A Campus Network Traffic Prediction Method Based on the ARIMA-CNN-LSTM

A secure, stable, and efficient campus network environment is the foundation of information-based teaching and research in schools. An efficient and accurate campus network traffic prediction model will help in timely forecasting of network trends and identifying potential issues. In this paper, we construct a combined model to achieve accurate predictions based on the real historical traffic data of the campus network. Firstly, we use the ARIMA model to fit the linear part of the data and extract the residuals to obtain their temporal dependencies and trends. Subsequently, the CNN-LSTM model is employed to correct the residuals, aiming to rectify errors in future ARIMA model predictions and thereby achieve more accurate network traffic prediction. The proposed model was tested in a real-world scenario using school data, achieving an R2 value of 0.9884. Additionally, we compared our model with seven widely used models under the same environment. The experimental results demonstrate that the model proposed in this paper outperforms the comparison models in terms of Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Coefficient of Determination (R2).