DRL-D: Revenue-Aware Online Service Function Chain Deployment via Deep Reinforcement Learning

Fan, Qilin; Pan, Pan; Li, Xiuhua; Wang, Sen; Li, Jian; Wen, Junhao

doi:10.1109/tnsm.2022.3181517

Cited by 11 publications

(2 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Regarding SFC deployment, extensive research has been conducted, and the problem has been proven to be an NPhard problem [27,28]. Various methods have been proposed to solve the SFC deployment problem, including mathematical methods [29][30][31], heuristic algorithms [32][33][34][35], artificial intelligence algorithms [36][37][38], and game theory methods [39,40]. Cohen et al [30] have presented a formula for an integer linear programing algorithm, as well as an approximate algorithm, which simplifies the SFC deployment problem by constraining it into a generalized assignment problem.…”

Section: Introductionmentioning

confidence: 99%

“…Wei et al [37] have proposed a resource management architecture for a multiagent service chain based on reinforcement learning using the classic Q-learning algorithm for SFC deployment. Fan et al [38] have presented an online method for SFC deployment based on deep reinforcement learning, improving the longterm average revenue. Liu et al [39] have proposed a game theory-based method to study SFC resource management in edge computing, aiming to minimize delay.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Intelligent Deployment of Delay-Sensitive Service Function Chain Based on Parallelization and Improved Cuckoo Search Algorithm

Zhao,

Kang,

Wang

et al. 2023

Wireless Communications and Mobile Computing

View full text Add to dashboard Cite

As one of the key focuses in 6G research, the space–air–ground integrated network incorporates a variety of technological frameworks. Network function virtualization allows network functions to be deployed on general servers in the form of software and creates a service function chain (SFC) according to user service requirements. In recent years, the deployment of SFC has become popular research due to the increasing demand for low delay in network application scenarios. Low delay is a crucial indicator of the quality of service, especially for delay-sensitive applications. To address this issue, we propose a method for the deployment of delay-sensitive SFC based on parallelization and the improved cuckoo search (ICS) algorithm (DDSSFC-PICS). This method optimizes the composition and deployment of SFC jointly. First, the serial structure of the SFC is transformed into a parallel structure by determining the dependency of virtual network functions, which reduces the length of the SFC and thereby reduces delay. Second, with the optimization goal of minimizing network delay, a parallel SFC deployment model is established under constraints including packet loss rate and resource availability. Finally, the ICS algorithm is applied for optimization, where delay is used as the fitness measure. By improving the Lévy flight step size and drawing inspiration from the whale algorithm, the performance of the cuckoo search (CS) algorithm is enhanced, leading to a further reduction in delay. The simulation results show that using the same CS deployment method, parallelized SFC has a significantly lower delay compared to serial SFC. Furthermore, the DDSSFC-PICS reduces the delay by 22.58% and 19.02%, respectively, compared with the CS deployment and particle swarm optimization SFC deployment methods.

show abstract