Just for interest sake, depending on the frequency used and the signal strength, LOS does not have to be an absolute requirement. Diffraction of a signal can even be used to your advantage i.e. over a building or hill/mountain. There are calculations for these. Also, if the volume/type of conductive objects in the way, i.e. leaves, can be estimated, penetration of the signal can be roughly determined. I mentioned earlier the frequency; the higher the frequency, the less penetration. Unfortunately the 5GHz frequency used is highly attenuated by these objects, so clean LOS is ideal. Less so for 2.4GHz. That is why Cell-C go on about their 900MHz frequency use as having more penetration than their competition. Software such as Radiomobile can partly factor in the effects of these types of interferences within the Fresnel pipe, but nothing beats a real signal test. If the signal is fairly strong, say within no less than about 15dBm of the receiver sensitivity, and is stable, that should be good enough for data. We have numerous private radio-links at 5GHz set up for our company that do not have clear LOS, but have been running absolutely stable for many years. For commercial use I can imagine the tolerances used being more strictly set.