Heuristic Subcarrier Allocation Algorithms with Multi-Slot Frames in Multi-user OFDM Systems

This work addresses radio resource allocation for downlink (DL) transmissions in a cellular system based on Orthogonal Frequency Division Multiple (OFDM). While scheduling based on multi-user diversity increases throughput, it decreases fairness across the users. Fairness is usually provided by Proportional Fair Scheduling (PFS). However, it was shown that optimal PFS in a Multi-Carrier (MC) system is prohibitively complex. In this paper we consider the practical assumption where there are multiple time slots (OFDM symbols) per scheduling frame, for which the optimal allocation becomes even more complex. Therefore, we derive the upper and lower proportional fairness limits which bound the optimal solution, by means of convex optimization. The results show that those derived proportional fairness limits are in fact very close, which gives an excellent bounding of the optimal solution. We also propose two heuristic algorithms which exploit the fact that several users can be multiplexed on one subcarrier in a time-division manner. We investigate their performance in terms of throughput, proportional fairness metric and latency. The additional overhead due to user multiplexing is taken into account. The results show that the proposed algorithms achieve an excellent tradeoff between throughput and proportional fairness.