A Novel Network Traffic Prediction Method Based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise

Jiayuan Zhong, Yuzhong Chen, Yiqing Shi, Yan Li, Peiqing Chen

Abstract: Network traffic prediction is fundamental for ensuring network reliability, security, and optimal management. Despite the insights from conventional prediction methodologies, the evolving network dynamics call for enhanced prediction techniques. To address this challenge, this paper proposes a novel method using the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN). The input time series is decomposed into multiple Intrinsic Mode Functions (IMFs) and a residual component, providing deep insight into the network traffic dynamics. The WaveNet model is then applied to forecast individual IMFs, and their integration produces the overall network traffic prediction. Notably, our approach surpasses conventional methods in prediction accuracy, empowering network administrators to further optimize network stability and security measures.

External IDs:dblp:conf/iccns/ZhongCSLC23