An Evolutionary Game for Distributed Resource Allocation in Self-Organizing Small Cells
An Evolutionary Game for Distributed Resource Allocation in Self-Organizing Small Cells
复制标题
DOI：
10.1109/tmc.2014.2318700
复制
发表时间：
2015-02-01
影响因子：
7.9
通讯作者：
Zhu, Kun
中科院分区：
文献类型：
--
作者：
Semasinghe, Prabodini;Hossain, Ekram;Zhu, Kun
我们提出了一种基于进化博弈论（EGT）的小细胞分布式资源分配方案。EGT是解决自组织小细胞中资源分配问题的合适工具，因为它允许具有有限理性的参与者从环境中学习，并采取个体决策，以最小的信息交换达到均衡。基于egt的资源分配还可以提供用户间的公平性。我们展示了EGT如何在基于正交频分多址（OFDMA）的小蜂窝网络中用于分布式子载波和功率分配，同时将对大蜂窝用户的干扰限制在给定阈值以下。考虑了两种博弈模型，其中每个小单元的效用分别取决于可实现的平均信噪比（SINR）和数据速率。对于所提出的分布式资源分配方法，基于随机几何分析得到了平均信噪比和数据速率。利用复制因子动力学对小蜂窝基站的策略适应过程进行了建模，得到了一种进化均衡解。根据随机几何的结果，分析了平衡的稳定性。我们还扩展了该公式，考虑了信息交换延迟，并研究了它对算法收敛性的影响。数值结果验证了我们的理论发现，并与集中式资源分配方案相比，显示了所提出方案的有效性。
We propose an evolutionary game theory (EGT)-based distributed resource allocation scheme for small cells underlaying a macro cellular network. EGT is a suitable tool to address the problem of resource allocation in self-organizing small cells since it allows the players with bounded-rationality to learn from the environment and take individual decisions for attaining the equilibrium with minimum information exchange. EGT-based resource allocation can also provide fairness among users. We show how EGT can be used for distributed subcarrier and power allocation in orthogonal frequency-division multiple access (OFDMA)-based small cell networks while limiting interference to the macrocell users below given thresholds. Two game models are considered, where the utility of each small cell depends on average achievable signal-to-interference-plus-noise ratio (SINR) and data rate, respectively. For the proposed distributed resource allocation method, the average SINR and data rate are obtained based on a stochastic geometry analysis. Replicator dynamics is used to model the strategy adaptation process of the small cell base stations and an evolutionary equilibrium is obtained as the solution. Based on the results obtained using stochastic geometry, the stability of the equilibrium is analyzed. We also extend the formulation by considering information exchange delay and investigate its impact on the convergence of the algorithm. Numerical results are presented to validate our theoretical findings and to show the effectiveness of the proposed scheme in comparison to a centralized resource allocation scheme.