Project Leader, Supervised by Prof. Shujuan Wang
Updates:
- Added follow-up study report (2020.06)
- Added follow-up study paper (2020.09)
Features:
Built an accurate mathematical model of the relation between network capability and video experience for network planning and assessment, successfully solving the problem proposed by Huawei Technologies
Constructed a nonlinear function based on both TCP mechanism and a dataset with 89,266 samples
Introduced unsupervised anomaly detection algorithm to remove noise data for regression analysis
Proposed and implemented a probabilistic method to solve the heteroscedasticity of video stalling data, significantly enhancing the prediction accuracy for video stalling ratio
To put our result into use, we released a Wechat App based on our model for network planning and assessment.