PSO-VMD-Informer: A New Edge Load Prediction Method

In edge computing systems, efficient load prediction plays an important role in optimizing resource allocation and efficient utilization of edge service providers, as well as maintaining stable service quality. Due to the high dynamics of edge computing environments, the load data characteristics of different edge servers are significantly different, which seriously affects the accuracy and generalization ability of load prediction. This paper proposes a new edge load prediction method, called PSO-VMD-Informer. This method first uses Variational Mode Decomposition (VMD) to decompose the edge load data sequence into stable intrinsic mode functions to effectively extract nonlinear features; then the Particle Swarm Optimization (PSO) algorithm is used to optimize the intrinsic mode function and penalty factor of VMD to improve the subsequence extraction quality and enhance the generalization ability of the model; finally, the Informer model is used to predict the edge load using the encoder-decoder architecture and self-attention mechanism. A large number of experimental results based on real edge load datasets show that compared with other mainstream methods, the PSO-VMD-Informer model shows the best performance on different edge server data and different prediction lengths, especially significantly outperforming other models in the MAE indicator. Compared with the other three mainstream methods, the proposed method reduces MAE, RMSE, and MAPE by 41.61%, 27.43%, and 11.21% respectively.