Rate Control Algorithm For Adaptive Bitrate Streaming

In recent years,with the development of multimedia technology and the improvement of the mobile network environment,video media has become one of the most important content production vehicles for the Internet.At the same time,users have increasingly higher quality requirements for online video.Due to the diversity of terminal devices and the heterogeneity of mobile networks,in order to enhance the user's viewing experience,a dynamic adaptive streaming technology that can adaptively adjust video quality according to device status and network bandwidth is being used by more and more video companies..In the dynamic adaptive streaming technology,video is stored in the server in the form of multi-rate chips for the user to select and download according to the network conditions.Since the client player cannot predict-the size of each slice,but the slice download time is estimated from the average bit rate of each stream,the fluctuation of the actual bit rate of the slice relative to the average bit rate will affect the correct decision of the adaptive algorithm..The most widely used x264 encoders in the industrial world are designed based on the continuous stream application scenario.The bit rate control method used does not consider the case where the video is segmented,and lacks the bit rate control at the slice level,resulting in a coded chip rate.Large fluctuations cause the adaptive algorithm to determine the misalignment,and it is easy to generate stalls or bandwidth waste in practical applications.In view of the shortcomings of traditional algorithms,based on the coding framework of x264,this paper proposes a slice-level code rate control method,which consists of two parts:code rate allocation and code rate control.In terms of code rate allocation,according to the reference relationship of each type of frame in inter-frame prediction,a code rate allocation method based on frame types is used,and the prediction of the frame type of the slice is predicted through forward prediction,and at the same time,The concept of key P-frames optimizes bit allocation under different video content by adaptively adjusting the number and distribution of key P-frames.In terms of rate control,the method establishes a linear prediction model based on the SATD value and the quantized coefficient q,and it uses the model to iteratively adjust the quantized coefficients of each row to control the entire frame's encoding size,thereby achieving the purpose of accurately controlling the encoding size of each frame,and Re-encode some of the frames with too low quality using the method of secondary coding check.This method is based on the industrial encoder x264(libx264 v20161024.2245),and a dynamic adaptive streaming media platform is set up for simulation experiments.Through statistics and analysis of experimental data,compared with the original rate control algorithm of x264,this method has better performance in controlling the video slice rate,and the slicing code rate fluctuation is reduced by 63%compared to the original algorithm of x264.At the same time,the objective quality of video is guaranteed.