Re: A side-effect
If you are talking about LEO constellations and using an omni antenna, basic physics starts to work against you. My company is working on a proof-of-concept mission based on a single (for testing purposes) 3U CubeSat using ground terminals with omni-antennas. The current system (it is still in the process of being designed) can hit a maximum data rate of about 10 kbps; however if the satellite is near the horizon then the data rate drops to a few hundred bps. Input power to the antenna is 20W (for both the ground station and the spacecraft), and the minimum safe distance from the antenna (i.e. how close you can get until your nadgers start to cook) is 40 cm.
10 kbps is hardly broadband, so what will it require to achieve, lets say, 10 Mbps. Well we would need a physically larger antenna on both the satellite and the ground station - at adds quite a lot of cost at either end. More importantly is that, in order to keep the required Eb/N0 (energy per bit divided by the noise energy) you have to increase the input power into the antenna to abut 20 kW (assuming your RF system and antenna have the same efficiencies). Minimum safe distance to the ground antenna also increases to something in the region of 3 metres.
The input power is OK for a ground terminal (but it will get expensive to run in the long term and is definitely not environmentally friendly) but is damn near impossible to achieve on a small satellite unless you start to fit the sort of massive solar arrays they use on the big GEO birds. Did I mention that these also cost significant $$$$.
Another thing to think of is the lifetime of your satellite in LEO. GEO birds are often designed for lifetimes of 15 to 20 years, are are generally only limited by consumables (e.g. thruster fuel). For LEO satellites, atmospheric drag is the biggest limiting factor; LEO missions often talk about having to completely replace their constellations every 2 or 3 years (maximum). That means lots of rocket launches (also not environmentally friendly).