Scholarlink

<p>The terahertz (THz) satellite communications technology will allow massive amounts of bandwidth; however, it is greatly attenuated by the atmospheric gases like water vapor, oxygen, and dynamic changes from the weather. Most current approaches to optimizing THz links have relied on abstract channel models, which fail to utilize actual, real-time measurements of the atmosphere. As a result of this limitation, these methodologies have a restricted applicability in the operational environment. This paper introduces a new methodology for intelligent atmospheric attenuation compensation through a combination of multi-source hyperspectral satellite observations and a deep learning model called SSACNet-THz, which directly measures atmospheric key parameters, including water vapor, temperature, pressure, and aerosol optical depth. These measured parameters are then used in conjunction with an adaptive THz link optimization (ATLO) algorithm to dynamically adjust the amount of transmitted power and allocated frequency to achieve maximum link performance, while ensuring that the action does not exceed the physical limitations of the satellite hardware. A large-scale test of the methodology has been completed, utilizing a total of 12,800 data samples collected from satellite platforms and ground-based validation networks. This demonstrates excellent prediction accuracy (<i>R</i><sup>2</sup> &gt; 0.93 across all parameters) and significant communication improvements, with 4.2 times greater throughput and 99.7% available link connections. The methodology can operate at near-real-time speeds, with a latency of 67.8&#xa0;ms for neural network inference and an end-to-end latency of approximately 3.2&#xa0;s. This enables deployment on low-Earth orbit (LEO) satellites with contact periods of 15&#xa0;min or less, and potentially on-board processing capabilities (projected latency of 180–250&#xa0;ms). This paper provides the first comprehensive integration of hyperspectral image processing and THz communication optimization, providing a means to develop reliable and high-capacity next-generation satellite communications systems.</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 Atmospheric Attenuation Mitigation in Terahertz Satellite Communication Using Adaptive Image Processing of Hyperspectral Data

  • 期刊论文

  • 发表日期：

    2025年12月4日

查看全文
  • Xinyu Cui,
  • Xinyue Zhang

摘要

The terahertz (THz) satellite communications technology will allow massive amounts of bandwidth; however, it is greatly attenuated by the atmospheric gases like water vapor, oxygen, and dynamic changes from the weather. Most current approaches to optimizing THz links have relied on abstract channel models, which fail to utilize actual, real-time measurements of the atmosphere. As a result of this limitation, these methodologies have a restricted applicability in the operational environment. This paper introduces a new methodology for intelligent atmospheric attenuation compensation through a combination of multi-source hyperspectral satellite observations and a deep learning model called SSACNet-THz, which directly measures atmospheric key parameters, including water vapor, temperature, pressure, and aerosol optical depth. These measured parameters are then used in conjunction with an adaptive THz link optimization (ATLO) algorithm to dynamically adjust the amount of transmitted power and allocated frequency to achieve maximum link performance, while ensuring that the action does not exceed the physical limitations of the satellite hardware. A large-scale test of the methodology has been completed, utilizing a total of 12,800 data samples collected from satellite platforms and ground-based validation networks. This demonstrates excellent prediction accuracy (R2 > 0.93 across all parameters) and significant communication improvements, with 4.2 times greater throughput and 99.7% available link connections. The methodology can operate at near-real-time speeds, with a latency of 67.8 ms for neural network inference and an end-to-end latency of approximately 3.2 s. This enables deployment on low-Earth orbit (LEO) satellites with contact periods of 15 min or less, and potentially on-board processing capabilities (projected latency of 180–250 ms). This paper provides the first comprehensive integration of hyperspectral image processing and THz communication optimization, providing a means to develop reliable and high-capacity next-generation satellite communications systems.