基于ARIMA-LSTM的容器云资源预测方法

Container cloud resource prediction method based on ARIMA-LSTM

边缘计算充分使用了网络边缘的网络资源、存储资源和计算资源,使任务可以在网络边缘进行处理,充分保证了服务的实时性和高健壮性。目前,边缘计算存在资源管理的高效性以及稳定性问题。因此,提出了一种Autoregressive Integrated Moving Average model(ARIMA)-Long Short-Term Memory(LSTM)混合模型的容器云资源预测方法,解决了单个预测算法不能同时求解容器云资源需求量数据中的线性分量和非线性分量问题。该模型使用自回归综合移动平均模型(ARIMA)来预测资源需求量中的线性分量,并使用长短期记忆模型(LSTM)来预测非线性分量。结合容器云平台监测数据,使用ARIMA-LSTM模型对未来容器云资源需求量进行预测,并与ARIMA模型,LSTM模型进行比较。最后,实验结果表明该混合模型可以有效提高预测的准确性。

Edge computing makes full use of the network resources, storage resources and computing resources at the network edge, so that tasks can be processed at the network edge, and fully ensures the real-time and high robustness of the service. At present, edge computing have some problems of efficiency and stability in resource management. So this paper provides a container cloud resource prediction method based on Autoregressive Integrated Moving Average model(ARIMA)-Long Short-Term Memory(LSTM) hybrid model, which solves the problem that a single prediction algorithm can’t solve the linear and nonlinear components of container cloud resource demand data at the same time. The model uses autoregressive comprehensive moving average model(ARIMA) to predict the linear component of resource demand, and uses long-term and short-term memory model(LSTM) to predict the nonlinear component. Combined with the monitoring data of container cloud platform, ARIMA-LSTM model is used to predict the future demand of container cloud resources, and compared with ARIMA model and LSTM model. Finally, the experimental results show that the hybrid model can effectively improve the accuracy of prediction.