Low SNR Capacity of Fading Channels with Peak and Average Power Constraints

TL;DR

This paper analyzes the capacity of fading channels with peak and average power constraints at low SNR, deriving bounds and exact scaling laws, and extends capacity-achieving signaling results to these constraints.

Contribution

It provides new bounds and exact capacity scaling laws for low SNR fading channels under peak and average power constraints, and extends capacity-achieving signaling results.

Findings

Derived a new upper bound on capacity per unit time.

Established the exact low SNR capacity scaling law.

Showed that non-bursty signaling achieves capacity under constraints.

Abstract

Flat-fading channels that are correlated in time are considered under peak and average power constraints. For discrete-time channels, a new upper bound on the capacity per unit time is derived. A low SNR analysis of a full-scattering vector channel is used to derive a complimentary lower bound. Together, these bounds allow us to identify the exact scaling of channel capacity for a fixed peak to average ratio, as the average power converges to zero. The upper bound is also asymptotically tight as the average power converges to zero for a fixed peak power. For a continuous time infinite bandwidth channel, Viterbi identified the capacity for M-FSK modulation. Recently, Zhang and Laneman showed that the capacity can be achieved with non-bursty signaling (QPSK). An additional contribution of this paper is to obtain similar results under peak and average power constraints.