Sixth-Order Statistics-Based Non-Data-Aided SNR Estimation

Abstract

Signal-to-noise ratio (SNR) estimation is an important task in many digital communication systems. With nonconstant modulus constellations, the performance of the classical second- and fourth-order moments estimate is known to degrade with increasing SNR. A new non-data-aided estimate is proposed, which makes use of the sixth-order moment of the received data, and which can be tuned for a particular constellation in order to extend the usable range of SNR values. The advantage of the new method is especially significant for constellations with two different amplitude levels, e.g. 16-amplitude-and-phase-shift keying (16-APSK)