Which distribution can be abbreviated as LD and which PDF is expressed as a formula with sum of erfc() functions?
$$p(o)=\frac{1}{4\ell} e^{\frac{\sigma^2}{2\ell^2}} \left[e^{\frac{o'}{\ell}}\operatorname{erfc}\left(\frac{\ell o'+\sigma^2}{\sqrt{2}\ell\sigma}\right) + e^{-\frac{o'}{\ell}}\operatorname{erfc}\left(\frac{-\ell o'+\sigma^2}{\sqrt{2}\ell\sigma}\right)\right]$$
where
$$o'=o-\mu^G-\mu^L$$
where $\mu^G$ is probably gaussian mean and $\mu^L$ and $\ell$ are entitled as LD's location and scale parameters respectively.
EDIT 1
Probably "D" stands for "distribution", if so then the question is what "L" stands for.
From the references of the paper you are reading, I find that "LD" means Laplacian distribution. To quote Wikipedia: The Laplacian distribution has been used in speech recognition to model priors on DFT coefficients.