This paper provides a simple approximation to pairwise error probability for memoryless fading channels when there is no channel state information at the decoder. The derived approximation is independent of the fading process and depends only on the mean and variance of the fading distribution. The new approximation is considerably better than an already existing approximation and a bound for low signal to noise ratio values. It can be applied to obtain average union upper bounds for block and turbo-like coding schemes. Plots for pairwise error probability for Rician, Rayleigh and Nakagami-m fading channels along with average union upper bounds for turbo codes with (1,5/7,5/7) generator polynomial are given to illustrate improvement in already published work

Campbell has given a generalized Kraft inequality and related results for decoding with respect to a fidelity criterion. His results are, however, restricted to "row-balanced" distortion matrices for finite-alphabet sources. We generalize these results to any distortion matrix.