Queue management algorithm for Hurst-weighted satellite networks based on traffic prediction

Combining the self-similarity characteristics of satellite network service traffic on queuing performance, this paper first develops a prediction model with optimised triple exponential smoothing. The model is based on a dynamic triple exponential smoothing model for network traffic prediction, and the smoothing coefficients of the model are optimised by a differential evolutionary algorithm. Further to accommodate the burstiness and self-similarity of network traffic, a Hurst-weighted Adaptive Random Early Detection (ARED) queue management algorithm based on traffic prediction is proposed. Simulation results show that in the case of bursty service flow data, the Hurst-weighted ARED algorithm based on traffic prediction can effectively reduce packet loss rate, improve throughput, control self-similarity in satellite networks network congestion caused by traffic, and shows better overall network performance.