Title :
On channel-constrained vector quantization and index assignment for discrete memoryless channels
Author :
Skoglund, Mikael
Author_Institution :
Dept. of Signals, Sensors & Syst., R. Inst. of Technol., Stockholm, Sweden
fDate :
11/1/1999 12:00:00 AM
Abstract :
This correspondence introduces an approach to the design of channel robust vector quantizers. Design criteria are derived based on a new expression for the channel distortion as a function of the code vectors of the quantizer. The introduced framework for design and analysis holds for arbitrary discrete memoryless channels. Simulations demonstrate good performance. The correspondence also presents new results regarding the channel distortion as a function of the index assignment given to the code vectors of a vector quantizer
Keywords :
combined source-channel coding; memoryless systems; source coding; vector quantisation; arbitrary channels; channel distortion; channel robust source coding; channel robust vector quantizers; channel-constrained vector quantization; code vectors; design criteria; discrete memoryless channels; index assignment; Algorithm design and analysis; Decoding; Discrete transforms; Error correction; Error correction codes; Memoryless systems; Performance evaluation; Robustness; Source coding; Vector quantization;
Journal_Title :
Information Theory, IEEE Transactions on