Adaptive rate control algorithm to improve the performance of H.264 video codec in varying network conditions

Abstract

This thesis involves proposing an efficient algorithm, which takes into consideration the NALU (Network abstraction Layer Unit) and tries to improve the H.264 codec performance over a network. Along with the bandwidth instabilities in the Internet and other issues like the packet drop cause the video frames to be dropped, which makes the video, appear distorted. The main idea behind the algorithm is to vary the bitrate by varying the Quantization parameter, both of which are stored in the reference table. The reference table is calculated based on the analysis of the sequences of different video clips. By changing the bitrate based on the bandwidth available the codec performs better and produces good results of (Signal to Noise Ratio)SNR. The research is based on the development of the reference table which when referred helps to change the quantization parameter of the codec for coding a Group of Frames in a video by periodically checking the network statistics. At the cost of reduction in the resolution due to the network conditions the video frame drop is avoided which helps improve the video quality at the decoder.. The entire simulations for this research were carried out using the JM Reference Software H.264 Encoder and decoder Version JM 10.2. Packet drop was simulated by dropping the individual NAL units from the video stream.