On near-far gain in multiuser diversity systems

Multiuser diversity scheduling in a single-cell system is studied. The scheduler chooses a constant size subset of users with best short term fading gains and coordinates transmissions using superposition coding. Rate allocation is performed according to userspsila channel states assuming equal power allocation over users. Asymptotic analysis of system capacity is applied to study system behavior at low and high spectral efficiencies. It is found that if users have non-symmetric channel distributions, i.e. path loss is present, with large enough user population it becomes beneficial in the sum rate sense to schedule more than one user at a time.