Performance of Adaptive Filtering Techniques Using the Fractional Fourier Transform for Non-Stationary Interference and Noise Suppression

The Fractional Fourier Transform (FrFT) performs better interference suppression than the fast Fourier Transform (FFT) when the signal-of-interest (SOI) or interference is nonstationary. Minimum mean-square error (MMSE) based filtering in the FrFT domain provides additional benefit in interference suppression in non-stationary environments. However, MMSE filtering requires computational covariance matrix inversion. Furthermore, non-stationary environments require fewer samples than needed to form the covariance matrix or to invoke most reduced rank techniques. Hence, MMSE-FrFT filtering results in errors. In this paper, we propose to apply the correlations subtraction architecture of the multistage Wiener filter (CSA-MWF) in the FrFT domain to overcome these problems. We compare the proposed MWF-FrFT algorithm to the MMSEFrFT algorithm and to the conventional MMSE-FFT algorithm by simulation. Using a BPSK signal in chirp noise and Gaussian pulse interference as examples, we show bit error rates (BERs) with 2 − 4 dB less Eb/N0 and just N = 4 samples per block.