Scholarlink

<p>This paper investigates a two-way multi-reconfigurable intelligent surface (RIS)-assisted fully mobile terahertz vehicular network under half-duplex time-division duplexing protocol, where both base station and RISs are vehicle-mounted. To address blockage and interference, rate-splitting multiple access (RSMA), non-orthogonal multiple access (NOMA), and orthogonal multiple access (OMA) are considered. Exact and asymptotic outage probability and ergodic rate expressions are derived for RSMA and NOMA. High signal-to-noise ratio (SNR) analysis shows downlink diversity order increases with number of RIS elements for RSMA and NOMA, while uplink diversity order is zero due to residual interference. Downlink RSMA ergodic rates saturate, whereas NOMA achieves slope one for the near user. To maximize joint uplink–downlink sum-rate, a non-convex problem is solved using twin delayed deep deterministic policy gradient-based deep reinforcement learning. Simulation results validate the analysis and demonstrate: (<i>i</i>) RSMA achieves the lowest downlink outage probability for the near user, outperforming NOMA and OMA; (<i>ii</i>) NOMA achieves a higher downlink ergodic rate than RSMA for the near user at high SNR, while in the uplink, the first near user message in RSMA and the near user in NOMA exhibit ergodic rate floors due to interference; (<i>iii</i>) RSMA consistently achieves the highest sum-rate across various RIS configurations. These findings highlight the effectiveness of RSMA in balancing reliability and spectral efficiency for fully mobile RIS-assisted 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

Two-way THz vehicular communications with mobile RISs: RSMA/NOMA performance analysis and DRL based optimization

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年8月1日

查看全文
  • Azadeh Khazali,
  • Mahrokh G. Shayesteh

摘要

This paper investigates a two-way multi-reconfigurable intelligent surface (RIS)-assisted fully mobile terahertz vehicular network under half-duplex time-division duplexing protocol, where both base station and RISs are vehicle-mounted. To address blockage and interference, rate-splitting multiple access (RSMA), non-orthogonal multiple access (NOMA), and orthogonal multiple access (OMA) are considered. Exact and asymptotic outage probability and ergodic rate expressions are derived for RSMA and NOMA. High signal-to-noise ratio (SNR) analysis shows downlink diversity order increases with number of RIS elements for RSMA and NOMA, while uplink diversity order is zero due to residual interference. Downlink RSMA ergodic rates saturate, whereas NOMA achieves slope one for the near user. To maximize joint uplink–downlink sum-rate, a non-convex problem is solved using twin delayed deep deterministic policy gradient-based deep reinforcement learning. Simulation results validate the analysis and demonstrate: (i) RSMA achieves the lowest downlink outage probability for the near user, outperforming NOMA and OMA; (ii) NOMA achieves a higher downlink ergodic rate than RSMA for the near user at high SNR, while in the uplink, the first near user message in RSMA and the near user in NOMA exhibit ergodic rate floors due to interference; (iii) RSMA consistently achieves the highest sum-rate across various RIS configurations. These findings highlight the effectiveness of RSMA in balancing reliability and spectral efficiency for fully mobile RIS-assisted networks.