Partial waterfilling with reduced feedback for sum discrete-rate maximization in SU-MIMO systems

Waterfilling enables a dramatic increase in capacity of multiple antenna systems, but requires perfect knowledge at transmitter.When a finite discrete rate is available, this algorithm is no more optimal since the interest lies in maximizing the sum discrete-rate. In this paper, motivated by a feedback reduction, we introduce a transceiver that partially decouples the channel into parallel subchannels. From the analogy between multi-user and single-user channels, we propose a discrete rate iterative waterfilling procedure. It first relies on the classical iterative waterfilling, and then rounds continuous rates to discrete rates with an efficient use of residual power. Simulations show that a substantial enhancement of the discrete bit loading process is provided. We further investigate the trade-off feedback-performance of different waterfilling-based systems. Simulations point out that the proposed scheme outperforms the waterfilling with nearly 60% of feedback reduction in the moderate to high SNR region, while achieving nearly 90% of the sum discrete-rate waterfilling at low SNR.