategies to enhance link robustness and spectral efficiency in next-generation networks.