Non-data-aided distorted constellation estimation and demodulation for mmWave communications

The signal constellation of 60 GHz millimeter wave (mmWave) communications is distorted by power amplifier (PA) nonlinearity seriously and the bit error rate (BER) performance declines, due to the huge bandwidth and high equivalent isotropic radiated power (EIRP), especially for quadrature amplitude modulation (QAM) signals. This paper aims at non-data-aided (NDA) distorted constellation (DC) estimation and demodulation to deal with PA nonlinearity. An estimation algorithm of expectation-maximization (EM) is proposed and the corresponding Cramer-Rao lower bounds (CRLBs) are calculated, then the estimated DC is used as reference for demodulation. Simulation results indicate that the means of the estimation results are unbiased and the mean square errors (MSEs) reach the CRLBs under easy requirements on signal-to-noise ratio (SNR) and output power back-off (OBO). Additionally, with PA nonlinearity, the BER performance of the presented DC demodulation is close to that of the ideal PA case, and much superior to that of the traditional standard constellation (SC) demodulation.