Base Station-Side Rate Estimation for Threshold-Based Feedback, and Design Implications in Multi-User OFDM Systems

Abstract

Rate adaptation and scheduling are essential in ensuring that contemporary orthogonal frequency division multiplexing systems achieve high downlink spectral efficiencies. They depend upon reduced feedback schemes to efficiently feedback channel state information from the users to the base station (BS). In the popular threshold-based quantized feedback scheme, a user feeds back to the BS the quantized value of the signal-to-noise ratio for each subchannel. For this scheme, we derive a novel, throughput-optimal discrete rate adaptation (TORA) policy, which enables a system designer to reduce the feedback overhead. We present it in closed form for different multi-antenna diversity modes for the exponentially correlated subchannel gains model. We also develop a computationally simpler suboptimal variant of it. We derive an insightful lower bound for the fading- and user location-averaged throughput gain achieved by TORA over conventional rate adaptation for 1-bit feedback. We present extensive results to benchmark the system-level performance of TORA for different numbers of feedback bits and modulation and coding schemes available at the BS, and various schedulers, quantizers, and multi-antenna modes.