A Stream Classification Method Using Mean Opinion Score Feature of Video Services

Abstract

The invention discloses a flow classification method using the average opinion score feature of video services, including three steps of data set preprocessing and feature extraction, calculation of PDF mean feature of three video service MOS and SVM classification. Data set preprocessing and feature extraction first use network packet analysis software to capture the video data stream in the required network, save the captured original data stream in a standard text format, perform a large number of traffic statistical feature calculations, and select the downstream bytes Rate is the feature with the largest information gain rate. SVM classification defines the mean feature of each video stream combined with the downlink byte rate as the optimal feature subset, supervises discretization of the feature subset, and uses the SVM-based ten-fold cross-validation method for stream identification and classification. The present invention can guarantee the experience quality of watching video by the client, analyze the distribution of MOS values ​​of three kinds of video services, and cooperate with effective feature combination to identify and classify the three kinds of videos, and can obtain better classification effect.

Description

technical field [0001] The invention belongs to the technical field of stream identification and classification, and in particular relates to a stream classification method using the average opinion score feature of video services. Background technique [0002] In recent years, with the rapid development of wireless communication technologies such as Wifi, LTE, 4G and WiMAX, the traffic of streaming video services has increased dramatically. increase. According to the statistical report on China's Internet development, as of June 2016, the number of online video users in my country reached 514 million, an increase of 10 million from the end of 2015. The number of people watching Internet videos accounts for 72.4% of all Internet users, and this value has been growing. At the same time, multimedia video services such as Web video have high requirements on transmission bandwidth, delay, and jitter, which will inevitably bring severe challenges to the transmission of video se...

AI Technical Summary

Problems solved by technology

The advantage of using this method is that the model is simple and the calculation process is relatively efficient, but the disadvantage of this method is that only the peak value of the downlink rate is considered, the selected feature is relatively single, and the recognition rate of untrained samples is low

Images