Scholarlink

<p>This paper introduces a novel Q-learning model designed for resource allocation within a heterogeneous vehicular network to elevate the Quality of Service (QoS) for diverse vehicular communication services. Integrating terrestrial Base Stations (BS), High Altitude Platform Stations (HAPS), and Low Earth Orbit (LEO) satellites, the network ensures extensive coverage. Vehicles are classified as fully automatic (FAVs), semi-automatic (SAVs), or manual vehicles (MVs), each assigned service priorities like autonomous driving (P0), vehicle health (P1), and infotainment (P2). On accumulating service requests in a centralized buffer, the cloud system processes them at fixed intervals within a strict latency threshold for timely allocation. Simulations assess the model’s resource allocation efficacy, following a priority scheme that aligns with each service’s latency sensitivity. Results confirm the model’s proficiency in upholding service reliability, particularly in high-demand scenarios, by prioritizing essential services without a notable QoS compromise. As request rates climbed, a predictable reliability dip emerged, yet the model adeptly maintained high reliability for critical services. The adaptability of the proposed Q-learning model to fluctuating resource statuses and demand patterns underscores its potential to enhance resource allocation decisions. This, in turn, contributes to the evolution of vehicular network resource management, paving the way for more robust and efficient communication frameworks in innovative transportation ecosystems.</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

Optimizing quality of service-based resource allocation in air-space-ground vehicular communication systems via Q-learning

  • 期刊论文

  • 发表日期：

    2026年4月17日

查看全文
  • Divyanshu Pandey,
  • K. L. V. Sai Prakash Sakuru

摘要

This paper introduces a novel Q-learning model designed for resource allocation within a heterogeneous vehicular network to elevate the Quality of Service (QoS) for diverse vehicular communication services. Integrating terrestrial Base Stations (BS), High Altitude Platform Stations (HAPS), and Low Earth Orbit (LEO) satellites, the network ensures extensive coverage. Vehicles are classified as fully automatic (FAVs), semi-automatic (SAVs), or manual vehicles (MVs), each assigned service priorities like autonomous driving (P0), vehicle health (P1), and infotainment (P2). On accumulating service requests in a centralized buffer, the cloud system processes them at fixed intervals within a strict latency threshold for timely allocation. Simulations assess the model’s resource allocation efficacy, following a priority scheme that aligns with each service’s latency sensitivity. Results confirm the model’s proficiency in upholding service reliability, particularly in high-demand scenarios, by prioritizing essential services without a notable QoS compromise. As request rates climbed, a predictable reliability dip emerged, yet the model adeptly maintained high reliability for critical services. The adaptability of the proposed Q-learning model to fluctuating resource statuses and demand patterns underscores its potential to enhance resource allocation decisions. This, in turn, contributes to the evolution of vehicular network resource management, paving the way for more robust and efficient communication frameworks in innovative transportation ecosystems.