一种Femtocell网络中的负载预测方法

A load prediction method for Femtocell network

在未来的5G移动通信系统中,Femtocell等小功率基站将承担起海量的数据通信业务。通过对Femtocell做出负载预测,进而完成小基站的休眠策略和对流量的管控等,但现有的研究中未出现此类探索方向。为了进一步实现移动通信系统的绿色通信和预知其对未来负载的发展态势,在Femtocell下提出了一种基于混沌特性和改进径向基函数(radial basis function,RBF)神经网络的预测方案。采用了C-C(Catmall-Clark)算法和G-P(GrassbergerProcaccia)算法计算时间序列的延迟时间和维度,根据最小数据量法求取Lyapunov指数,验证了Femtocell的负载变化存在混沌特性,并对Femtocell负载的时间序列进行相空间重构,采用了粒子群优化(particle swarm optimization,PSO)算法和改进的RBF进行优化、学习和预测。通过对比仿真实验的结果表明,该方案预测效果较好,可以应用到实际的预测工作中。

In the 5 G mobile communication system,low power stations like Femtocell will shoulder the responsibility of vast communication data. Predicting the load of Femtocell will benefit the sleep strategy and traffic control,but it’s difficult to find the analogous direction in existing research. In order to further implement the green communication of mobile communication system and predict the developing trend of the future load,a prediction scheme based on chaotic characteristics and radial basis function( RBF) neural network is proposed in Femtocell. Firstly,the delay time and dimension of time series can be calculated by C-C algorithm and G-P algorithm and the Lyapunov index is calculated according to the small data method,and the existence of chaotic characteristics in load variation of Femtocell is verified. Then the phase space reconstruction of the time series of the Femtocell load is carried out. Finally,PSO( Particle Swarm Optimization) algorithm and the improved RBF are used to optimize,learn and predict the developing trend. By comparing the simulation results,the results show that the proposed scheme has the better performance of prediction and can be applied to the actual forecast work.