Embedded image transmission over block fading channels.

摘要

Motivated by the progress in the area of embedded image compression, adaptive modulation, and diversity techniques; we propose the problem of embedded image transmission over block fading channels. The objective of embedded image transmission is to have progressive image reconstruction at the output of wireless link according to the user's individual requirements.; Our transmission strategy includes three subsystems: embedded image compression with wavelet P&barbelow;ixel C&barbelow;lassification A&barbelow;nd S&barbelow;orting to provide the embedding property; forward transmission with adaptive modulation or time diversity to improve the link spectral efficiency, and enhanced feedback retransmission to guarantee the transmission reliability with limited delay.; Throughout this thesis, we emphasize the essential role of known information for embedded image compression; forward transmission, and feedback retransmission. The concepts of classification and sorting are proposed to better understand the previous and current works. Dualities in source coding and channel coding, including models, known information, and classification, accompany the descriptions of system design and performance analysis.; There are two main contributions in this thesis. One is a new embedded image compression algorithm based on wavelet pixel classification and sorting; which is able to provide quality or spatial scalability with flexible complexity. The other is adaptive modulation over correlated block fading channels based on partial channel state information at transmitter, where the prediction loss due to delayed feedback is sensitive to channel fading; channel correlation, and target reliability of forward transmission.