Description
With the rapid growth of multimedia services, increasing demands on video transmissions have driven numerous researchers to develop error resilient techniques. The perceptual video quality is influenced by not only compression artifacts, but also the channel errors. Hence a desired video codec has to accommodate the contradictory requirements which are coding efficiency, robustness to data loss, as well as other limitations such as memory, bandwidth and complexity. In my thesis, I have addressed a priority and packetization scheme to transport H.264 video encoded packets over wireless networks. The objective of this research is to efficiently exploit the existing error resiliency features in H.264, like Data Partition, Flexible Macroblock Ordering, Constrained Intra Prediction and slice structure in order to come up with a priority scheme to transport packets based on their importance in decoding a video and also in playback. The novel scheme gives protection to video in terms of importance of frame location, partition preferences, slice grouping benefits and also the advantage of using small packets for more important frames thereby reducing the overhead of using small packets for all frames which in turn affects the compression efficiency. A thorough analysis was done on different packet sizes to find the optimum packet size based on the network behavior. Study of non uniform packet sizes was also carried out which proves to be very effective in a bandwidth constrained scenario as the bit rate overhead in this case is less. Also the analysis on fragmentation at Application layer helps us to intelligently transport packets without having to lose any important fragment during MAC layer fragmentation.
