A Novel MMSE Based Predictive Method for Narrow Band Interference Removal

Abstract

A new system model based on state space representation of the received signal for Narrow-band interference suppression in Direct Sequence-Spread Spectrum (DS-SS) system is given. This model recursively estimates the interfering signal in the presence of desired information and white Gaussian noise. To further improve the performance, we employ MMSE criteria to estimate the transmitted symbols. The MMSE estimate requires low computational complexity than Code-aided techniques. Our simulation results demonstrate that this technique provides better SINR improvement over to Nonlinear predictor, Linear predictor, Kalman-Bucy predictor for-65 dB to-5 dB input SINR and up to 10 dB AWGN power. For high interfering signal and noise power, our results clearly show that our SINR improvement performance exceeds the SINR improvement upper-bound calculated for the prediction based techniques. Besides reducing the effect of interfering signal, our technique also reduces AWGN noise. Our model also provides better BER performance than Nonlinear predictor.