Scholarlink

<p>Network Function Virtualization (NFV) has emerged as a transformative technology that decouples network functions from dedicated hardware, enabling flexible and scalable network service deployment. One of the key challenges in NFV is the placement of Service Function Chains (SFCs) in Mobile Edge Computing (MEC) networks, where traffic must traverse a sequence of Virtual Network Functions (VNFs) while minimizing latency. Although parallelization of VNFs plays an important role in reducing processing latencies, reuse of sharable VNF instances and efficient resource management should not be neglected. With this motivation, this paper presents a Latency-aware Parallelized SFCs Placement framework through VNF Instance sharing and Flexible resource allocation (LPPIF). By identifying the dependencies between VNFs, LPPIF transforms an original SFC into a parallelized SFC and solves the deployment problem using deep reinforcement learning approaches. By leveraging the concept of Network Function Parallelism (NFP), our framework reduces end-to-end service latency by allowing multiple VNFs to process traffic concurrently. Furthermore, the framework introduces a resource allocation mechanism that flexibly adapts to the varying demands of SFC requests, optimizing both resource usage and performance. Meanwhile, LPPIF enables the sharing of reusable VNFs across different SFCs, further enhancing resource efficiency. Extensive simulations demonstrate that LPPIF significantly improves the service latency by 3.1% and the acceptance rate of SFC requests by 0.53% compared to the best existing state-of-the-art method.</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

LPPIF: Latency-Aware Placement of Parallelized Service Function Chains Through Instance Sharing and Flexible Resource Allocation

  • 期刊论文

  • 发表日期：

    2025年9月8日

查看全文
  • Yechuan Lin,
  • Xianglan Jin,
  • Xiaojian Liu,
  • Zhixuan Zhang,
  • Amin Rezaeipanah

摘要

Network Function Virtualization (NFV) has emerged as a transformative technology that decouples network functions from dedicated hardware, enabling flexible and scalable network service deployment. One of the key challenges in NFV is the placement of Service Function Chains (SFCs) in Mobile Edge Computing (MEC) networks, where traffic must traverse a sequence of Virtual Network Functions (VNFs) while minimizing latency. Although parallelization of VNFs plays an important role in reducing processing latencies, reuse of sharable VNF instances and efficient resource management should not be neglected. With this motivation, this paper presents a Latency-aware Parallelized SFCs Placement framework through VNF Instance sharing and Flexible resource allocation (LPPIF). By identifying the dependencies between VNFs, LPPIF transforms an original SFC into a parallelized SFC and solves the deployment problem using deep reinforcement learning approaches. By leveraging the concept of Network Function Parallelism (NFP), our framework reduces end-to-end service latency by allowing multiple VNFs to process traffic concurrently. Furthermore, the framework introduces a resource allocation mechanism that flexibly adapts to the varying demands of SFC requests, optimizing both resource usage and performance. Meanwhile, LPPIF enables the sharing of reusable VNFs across different SFCs, further enhancing resource efficiency. Extensive simulations demonstrate that LPPIF significantly improves the service latency by 3.1% and the acceptance rate of SFC requests by 0.53% compared to the best existing state-of-the-art method.