Maximum Latency Prediction Based on Random Forests and Gradient Boosting Machine for AVB Traffic in TSN

In this letter, we propose a latency prediction method based on random forests (RFs) and gradient boosting machine (GBM) to estimate the maximum latency of audio-video bridging (AVB) traffic in time-sensitive networks. We use a dataset collected from actual networks as the input features for the RFs. Then, we use the prediction results of the RFs as input features for the GBM to complete the model training. The experimental results show that the proposed method, compared to network calculus, performs better in terms of deviations as well as error metrics when the link utilization is 25%.