Towards User QoE-Centric Elastic Cellular Networks: A Game Theoretic Framework for Optimizing Throughput and Energy Efficiency

Towards User QoE-Centric Elastic Cellular Networks: A Game Theoretic Framework for Optimizing Throughput and Energy Efficiency
复制标题

DOI:
10.1109/pimrc.2018.8580747
复制
发表时间:
2018-09
期刊:
2018 IEEE 29th Annual International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC)
影响因子:
--
通讯作者:
U. Hashmi;A. Islam;K. Nasr;A. Imran
U. Hashmi;A. Islam;K. Nasr;A. Imran
中科院分区:
其他
文献类型:
--
作者:
U. Hashmi;A. Islam;K. Nasr;A. Imran

文献摘要

被引文献

相似文献

以用户为中心的网络架构是实现未来密集异构网络(HetNet)部署的统一体验质量(QoE)要求的关键支持者。然而,在这样的网络架构中,迎合由过多的不同移动的应用引起的时空变化的用户服务需求仍然是一个挑战。在本文中,我们提出了一个密集的多层蜂窝网络部署的QoE为中心的弹性框架。该框架利用控制和数据平面分离架构(CDSA)来实现以用户设备(UE)为中心的虚拟小区(也称为服务区)内的选择性数据基站(DBS)激活。围绕所选UE的这些虚拟弹性服务区的分配经由中央控制基站(CBS)进行,并通过两种游戏技术(即进化游戏和拍卖游戏)来建模。这两个游戏都基于效用最小化问题,效用最小化问题是加权平均UE吞吐量和基于使用的UE服务需求的函数。为了说明博弈模型之间的权衡,在总吞吐量、能量效率、算法收敛速度和平均UE调度概率方面比较了网络级性能。
User-centric network architectures are a key proponent to enable the uniform Quality of Experience (QoE) requirement for future dense heterogeneous network (HetNet) deployments. However, catering to spatio-temporally varying user service demands arising from the plethora of diverse mobile applications remains a challenge in such network architectures. In this paper, we propose a QoE-centric elastic framework for a dense multi-tier cellular network deployment. The framework leverages the control and data plane separation architecture (CDSA) for enabling selective data base station (DBS) activation within user equipment (UE)-centric virtual cells (also referred to as service zones). The allocation of these virtually elastic service zones around selected UEs is conducted via a central control base station (CBS) and modeled through two game techniques, namely evolutionary and auction games. Both the games are based on a utility minimization problem which is a function of weighted mean UE throughput and usage based UE service demands. To illustrate the trade-offs between the game models, network level performance is compared in terms of aggregate throughput, energy efficiency, algorithm convergence speed and mean UE scheduling probabilities.