Hey guys,
I'm in charge of research in Corona so I took a thorough look at the paper. I came to the conclusion that it is not very useful for Corona at this point, for the following reasons.
First there is the overhead: for some rough refractive surfaces, the render time can get almost twice longer with the method in the paper - definitely not something we would want to see in Corona.
Second, what does the paper really has to offer in exchange for this overhead? In fact, nothing that Corona wouldn't already have.
What they use is the usual microfacet BRDF (such as GGX) but they can ensure that energy is preserved as the roughness/glossiness changes. In other words, the material does not get darker/brighter as you tweak glossiness.
But we already have a different solution for exactly the same problem in Corona which works just fine. Try to play around with CoronaMtl and you should be able to see that there is actually no darkening as you change the reflection glossiness.
Current version of Corona still has some issues with energy preservation of rough *refraction*, but we already have implemented a solution that works very well and we will release it in a daily build soon after 1.4.
So there's probably not much to get out of this paper for Corona.
Cheers,
Jaroslav