Design of Data-Aided SNR Estimator Robust to Frequency Offset for MPSK Signals

Abstract

Data-aided (DA) signal-to-noise ratio (SNR) estimation is required especially at low SNR. The conventional maximum likelihood (ML) DA SNR estimator requires perfect carrier phase estimation and frequency recovery. In this paper, we propose a novel carrier frequency robust DA SNR estimator with its improved variant using autocorrelation of received MPSK symbols. Computer simulations are used to examine their performance in terms of mean estimation value (MEV) and normalized mean square error (NMSE). For the example system in simulations, the MEV of proposed estimator is accurate enough with normalized frequency error on the order of symbol rate. However, its NMSE can not reach DA normalized Cramer-Rao bound (NCRB) even with large observatory length, whereas its NMSE may perform a little worse at high SNR for short pilot symbols. On the other hand, fortunately the its improved variant can reach NCRB with enough pilot symbols. What's more, the proposed DA SNR estimators can operate under large frequency errors or before the frequency recovery unit with baud rate. The implementation complexity is also analyzed.Index Terms-SNR estimator; data-aided; carrier frequency robust; MPSK; DVB-S2.

Performance analysis of EM-based iterative SNR estimation for coded signals

“…For simplicity, BPSK is considered as the modulation scheme in the following, so k U in (4) always equals 1, and k K in (3) is replaced by ( 1) ( 1) (6) In this section, we consider the CA SNR estimation mode described in (2) as a unified interpretation of the EM-based SNR Estimation, in other words, the conventional NDA mode is considered as a particular approach within the framework, only differing in the APP, or the corresponding k K .…”

“…To sum up, the unified interpretation of EM-based NCA and CA SNR estimation algorithms can be generalized as ( 1) ( can also be seen in Fig. 1 …”

“…A lot of SNR estimation methods have been developed in recent years. Earlier SNR estimators often work in the data-aided (DA) modes [1], which make use of the presence of some known training or pilot symbols in the data frame. Although the DA estimators tend to be more accurate, they suffer from the defect that additional bandwidth and power should be provided for training sequences [2].…”

Code-Aided Iterative SNR Estimator for M-APSK Signals Based on Expectation Maximization Algorithm

“…A large quantity of DA and NDA estimators has been developed for phase shift keying (PSK) signals [3][4][5][6]. Recently, M-ary amplitude phase shift keying (APSK) signals [7] have been proved to be superior to MPSK signals in nonlinear satellite channels and have been adopted in the second generation of Digital Video Broadcasting via Satellite standard (DVB-S2) [8].…”

“…Recently, M-ary amplitude phase shift keying (APSK) signals [7] have been proved to be superior to MPSK signals in nonlinear satellite channels and have been adopted in the second generation of Digital Video Broadcasting via Satellite standard (DVB-S2) [8]. However, the estimators in [3][4][5][6] cannot be directly applied to M-APSK signals. In [9] and [10], NDA SNR estimators based on six-and eight-order statistics were evaluated for two-and three-level constellations, like 16-and 32-APSK signals.…”