Video header compression for wireless communications

• Main Author: Ahmad, Zaheer
• Published: University of Surrey 2006
• Subjects: 621.38833
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.436076

The delivery of high quality video to wireless users depends on achieving high compression efficiency and high robustness to wireless channel errors. Research on these topics has led to the introduction of a number of video codec standards. However, most of these standards incorporate redundant syntactical information that renders the video more susceptible to channel errors, and reduces the compression efficiency. This thesis presents a new approach to video compression that removes most of the problems associated with the excess syntax, a header compressor and decompressor is placed adjacent to encoder and decoder respectively. The compressor removes the excess header redundancy and decompressor regenerates the original header using already stored reference. The thesis investigates the syntactical header information of video coding standards MPEG-4 and H.264. The analysis shows that the overheads in small video packet may contribute up to 50% of the texture data. The video packet header fields are classified as static and dynamic. Based on the header analysis, a comprehensive header compression scheme is designed for MPEG-4 and H.264 video packets. For a practical scenario, simulations of video packets are carried out including the compressed IP/UDP/RTP overheads. The ROHC (RObust Header Compression) standard compresses the IP/UDP/RTP headers. In this thesis, ROHC parameters have been optimised for transmission over 3GPP simulated downlink channel. Furthermore, an improvement in the ROHC U-mode has been proposed to reduce the effects of unnecessary packet loss due to false context damage. Results show better video quality and lower packet loss rates with proposed scheme. The efficiency of the proposed video header compression scheme is evaluated with different combinations of encoding parameters. Experiments show the using the proposed video header compression scheme, up to 95% of header redundancy may be removed. Extensive simulations illustrate improvements in video quality due to the proposed header compression scheme under various channel conditions. The video quality is further enhanced by using a header-compression based unequal error protection scheme. The bits saved due to header compression can be utilised adaptively to protect the critical data at a fixed transmission rate. The results show significant improvement in objective and subjective video quality.