Software-Defined Device-to-Device (D2D) Communications in Virtual Wireless Networks With Imperfect Network State Information (NSI)
Software-Defined Device-to-Device (D2D) Communications in Virtual Wireless Networks With Imperfect Network State Information (NSI)
复制标题
DOI：
10.1109/tvt.2015.2483558
复制
发表时间：
2016-09
影响因子：
6.8
通讯作者：
Yegui Cai;F. Yu;C. Liang;Bo Sun;Qiao Yan
中科院分区：
文献类型：
--
作者：
Yegui Cai;F. Yu;C. Liang;Bo Sun;Qiao Yan
软件定义的网络（SDN）和网络功能虚拟化（NFV）是一种有希望的未来网络的系统体系结构和控制机制。尽管在无线SDN和NFV上已经完成了一些工作，但是在这个新颖的框架中，设备到设备（D2D）通信的最新进步在很大程度上被忽略了。在本文中，我们研究了D2D通信在SDN和NFV框架中的整合。支持软件定义的D2D的固有挑战是网络状态信息的不完善性，包括渠道状态信息（CSI）和排队状态信息，在虚拟无线（QSI）网络中。为了应对这一挑战，我们将此框架中的资源共享问题作为离散的随机优化问题，并开发出离散的随机近似算法来解决此问题。与详尽的搜索相比，这种算法可以降低计算复杂性，同时达到令人满意的性能。都考虑了静态无线通道和时变通道。广泛的模拟表明，用户可以从无线网络虚拟化和软件定义的D2D通信中受益，而我们提出的计划可以在实际网络设置下在系统吞吐量和用户实用程序中获得可观的性能提高。
Software-defined networking (SDN) and network function virtualization (NFV) are a promising system architecture and control mechanism for future networks. Although some works have been done on wireless SDN and NFV, recent advancements in device-to-device (D2D) communications are largely ignored in this novel framework. In this paper, we study the integration of D2D communication in the framework of SDN and NFV. An inherent challenge in supporting software-defined D2D is the imperfectness of network state information, including channel state information (CSI) and queuing state information, in virtual wireless (QSI) networks. To address this challenge, we formulate the resource sharing problem in this framework as a discrete stochastic optimization problem and develop discrete stochastic approximation algorithms to solve this problem. Such algorithms can reduce the computational complexity compared with exhaustive search while achieving satisfactory performance. Both the static wireless channel and time-varying channels are considered. Extensive simulations show that users can benefit from both wireless network virtualization and software-defined D2D communications, and our proposed scheme can achieve considerable performance gains in both system throughput and user utility under practical network settings.