A Non-data-aided SNR Estimation Method for Square-QAM System

Envelop-based (EVB) SNR estimators are robust to carrier offset and phase jitter. The performance of such estimators however, degrades with increasing SNR for nonconstant modulus constellations. In this work, we consider the SNR estimation as an optimum scale-invariant rank discrimination testing problem and propose an EVB algorithm that does not fail at high SNR. The performance of this algorithm is evaluated for 16 and 64 square-QAM and compared with popular EVB algorithms, i.e., M_2M_4 and M_6. Under the assumption of no phase/frequency offset errors, the simulation results show comparable performance at low SNR with existing algorithms and a significant improvement at higher SNR.