Motivated by the applications in the field of mobile and broadcast communications, this research investigates techniques for joint design of source and channel coding schemes. A special emphasis is placed on the design of vector quantizer (VQ) based communication systems.; The design of VQ for operation over binary symmetric channels is first addressed. A novel design method, the noisy channel relaxation (NCR), is presented. It is demonstrated that NCR-based design of channel optimized VQ and index assignment VQ offers substantial performance gains over existing design methods.; The second line of research concerns VQ index transmission over time-varying channels. Here, a powerful technique called transmission energy allocation (TEA) is developed. The basic idea underlying this approach involves varying the level of transmission energy allocated to the different VQ index bits depending on their sensitivity. Efficient solutions are developed for the following problems: (i) joint optimization of TEA and index assignment, (ii) implementation of TEA with low peak-to-average ratio, (iii) design of an adaptive channel optimized VQ system with TEA, and, (iv) application of TEA to CDMA communication. It is demonstrated that the implementation of TEA is practically feasible and promises very substantial performance gains.; The third line of research is concerned with the design of source-channel coding schemes for operation in communication scenarios where the state of the channel is known to the receiver, but not to the transmitter. Typical applications are encountered in broadcast communication and in mobile communication applications without a feedback path from the receiver to the transmitter. An appropriate source-channel coding scheme for these channels consists of a multi-resolution source-coder followed by unequal error protection channel coding. A new method for optimal design of a system based on multi-stage VQ with unequal protection provided via transmission energy allocation, is presented. The scope of the research is then extended to include an in-depth investigation into the design of coded modulation schemes for unequal error protection. Specifically, new analytical and experimental approaches are developed to compare: superposition coded modulation (SCM), and time-division coded modulation (TDCM). A significant result of this study shows that although asymptotically (i.e., when Shannon's capacity is achieved) SCM is expected to outperform TDCM, practically, TDCM is often comparable, or superior, in its performance. Finally, a hybrid modulation method is developed and shown to be almost always superior to TDCM or SCM.
展开▼
