Hidden Markov model algorithms for narrowband interference suppression in CDMA spread spectrum systems

Abstract

Prefiltering of received CDMA-spread spectrum signals to suppress narrowband interference is known to significantly improve the receiver performance. An adaptive nonlinear filtering algorithm is proposed, based on the cross-coupling of two optimal filters, a hidden Markov model filter for signal estimation, and a recursive prediction error estimator for interference determination. Comparisons with the EM-derived HMM-EM scheme presented in Krishnamurthy and Logothetis (1999) demonstrate the improved performance of our HMM-RPE algorithm, both in relation to SNR improvement and properties of convergence.