Noisy Graph Signal Vector Quantization with Codebook Allocation

We study the problem of vector quantization and codebook allocation for noisy graph signals. To solve the communication resource constraint, this paper considers vector quantization for noisy graph signals. Kernel regression over graphs is utilized to minimize distortion of the noisy graph signal vector quantization. For further distortion reduction and resource limitation settlement, we also propose a smoothness-based codebook allocation algorithm to allocate the codebook size to each vector quantizer. Simulation results show the effectiveness of the proposed vector quantization for noisy graph signal and the advantage of the proposed codebook allocation algorithm over the existing algorithms.