Blind SNR Estimation for Arbitrary Linear Modulated Signals

Three novel blind SNR estimators for arbitrary linear modulated signals are proposed in this paper. The first fixed dimension ED (FDED) estimator is constructed by building the correlation matrix of the oversampled signal with a specific dimension equals to the oversampling rate, which shows less complexity and better performance than the traditional one. Another low complexity (LC) estimator can be built if the synchronization can be achieved, which will avoid the step of eigenvalue decomposition. The third reusing data fixed dimension PASTD (RDFD-PASTD) algorithm based SNR estimator is proposed to solve the drawbacks of the traditional PASTD estimators. Simulation results show that the FDED estimator outperforms the traditional ED estimator especially at low SNR, while the LC estimator only has an evidently higher bias at high SNR The RDFD-PASTD estimator exhibits good performance under the limited data.