Diversity-multiplexing-tradeoff-optimal 2-user scheduling in an M-user Gaussian multiple-access channel

The problem of user scheduling in a Gaussian i.i.d. Rayleigh faded multiple-access channel (MAC) is considered, whereby the receiver uses perfect knowledge of the channels of all the transmitters to select and schedule two-out-of-M users. The transmitters are assumed to have one antenna each and the two selected users are allocated equal rates. The receiver is equipped with Nr antennas. For user selection that is based on minimizing outage probability (say via an exhaustive search over (M over 2)possibilities), the diversity-multiplexing tradeoff (DMT) of the corresponding outage probability is computed. Moreover, a novel user-selection algorithmis proposed that achieves the same DMT as does the minimum outage probability method, but with only a O(M) complexity instead of O(M2) required for an exhaustive search to minimize outage probability. The DMT optimality of the proposed algorithm is also proved for 2-user scheduling in the dual M-user Gaussian broadcast channel assuming availability of global channel state information. Simulation results are presented which demonstrate that the proposed algorithm achieves an outage probability close to that of the exhaustive search.