Capacity of Sparse Wideband Channels with Partial Channel Feedback

TL;DR

This paper investigates how sparse multipath structures in wideband channels can be exploited with limited feedback to achieve capacity gains comparable to perfect CSI scenarios, even under realistic power constraints.

Contribution

It demonstrates that channel sparsity and coherence can be leveraged with limited feedback and power constraints to attain benchmark capacity gains in wideband channels.

Findings

Sparse multipath channels enable capacity gains with limited feedback.

Channel coherence scaling is crucial for achieving capacity benefits.

Training schemes can exploit sparsity for improved performance.

Abstract

This paper studies the ergodic capacity of wideband multipath channels with limited feedback. Our work builds on recent results that have established the possibility of significant capacity gains in the wideband/low-SNR regime when there is perfect channel state information (CSI) at the transmitter. Furthermore, the perfect CSI benchmark gain can be obtained with the feedback of just one bit per channel coefficient. However, the input signals used in these methods are peaky, that is, they have a large peak-to-average power ratios. Signal peakiness is related to channel coherence and many recent measurement campaigns show that, in contrast to previous assumptions, wideband channels exhibit a sparse multipath structure that naturally leads to coherence in time and frequency. In this work, we first show that even an instantaneous power constraint is sufficient to achieve the benchmark gain when perfect CSI is available at the receiver. In the more realistic non-coherent setting, we study the performance of a training-based signaling scheme. We show that multipath sparsity can be leveraged to achieve the benchmark gain under both average as well as instantaneous power constraints as long as the channel coherence scales at a sufficiently fast rate with signal space dimensions. We also present rules of thumb on choosing signaling parameters as a function of the channel parameters so that the full benefits of sparsity can be realized.