Memory efficient adaptation of vector quantizers to time-varying channels

Channel-optimized vector quantization (COVQ) is approximated by the novel channel-adaptive scaled vector quantization (CASVQ). This new method uses a reference codebook that is optimal for one specific channel condition. However, for a bit-error rate being different from the design assumption for the reference codebook, all codevectors are scaled by a common factor, which depends on the channel condition. It is shown by simulations that a performance close to that of COVQ can be achieved in many practically important situations. Without a significant increase in complexity, the new CASVQ-scheme can be adapted to time-varying channels by adjusting the scaling factor to the current bit-error probability. Another advantage is that only one codebook needs to be stored for all error probabilities, while for COVQ either the performance degrades significantly due to channel mismatch, or a large set of codebooks must be available at the encoder and the decoder.