Scholarlink

<p>The rapid rise of IoT devices, which are expected to reach over 75&#xa0;billion by 2025 and generate 175 zettabytes of data each year, has shown that traditional cloud computing has problems with latency and bandwidth. This means that fog-cloud architectures are needed for IoT processing in real time. This paper suggests the DLSFC-Enhanced (DLSFC-E) algorithm, which builds on the Data-Locality Aware Job Scheduling in Fog-Cloud (DLSFC) technique and uses a multi-objective optimization framework to solve these problems. DLSFC-E uses a Directed Acyclic Graph (DAG) to show how tasks depend on each other, adds dynamic data replication based on how people use the system, and includes realistic network dynamics (bandwidth 10–100 Mbps ± 20%, latency 1–10 ms ± 15%) in simulations of a three-layer IoT-fog-cloud system with 10 fog nodes. CloudSim 4.0 simulations and real-world traffic statistics from Amsterdam on a 5-node physical testbed are used to check the method. The results reveal that DLSFC-E is 85% in line with the best Linear Programming (LP) solutions, cuts the makespan by 2.8 to 5.2 times compared to centralized methods, and lowers migration expenses by 15% compared to DLSFC. It improves runtime scalability by 40% for 1000 or more activities <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="13677_2025_772_Article_IEq1.gif" Format="GIF" Height="19" Rendition="HTML" Resolution="72" Type="Linedraw" Width="71" /> </InlineMediaObject> <EquationSource Format="TEX">\(\:O\left(n\text{log}n\right)\)</EquationSource> </InlineEquation> complexity and energy efficiency by 14%, with 92% of tasks keeping latency below 10 ms. These results, which were tested on datasets ranging from 30 to 600&#xa0;MB, show that DLSFC-E is strong enough for IoT installations on a broad scale. The study finds that DLSFC-E is a scalable and efficient way to schedule things, but there are still problems to solve, such as making sure it works when the network goes down and adjusting the weight of tasks as needed.</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

A novel Location-Aware job scheduling framework for optimizing Fog-Cloud IoT systems: insights from dynamic traffic management

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年9月30日

查看全文
  • Xiaomo Yu,
  • Mingjun Zhu,
  • Menghan Zhu,
  • Xiaomeng Zhou,
  • Long Long

摘要

The rapid rise of IoT devices, which are expected to reach over 75 billion by 2025 and generate 175 zettabytes of data each year, has shown that traditional cloud computing has problems with latency and bandwidth. This means that fog-cloud architectures are needed for IoT processing in real time. This paper suggests the DLSFC-Enhanced (DLSFC-E) algorithm, which builds on the Data-Locality Aware Job Scheduling in Fog-Cloud (DLSFC) technique and uses a multi-objective optimization framework to solve these problems. DLSFC-E uses a Directed Acyclic Graph (DAG) to show how tasks depend on each other, adds dynamic data replication based on how people use the system, and includes realistic network dynamics (bandwidth 10–100 Mbps ± 20%, latency 1–10 ms ± 15%) in simulations of a three-layer IoT-fog-cloud system with 10 fog nodes. CloudSim 4.0 simulations and real-world traffic statistics from Amsterdam on a 5-node physical testbed are used to check the method. The results reveal that DLSFC-E is 85% in line with the best Linear Programming (LP) solutions, cuts the makespan by 2.8 to 5.2 times compared to centralized methods, and lowers migration expenses by 15% compared to DLSFC. It improves runtime scalability by 40% for 1000 or more activities \(\:O\left(n\text{log}n\right)\) complexity and energy efficiency by 14%, with 92% of tasks keeping latency below 10 ms. These results, which were tested on datasets ranging from 30 to 600 MB, show that DLSFC-E is strong enough for IoT installations on a broad scale. The study finds that DLSFC-E is a scalable and efficient way to schedule things, but there are still problems to solve, such as making sure it works when the network goes down and adjusting the weight of tasks as needed.