Ensemble Deep Learning Assisted VNF Deployment Strategy for Next-Generation IoT Services

Abstract

Due to the massive Internet of Things (IoT) connectivity and substantial growth of communication traffic, Virtual Network Function (VNF) orchestration scheme is anticipated to function promptly, dynamically, and intelligently for next-generation networks. Hence, we urge the necessity to move beyond the traditional paradigm and employ VNFs on the network edge located cloudlet. Overall, multi-access edge computing can intensify the performance of delay-sensitive IoT applications compared to the core cloud based VNF deployments. In this paper, we intend to investigate how to simultaneously leverage the ensembling of multiple deep learning models for proper calibration to provide real-time VNF placement solutions. We also address the challenges associated with state-of-the-art approaches to deal with dynamic network traffic and topology patterns. Our envisioned methods, based on Convolutional Neural Networks and Artificial Neural Networks named as E-ConvNets and E-ANN respectively, suggest two proactive VNF deployment strategies. These ensembled VNF deployment strategies demonstrate encouraging performance (optimality gap nearly 7%) in terms of minimizing relocation and communication costs, and high scalability intelligence factor (around 0.93) through simulation results compared to standalone deep learning models. Furthermore, the presented results indicate the potentialities of applying deep learning-based strategies into similar research enigmas for future telecommunication network researches.

Multi-Agent Deep Reinforcement Learning for Cost- and Delay-Sensitive Virtual Network Function Placement and Routing

“…Li et al [17] studied the VNF chaining in an optical network by employing a deep learning (DL)based prediction technique. Emu et al [18] used convolutional neural network (CNN) to minimize communication costs of VNF deployment. Pei et al [19] divided the VNF selection and chaining problem into two stages: first VNF selection, and then VNF chaining, and developed a DL-based two-phase algorithm comprising VNF selection and VNF chaining networks.…”