Channel-Optimized Strategic Quantizer Design via Dynamic Programming

We consider the design problem of a strategic quantizer over a noisy channel, extending the classical work on channel-optimized quantization to strategic settings where the encoder and the decoder have misaligned objectives. Building on our recent work on strategic quantization over noiseless channels, we employ a random channel index assignment mapping, as done in prior work on classical channel-optimized quantizer design literature, combined with a dynamic programming approach to optimize quantization boundaries. Our analysis and numerical results demonstrate several interesting aspects of channel-optimized strategic quantization which do not appear in its classical (nonstrategic) counterpart. The codes are available at: https://tinyurl.com/ssp2023dpnoise.