Rate Maximization for Downlink OFDMA With Proportional Fairness

In this paper, we study the sum throughput maximization with access proportional fairness (APF) for downlink orthogonal frequency-division multiple access (OFDMA) channels. We propose selective multiuser diversity (SMuD) schemes with normalized signal-to-noise ratio (n-SNR)-based ranking for user selection at each carrier to achieve long-term fairness. We also propose a modified absolute signal-to-noise ratio (a-SNR)-ranking-based SMuD scheme, which provides improved performance over the original a-SNR SMuD scheme. The total transmit power is assigned to the allocated carriers using either equal power allocation (EPA) or water filling. Closed-form throughput and fairness expressions for both the n-SNR and a-SNR SMuD schemes with EPA over Rayleigh channels are derived, which are accurate for even highly correlated frequency channels. Numerical results show that the long-term fairness gives a substantially higher rate than the more stringent short- term fairness. The n-SNR scheme brings perfect access fairness; the sum rate is slightly degraded compared to the a-SNR scheme without fairness, and it is not very sensitive to the imperfect estimation of the average channel signal-to-noise ratio. These results put new insight into the achievable downlink OFDMA performance with proportional fairness and the effects of various system and channel parameters.