Flow-Level Capacity of Fractionally Loaded OFDMA Networks with Proportional Fair Scheduling

We define and evaluate the downlink flow-level capacity of a fractionally loaded OFDMA network utilizing round robin (RR) or proportional fair (PF) scheduling. The flow-level capacity is the maximum traffic intensity that can be supported by a sector while satisfying a minimum flow throughput requirement for bursty, elastic traffic. We obtain a simple, insightful capacity expression for RR scheduling; for PF scheduling, we exploit a fast convergence property of the multi-user diversity gain to implement a computationally efficient, iterative algorithm to solve for the capacity. A hybrid simulation/analysis approach is applied, where a detailed rate distribution obtained via simulation is used as input to a generalized processor sharing queue model. The analytical expressions and numerical results indicate that enhancement of edge throughputs can significantly improve the flow-level capacity.