A network bandwidth prediction method and system

Abstract

The application discloses a network bandwidth prediction method and system, and the method comprises the following steps: collecting historical bandwidth data of a target network, using a deep reinforcement learning algorithm to perform prediction, and obtaining a reinforcement learning bandwidth prediction value at each moment; using a heuristic algorithm to perform prediction, and obtaining a heuristic bandwidth prediction value at each moment; weighting and fusing the reinforcement learning bandwidth prediction value and the heuristic bandwidth prediction value to obtain a fused prediction bandwidth at each moment; performing serialization processing on the fused prediction bandwidths at all moments to obtain a fused prediction bandwidth sequence; using an autoregressive moving average model to perform one-level moving smoothing processing on the fused prediction bandwidth sequence to obtain a one-level smoothed prediction bandwidth sequence; and using a Kalman filtering algorithm to perform two-level dynamic smoothing processing on the fused prediction bandwidth sequence to obtain a two-level smoothed prediction bandwidth sequence. The network bandwidth prediction method and system provided by the application improves the prediction accuracy of network bandwidth.

Images