Scholarlink

<p>This paper presents the Congestion-aware SDN–Fog with Deep Reinforcement Learning (CSFD) framework, a novel hybrid architecture that integrates SDN’s centralized control with Fog-layer adaptive learning to mitigate congestion in vehicular networks. Unlike existing DRL-based routing or SDN-only models, CSFD introduces a DRL-enabled Fog decision layer that performs congestion prediction and dynamic route optimization in coordination with SDN controllers. Simulation results demonstrate that CSFD achieves higher delivery ratio, lower latency, and reduced routing overhead compared to recent DRL and SDN–Fog baselines. This paper presents the approach, with congestion reduction SDN and Fog integration using Deep reinforcement learning to minimize the congestion in vehicular newtwork. It elucidates the intricacies of the network model, SDN processing, device management, and routing mechanisms within an SDN-enabled VANET environment, highlighting dynamic resource allocation and adaptive routing strategies. Additionally, it empowers autonomous decision-making for data scheduling and resource optimization and implements parked vehicle routing to manage communication in remote areas. The proposed approach, Speed and Position-aware Dynamic Routing (SPDR) is introduced to facilitate the efficient dissemination of Emergency Messages (EMs) on motorways. SPDR incorporates a dynamic greedy routing mechanism based on speed metrics, positional awareness, and a collaborative forwarding strategy. EMs are relayed incrementally from the source vehicle to the intended recipient through the hop-by-hop transmission. This strategy effectively mitigates the risk of selected forwarders moving out of the reception range during message forwarding, ensuring reliable transmission and timely delivery of EMs. From the results, it is evident that the CSFD protocol significantly improves the performance compared to SDN. While lowering the average end-to-end delay and packet dropping ratio by 26.35 and 21.09%, respectively, the goodput, throughput, and packet delivery ratio are enhanced by 26.353, 5.277, and 5.27%.</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

CSFD: Enhancing Congestion Reduction Vehicular Communication with SDN and Fog integration using Deep Reinforcement Learning

  • 期刊论文

  • 发表日期：

    2025年12月27日

查看全文
  • R. Ramesh,
  • K. V. Anusuya,
  • R. Sanjeev Kumar

摘要

This paper presents the Congestion-aware SDN–Fog with Deep Reinforcement Learning (CSFD) framework, a novel hybrid architecture that integrates SDN’s centralized control with Fog-layer adaptive learning to mitigate congestion in vehicular networks. Unlike existing DRL-based routing or SDN-only models, CSFD introduces a DRL-enabled Fog decision layer that performs congestion prediction and dynamic route optimization in coordination with SDN controllers. Simulation results demonstrate that CSFD achieves higher delivery ratio, lower latency, and reduced routing overhead compared to recent DRL and SDN–Fog baselines. This paper presents the approach, with congestion reduction SDN and Fog integration using Deep reinforcement learning to minimize the congestion in vehicular newtwork. It elucidates the intricacies of the network model, SDN processing, device management, and routing mechanisms within an SDN-enabled VANET environment, highlighting dynamic resource allocation and adaptive routing strategies. Additionally, it empowers autonomous decision-making for data scheduling and resource optimization and implements parked vehicle routing to manage communication in remote areas. The proposed approach, Speed and Position-aware Dynamic Routing (SPDR) is introduced to facilitate the efficient dissemination of Emergency Messages (EMs) on motorways. SPDR incorporates a dynamic greedy routing mechanism based on speed metrics, positional awareness, and a collaborative forwarding strategy. EMs are relayed incrementally from the source vehicle to the intended recipient through the hop-by-hop transmission. This strategy effectively mitigates the risk of selected forwarders moving out of the reception range during message forwarding, ensuring reliable transmission and timely delivery of EMs. From the results, it is evident that the CSFD protocol significantly improves the performance compared to SDN. While lowering the average end-to-end delay and packet dropping ratio by 26.35 and 21.09%, respectively, the goodput, throughput, and packet delivery ratio are enhanced by 26.353, 5.277, and 5.27%.