Design of Optimal Entropy-constrained Scalar Quantizer for Sequential Coding of Correlated Sources

This work addresses the design of a sequential code for correlated sources using entropy-constrained scalar quantization at each encoder. We consider discrete sources and propose a globally optimal algorithm to minimize a weighted sum of distortions and rates. Our algorithm is based on solving the minimum weight path problem in a series of appropriately constructed weighted directed acyclic graphs. Its asymptotical time complexity is $O(N_{1}^{2}N_{2}^{2})$, where $N_{1}$ and $N_{2}$ denote the alphabet sizes of the two sources, respectively.