Reduced Feedback Schemes in 4g/5g Ofdm Systems: Modeling, Performance Analysis and Redesign
Reduced feedback schemes are a crucial component of orthogonal frequency division multiplexing-based 4G and 5G cellular systems that use downlink scheduling, adaptive modulation and coding, and multiple-input-multiple-output (MIMO). In these systems, the system bandwidth is divided into several subbands. The reduced feedback schemes ensure that the overhead of feeding back the subband-level channel state information from the users to the base station (BS) does not overwhelm the limited-bandwidth uplink feedback channel. Differential feedback is a key component of the reduced feedback schemes specified in both 4G and 5G standards. In it, the user feeds back a 4-bit wideband channel quality indicator (CQI), which indicates the rate that the user can decode if the BS were to transmit to it over the entire system bandwidth, and a 2-bit differential CQI for each subband relative to the wideband CQI. We present a novel modeling and a comprehensive analysis of the throughput of the differential feedback scheme. It incorporates several key and unique aspects of the 4G and 5G standards and encompasses different multi-antenna modes and schedulers. The analysis brings out several insights about how the throughput increases as the correlation between subbands increases and how the feedback scheme reduces the overhead while only marginally reducing the throughput. We then develop a BS-side estimation technique for the widely studied and simpler threshold-based quantized feedback scheme, in which a user feeds back a quantized value of the signal-to-noise ratio for each subband. For it, we derive a novel, throughput-optimal rate adaptation rule in closed-form for different multi-antenna modes. We also develop a computationally simpler near-optimal variant of it and derive an insightful lower bound that characterizes the average throughput gain achieved by the proposed scheme over conventional rate adaptation. Using the insights obtained from the throughput analysis and the BS-side estimation technique for threshold-based quantized feedback, we then develop BS-side estimation for differential feedback. We do so for a more general variant of it in which the differential CQI overhead can be different for different subbands. This provides a new flexibility to the BS to control the feedback overhead and leads to a new throughputoptimal rate adaptation rule for it. It brings out the sub-optimality of the conventional method used in the literature. Our approach incorporates small-scale fading, largescale shadowing, path-loss, and different MIMO modes in single-user and multi-user deployment scenarios. It achieves nearly the same throughput as the scheme currently employed in 4G and 5G, but with much less feedback overhead.