Cyclic-delay diversity (CDD)-aided frequencyselective scheduling has been known as an effective technique for increasing the system capacity of orthogonal frequency-division multiple-access (OFDMA) systems. By increasing the channels' frequency selectivity with different cyclic delays applied on transmit antennas, the merit of multiuser diversity can be more effectively exploited in the system. In this paper, we aim to optimize the design of the CDD-aided frequency-selective scheduling for OFDMA downlink systems. Optimization is done from two aspects: cyclic-delay search and scheduling-information feedback. In the optimization of cyclic-delay search, three new methods are proposed, including antenna-wise sequential search for frequency-nonselective (AWSS-FNS) channels, AWSS for frequency-selective (AWSS-FS) channels, and a genetic-algorithmbased search (GAS). AWSS-FNS achieves the optimal performance if a channel is frequency nonselective over a subchannel, whereas AWSS-FS and GAS provide good tradeoffs between system performance and complexity for channels that are frequency selective over a subchannel. In the optimization of scheduling information feedback, a new channel-dependent feedback (CDF) method is proposed, where the feedback resource is allocated to users according to their frequency-selectivity rates and a principle of proportional fairness. The optimized scheduler is shown to provide significant improvement in both system sum rate and user fairness over previous methods under fixed feedback overhead.
- Cyclic-delay diversity (CDD)
- Frequencyselective scheduling
- Orthogonal frequency-division multiple access (OFDMA)