The scenario I consider is a number of transmitters and receivers that
communicate with eachother. Therefore, the free space model does work,

Using the freespace model will, in most cases, give an very optimistic
result; unless there is a true optical path between both antennas and they
are both very well clear of the ground and other obstructions.


but each non-intended receiver receives the interfering power. However,
in my case transmitters will receive requests from receivers that can
be in any direction and thus I believe omnidirectional antennas may be
more appropriate. Omnidirectional antennas allow us to receive and
transmit in all directions equally. Therefore, there isn't a need for
directivity and the use of directional antennas. Are 25 dB
omnidirectional antennas realistic for 802.16a?

It sounds like omni directional is what you want, and 25dBi gain is
unrealistic, stick with 6 to 8 dBi max for a very good omni.

Regards
Jeff