Load Forecasting of Distributed System Host Based on Linear Time Series Model

Abstract

With the development of communication network, wireless data transmission technology is becoming more and more important in people's daily life. As a new structure, distributed system has strong applicability and adaptability. This paper mainly studies the load forecasting of urban cloud computing platform based on time series model. Firstly, the load forecasting method and principle are introduced. Secondly, according to the actual situation, the appropriate range of random disturbance parameters is determined and a stable and reliable index system is established. Finally, through simulation experiments, the load rate curves of different time resolution and different types of distributed systems under uniform load distribution are verified, and the stability and accuracy are improved by adjusting the load degree. From the test results, it can be seen that the host load prediction results for linear time series have small errors, which indicates that the prediction model has high accuracy.