Distributed coding via folding functions

Summary form only given. We propose an innovative scheme for distributed coding of two correlated variables X and Y. After analyzing the proposed scheme in some detail, we compare it with joint coding via Karhunen-Loeve transform followed by uniform quantization and entropy coding. We carried out some experiments with X = Y + N, where X, Y and W are all uniform or all Gaussian, with W independent from Y. Experimental results show that the proposed distributed coding scheme has very close or better performance than the joint coding procedure in the considered cases