Computing the Rate-Distortion Function of Gray-Wyner System

In this paper, the rate-distortion theory of Gray-Wyner lossy source coding system is investigated. An iterative algorithm is proposed to compute rate-distortion function for general successive source. For the case of jointly Gaussian distributed sources, the Lagrangian analysis of scalable source coding in [1] is generalized to the Gray-Wyner instance. Upon the existing single-letter characterization of the rate-distortion region, we compute and determine an analytical expression of the rate-distortion function under quadratic distortion constraints. According to the rate-distortion function, another approach, different from Viswanatha et al. used, is provided to compute Wyner's Common Information. The convergence of proposed iterative algorithm, RD function with different parameters and the projection plane of RD region are also shown via numerical simulations at last.