论文部分内容阅读
针对长程突发通信量提出了两种基于α-平稳信息的预测方法:根据α-平稳过程的协变概念,推导出双曲线偏差渐近意义下的FARIMA(fractionally autoregressive integrated moving average)预测,采用自回归神经网络模拟ARMA过程,并利用遗传算法的全局优化能力与人工免疫算法的多种群快速局部收敛能力对神经网络权值进行准确估计,从而实现对通信量的FARIMA预测.这两种预测方法均能在无限方差准则下实现偏差最小,合并这两种预测值以获得最后的预测结果.对实际踪迹的预测结果证实了两种独立的预测方法有效准确,最后的混合预测能进一步提高最后的预测精度.
Two kinds of prediction methods based on α-stationary information are proposed for long-range burst traffic. According to the concept of covariance of α-stationary process, FARIMA (Fractional autoregressive integrated moving average) prediction is derived asymptotically in hyperbolic deviations. ARMA is used to simulate the ARMA process and the FARIMA prediction of traffic is realized by using the global optimization ability of genetic algorithm and the multi-swarm fast local convergence of artificial immune algorithm to accurately estimate the weights of the neural network. And the two predictors can be combined to obtain the final prediction result.The prediction results of the actual trace confirm that the two independent prediction methods are effective and accurate and the final mixed prediction can further improve the final Prediction accuracy.