Abstract
We consider the design of a mutually refining DPCM (MR-DPCM) system and its applications to frame erasure channels. MR-DPCM is a system that uses diversity to overcome channel impairments. In this thesis the design for a two channel system is addressed. Two descriptions of the source sequence is sent to the receiver over the two available channels. If both channels are working a high quality reconstruction of the source sequence is obtained. In case one of the channels is down a lower quality description is obtained. The performance of the MR-DPCM system is analyzed theoretically, and verified by simulation using Gauss-Markov sources. A sub-optimum MR-DPCM scheme is also analyzed, and conditions necessary for mutual refinement are derived. The MR-DPCM system is modified to code speech, and speech coders operating at 24 kbps and 32 kbps are constructed. The performance of MR-DPCM speech coders is compared to the performance of a single channel ADPCM system. The MR-DPCM based speech coders are used for the transmission of speech frames over frame erasure channels. For simple frame erasures we show that the MR-DPCM system is more robust than the single channel ADPCM system. We then apply the MR-DPCM system to transmit speech over a mobile radio channel model where frame erasures occur in bursts. A Reed-Solomon channel code is applied to decrease the operating channel SNR significantly. The performance of the Reed-Solomon coded MR-DPCM system is then obtained for different code rates and vehicle speeds.
Sutharsan, Thirunathan (1996). Mutually refining DPCM system with applications in frame erasure channels. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -1996 -THESIS -S88.