NeTS: Medium: Collaborative Research: Big Data Enabled Wireless Networking: A Deep Learning Approach

NeTS：媒介：协作研究：大数据支持的无线网络：深度学习方法

• 批准号: 1704662
• 负责人: Jian Tang
• 金额: $ 70万
• 依托单位: Syracuse University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-08-15 至 2021-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1704662&HistoricalAwards=false
• 关键词: NeTS; Medium; Collaborative; Research; Big

Wireless networks are becoming larger and more complicated, generating a huge amount of runtime statistics data (such as traffic load, resource usages, etc.) every second. Instead of treating big data in wireless networks as an unwanted burden, we aim to leverage them as a great opportunity for better understanding user demands and system capabilities such that we can optimize resource allocation to better serve mobile users. In addition, Cloud Radio Access Networks (C-RANs) have become a key enabling technology for the next generation wireless communication systems. Their centralized architecture makes it easy to collect and analyze various runtime system data. This project aims to exploit how the powerful new machine learning techniques, including Deep Learning (DL) and Deep Reinforcement Learning (DRL), can be leveraged to grasp the exciting opportunity provided by big data to enable future wireless networks to better serve their users. The proposed research is expected to significantly improve resource utilization of wireless networks and reduce their operational costs (such as power consumption), which can substantially benefit wireless network carriers and mobile users, and more importantly, is good for global environment. Beyond wireless networking, the proposed DL models and algorithms may find its applications in a large variety of domains, including video content analysis, user behavior study, etc. Moreover, the proposed project is expected to advance public understanding of the emerging 5G wireless communications, DL and DRL via publications, seminars and workshops, and international and industrial collaborations. The objective of this project is to develop a novel deep learning approach to enable efficient design and operations of future wireless networks with big data. Specifically, we will propose DL models and algorithms for spatiotemporal analysis and prediction of key system parameters, which can provide accurate and useful input information for existing resource allocation algorithms to better operate a wireless network. Moreover, we will develop a novel DRL-based control framework for a wireless network to efficiently allocate its resources by jointly learning the system environment and making decisions under the guidance of a powerful deep neural network. To achieve the above object, the project is organized into three cohesive thrusts: Thrust 1 Deep Learning based Modeling and Prediction; Thrust 2 Deep Reinforcement Learning based Dynamic Resource Allocation; and Thrust 3 Validation and Performance Evaluation.

无线网络正变得越来越大和复杂，产生大量的运行时统计数据（如流量负载、资源使用等）。每秒我们的目标不是将无线网络中的大数据视为不必要的负担，而是将其作为更好地了解用户需求和系统功能的绝佳机会，以便优化资源分配，更好地为移动的用户提供服务。此外，云无线电接入网络（C-RAN）已经成为下一代无线通信系统的关键使能技术。它们的集中式架构使收集和分析各种运行时系统数据变得容易。该项目旨在探索如何利用强大的新机器学习技术，包括深度学习（DL）和深度强化学习（DRL），来抓住大数据提供的令人兴奋的机会，使未来的无线网络能够更好地为用户服务。该研究有望显著提高无线网络的资源利用率，降低其运营成本（如功耗），这将大大有利于无线网络运营商和移动的用户，更重要的是，有利于全球环境。除了无线网络之外，拟议的DL模型和算法可能会在各种领域中找到应用，包括视频内容分析，用户行为研究等。此外，拟议的项目预计将通过出版物，研讨会和研讨会以及国际和行业合作来促进公众对新兴的5G无线通信，DL和DRL的理解。该项目的目标是开发一种新的深度学习方法，以实现未来大数据无线网络的高效设计和运营。具体来说，我们将提出DL模型和算法的时空分析和预测的关键系统参数，这可以提供准确和有用的输入信息，现有的资源分配算法，以更好地运行无线网络。此外，我们将为无线网络开发一种新的基于DRL的控制框架，通过联合学习系统环境并在强大的深度神经网络的指导下做出决策来有效地分配其资源。为了实现上述目标，该项目被组织成三个有凝聚力的推力：推力1基于深度学习的建模和预测;推力2基于深度强化学习的动态资源分配;推力3验证和性能评估。

项目成果

EXTRA: An Experience-driven Control Framework for Distributed Stream Data Processing with a Variable Number of Threads

EXTRA：用于具有可变线程数的分布式流数据处理的体验驱动控制框架

• DOI: 10.1109/iwqos52092.2021.9521325
• 发表时间: 2021
• 期刊: IEEE/ACM IWQoS
• 作者: Li, Teng;Xu, Zhiyuan;Tang, Jian;Wu, Kun;Wang, Yanzhi
• 通讯作者: Wang, Yanzhi

Experience-driven Networking: A Deep Reinforcement Learning based Approach

• DOI: 10.1109/infocom.2018.8485853
• 发表时间: 2018-01
• 期刊: IEEE INFOCOM 2018 - IEEE Conference on Computer Communications
• 作者: Zhiyuan Xu;Jian Tang;Jingsong Meng;Weiyi Zhang;Yanzhi Wang;C. Liu;Dejun Yang

PnP-DRL: A Plug-and-Play Deep Reinforcement Learning Approach for Experience-Driven Networking

• DOI: 10.1109/jsac.2021.3087270
• 发表时间: 2021-08
• 期刊: IEEE Journal on Selected Areas in Communications
• 影响因子: 16.4
• 作者: Zhiyuan Xu;Kun Wu;Weiyi Zhang;Jian Tang;Yanzhi Wang;G. Xue

Towards Ultra-High Performance and Energy Efficiency of Deep Learning Systems: An Algorithm-Hardware Co-Optimization Framework

• DOI: 10.1609/aaai.v32i1.11653
• 发表时间: 2018-02
• 期刊: ArXiv
• 作者: Yanzhi Wang;Caiwen Ding;Zhe Li;Geng Yuan;Siyu Liao;Xiaolong Ma;Bo Yuan;Xuehai Qian;Jian Tang;Qinru Qiu;X. Lin

A Deep Recurrent Neural Network Based Predictive Control Framework for Reliable Distributed Stream Data Processing

基于深度循环神经网络的可靠分布式流数据处理的预测控制框架

• DOI: 10.1109/ipdps.2019.00036
• 发表时间: 2019
• 期刊: Proceedings of the IEEE Symposium on Parallel and Distributed Processing
• 作者: Xu, Jielong Xu;Tang, Jian;Xu, Zhiyuan Xu;Yin, Chengxiang;Kwiat, Kevin;Kamhoua, Charles
• 通讯作者: Kamhoua, Charles