Self-adaptive shunting decision method based on control theory and data driving

Abstract

The invention discloses a self-adaptive shunting decision method based on a control theory and data driving, and the method comprises the steps: defining a proper network environment as the state input of an intelligent agent module, and enabling an intelligent agent to make an action decision in a self-adaptive manner through the training and testing process of a machine learning model; distributing a corresponding number of video data to the transmission terminal after the optimal distribution ratio of the multiple paths is decided; meanwhile, calculating the code rate control range based on a dynamic change model of the buffer area, adjusting the video code rate to adapt to the change of occupation of the buffer area, and therefore, the video playing fluency is guaranteed; through the control system, further optimizing the application accuracy of the machine learning algorithm decision model in an actual system; and in a system test stage, transmitting corresponding video data by using an optimal split ratio, calculating a video code rate range, and selecting a proper value by a control system to be matched with the dynamically changing buffer area occupation length.

Description

technical field [0001] The invention relates to the technical field of streaming media playback, and mainly relates to a control theory-based and data-driven adaptive distribution decision-making method. Background technique [0002] According to the report of Cisco Technology, the wireless video traffic is growing rapidly and will dominate the wireless data traffic. These include Internet video, IP VoD, video files exchanged via file sharing, video streaming games, and video conferencing, among others. Global IP video traffic is expected to account for 82% of traffic by 2022. The huge growth of video traffic is mainly due to the rapid development of wireless communication technology (such as 4G, 5G technology), the popularity of mobile devices (such as smart phones, tablet computers, etc.) and the growth of Internet business demand (such as online video conferencing). Therefore, it is of great significance to provide users with smooth video playback and improve user exper...

AI Technical Summary

Problems solved by technology

[0009] Purpose of the invention: In order to solve the fluency of video playback under heterogeneous wireless networks and improve user QoE, the present invention provides an adaptive distribution decision method based on cybernetics and data-driven, based on deep reinforcement learning algorithm as the theoretical basis, And use control theory to optimize, solve the problem of video distribution decision-making in heterogeneous wireless networks, realize the smoothness of multi-terminal cooperative video transmission, and ensure high user QoE

Images