A COMPARISON OF RECURSIVE LEAST SQUARES AND KALMAN FILTERING EXCISORS FOR SWEPT TONE INTERFERENCE

This technical note presents a Kalman filtering approach that is used for filtering narrow-band interferers out of direct sequence spread spectrum signals. The approach is based on the digital phase-locked loop Kalman filter and is close to optimum in so far as de-modulating an FM-type of interferer. Because the interference is assumed to be much stronger than either the signal or noise, the Kalman filter locks onto the interference and produces estimates of its phase and envelope. The algorithm is compared, through computer simulation, with the recursive least squares lattice algorithm for the case of a swept tone interferer. Examples of the phase- and envelope-tracking capabilities of the algorithm are presented, followed by bit error rate curves for the case of an interferer of bandwidth equal to 20 % of the chip rate of the spread spectrum signal. The results show that with proper selection of the Kalman filter parameters, the Kalman filter excisor can outperform the recursive least squares algorithm by 2 dB. It is suggested that performance could perhaps be improved further by using a higher order Kalman filter to remove the residual phase-tracking error produced by the second order Kalman filter