Phase estimation of PSK signals using XTFD: A performances comparison between local and global adaptive methods

The quadratic time-frequency distribution (TFD) provides distribution of energy over the time-frequency plane for time-varying signals. Since phase information is not represented, the cross TFD (XTFD) is proposed to analyze phase shift keying (PSK) signals by providing localized phase information. However, the phase estimation does not yield desirable performances as the time-frequency representation is interfered by duplicated terms. The problem is solved by the proposed XTFD which uses an adaptive window to remove the duplicated term. The local and global adaptive algorithms are proposed to estimate the window width. It is shown that both algorithms meet the theoretical limit at a minimum signal-to-noise ratio (SNR) of 5dB. At lower SNR, the local adaptive method outperforms the global adaptive method at the expense of higher number of computation.