Scholarlink

<p>Intelligent transportation systems (ITS) enhance road safety by transmitting messages, particularly decentralized environmental notification messages (DENM), that inform users of incidents or hazards. Integrating mobile edge computing (MEC) brings computational power closer to vehicles, reducing latency and optimizing efficiency. A key challenge is load balancing, which dynamically distributes workloads among neighboring MEC nodes to prevent local server congestion. Artificial intelligence strategies are being explored to forecast resource demands and improve data flow. Long-short-term memory (LSTM) algorithms predict future MEC load by analyzing historical traffic patterns and resource use. In this article, we present three load balancing approaches applied to MEC servers using the LSTM algorithm. Such a prediction facilitates data redistribution following the proposed load balancing methodologies. A software-defined vehicular network (SDVN) controller is adopted to help the local MEC server balance the load using our three proposed approaches according to varied use cases: LBA-MC, LBA-MDS, and LBA-TPS. LBA-MC focuses on the total available capacities of all available servers, optimizing resource usage. LBA-MDS takes into consideration the proximity between the neighboring server and the local MEC, prioritizing the geographically closer server to minimize latency. LBA-TPS proposes to update the local traffic priorities based on the average priority of the traffic treated by each neighboring MEC, thereby prioritizing the critical traffic. The results demonstrate that the proposed approaches improve decision-making regarding resource allocation, mitigate congestion, and enhance the quality of service (QoS) within the Internet of Vehicles (IoV), including considerations of end-to-end (E2E) delays.</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

Enhanced Traffic Management Approaches in Internet of Vehicles through LSTM-Driven Load Balancing for Mobile Edge Computing

  • 期刊论文

  • 发表日期：

    2026年1月23日

查看全文
  • Amal Hmaidi,
  • Hend Marouane,
  • Hassene Mnif,
  • Mohamed Mosbah

摘要

Intelligent transportation systems (ITS) enhance road safety by transmitting messages, particularly decentralized environmental notification messages (DENM), that inform users of incidents or hazards. Integrating mobile edge computing (MEC) brings computational power closer to vehicles, reducing latency and optimizing efficiency. A key challenge is load balancing, which dynamically distributes workloads among neighboring MEC nodes to prevent local server congestion. Artificial intelligence strategies are being explored to forecast resource demands and improve data flow. Long-short-term memory (LSTM) algorithms predict future MEC load by analyzing historical traffic patterns and resource use. In this article, we present three load balancing approaches applied to MEC servers using the LSTM algorithm. Such a prediction facilitates data redistribution following the proposed load balancing methodologies. A software-defined vehicular network (SDVN) controller is adopted to help the local MEC server balance the load using our three proposed approaches according to varied use cases: LBA-MC, LBA-MDS, and LBA-TPS. LBA-MC focuses on the total available capacities of all available servers, optimizing resource usage. LBA-MDS takes into consideration the proximity between the neighboring server and the local MEC, prioritizing the geographically closer server to minimize latency. LBA-TPS proposes to update the local traffic priorities based on the average priority of the traffic treated by each neighboring MEC, thereby prioritizing the critical traffic. The results demonstrate that the proposed approaches improve decision-making regarding resource allocation, mitigate congestion, and enhance the quality of service (QoS) within the Internet of Vehicles (IoV), including considerations of end-to-end (E2E) delays.