System-Level Impact of Multi-User Diversity in SISO and MIMO-based Cellular Systems

We quantify cell-wide mean throughputs of single-input-single-output (SISO) and multiple-inputmultiple-output (MIMO)-based cellular systems which employ multi-user diversity (MuD). Our study considers several practical and useful system-level design dimensions, including: number of transmit/receive antennas; antenna-pattern (omni-directional or sectorized); degree of error-protection (Shannon coding, no coding or intermediate coding strategies); allowable constellation size; Rician κ-factor; number of users and schedluling algorithm (Greedy (i.e. MAX C/I), Proportional Fair, or Equal Grade of Service) in singlecell (noise-limited) and multi-cell (co-channel-interferencelimited) environments. We also provide a comparison between single-user systems having excess receive antennas and multi-user diversity systems with no excess receive antennas. Both strategies improve signal quality. Since economic costs of RF chains, mobile size and form factor limit the number of antennas a mobile receiver can have, multi-user diversity can be a more practical option. We observe that MuD with only a few scheduled users leads to comparable throughputs as receivers with excess receive antennas. By quantifying the average throughput gains that accrue from using multi-user SISO and MIMO-based cellular systems, this study serves the needs of operators to assess these promising technologies.