Enhancing the robustness of distributed compression using ideas from channel coding

We investigate the self error resilience of practical distributed source coding (DSC) approaches at the presence of residual transmission errors. It is firstly shown that the existing asymmetric syndrome-former inverse-syndrome-former framework (ASIF), although simple, general and optimal for the noiseless-channel case, is not sustainable to transmission errors. The vulnerability stems from the underlying binning approach. Using ideas from channel coding, we illuminate a subtle relation between the binning approach and the parity approach, and demonstrate, through the examples of convolutional codes, how the former can be transformed to the latter for stronger self error resilience. Simulation results confirm that the new scheme is much more robust and less error-sensitive than the existing ASIF scheme