Fairness and throughput analysis for generalized proportional fair frequency scheduling in OFDMA

In this paper, a generalized proportional fair (GPF) scheduling algorithm is presented, which allows tweaking the trade-off between fairness and throughput performance for best effort traffic in a cellular downlink scenario. The GPF is extended to frequency scheduling in an OFDMA system by performing dynamic channel allocation on a subband basis including link adaptation by adaptive modulation and coding. In this way, multiuser diversity can be utilized in time domain - as for CDMA - and in frequency domain. Compared to a system without frequency scheduling, this increases the system throughput and yields an improved fairness with respect to the allocated resources and with respect to the achieved data-rate per user. OFDMA system level simulations are carried out in order to analyze short/long-term fairness, multiuser diversity gain and system throughput of various GPF configurations with and without applying frequency scheduling.