TDMA networks to CDMA
When carriers convert their TDMA networks to CDMA they'll be tempted to reuse their cell site antennas to cut costs. Bad idea. Although we think of antennas as being technology independent they really aren't in cellular radio. And there's a different way to work with CDMA antennas. Mark van der Hoek (internal link) relates:
"I haven't worked on GSM systems myself. As to antennas, the main thing is to CONTROL YOUR PROPAGATION!!!! The GSM operators of the world are going to find out the hard way that they cannot use the same antennas for W-CDMA that they have used for GSM. You cannot frequency plan around bad sidelobes in CDMA! Be sure that you understand the different behavior of sidelobes when down tilting antennas -- few do! There is a widespread notion that down tilting an antenna causes the sidelobes to increase. This is false.
The sidelobes do NOT increase relative to where they were before. They SEEM to because the main lobe decreases, so the side lobes increase RELATIVE TO THE MAIN LOBE. But in truth, the energy at 90 degrees is NOT traveling any farther than it was.
Also, understand the difference, especially with sidelobes, in electrical vs. mechanical downtilt. And don't be afraid to combine electrical with mechanical tilt.
When mounting antennas on a rooftop, be careful to pay attention to the shadowing (and possible upward reflections) of the rooftop itself. I've seen a lot of antennas pointed into a roof, and engineers wondering why the site doesn't cover like it should!
Know your pattern, and know where you want to serve with a given sector or site, then choose an antenna that only goes where you want it! You can't adjust parameters enough to make up for poor design, especially in CDMA. Do a good design, and default parameters will work well enough. Do a poor design, and you're sunk.
More on antennas and patterns and some neat looking graphics from Mark here (internal link)