Adapting Wireless Mesh Network Configuration from Simulation to Reality via Deep Learning based Domain Adaptation
Adapting Wireless Mesh Network Configuration from Simulation to Reality via Deep Learning based Domain Adaptation
复制标题
DOI：
--
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
Junyang Shi;M. Sha;Xi Peng
中科院分区：
文献类型：
--
作者：
Junyang Shi;M. Sha;Xi Peng
近年来，无线网状网络（WMN）在工业自动化、军事行动、智能能源等领域得到了快速部署。尽管经过多年的研究，WMN在大多数情况下都能令人满意地工作，但它们通常很难配置，因为配置WMN是一个复杂的过程，涉及理论计算、仿真和现场测试等任务。在识别良好的网络配置时，模拟WMN比在物理网络上进行实验具有明显的优势。不幸的是，我们的研究表明，由于模拟与现实之间的差距，从模拟中学习的网络配置预测模型并不总能帮助物理网络满足性能要求。在本文中，我们采用基于深度学习的领域自适应来缩小差距，并利用师生神经网络将从模拟网络学习到的网络配置知识转移到相应的物理网络。实验结果表明，我们的方法有效地缩小了差距，并提高了预测一个良好的网络配置的准确性，使网络满足性能要求的30.10%至70.24%，通过学习强大的机器学习模型，从大量的廉价的仿真数据和一些昂贵的现场测试测量。
Recent years have witnessed the rapid deployments of wireless mesh networks (WMNs) for industrial automation, military operations, smart energy, etc. Although WMNs work satisfactorily most of the time thanks to years of research, they are often difficult to configure as configuring a WMN is a complex process, which involves theoretical computation, simulation, and field testing, among other tasks. Simulating a WMN provides distinct advantages over experimenting on a physical network when it comes to identifying a good network configuration. Unfortunately, our study shows that the models for network configuration prediction learned from simulations cannot always help physical networks meet performance requirements because of the simulation-to-reality gap. In this paper, we employ deep learning based domain adaptation to close the gap and leverage a teacher-student neural network to transfer the network configuration knowledge learned from a simulated network to its corresponding physical network. Experimental results show that our method effectively closes the gap and increases the accuracy of predicting a good network configuration that allows the network to meet performance requirements from 30.10% to 70.24% by learning robust machine learning models from a large amount of inexpensive simulation data and a few costly field testing measurements.