Joint EM channel and covariance estimation with sufficient-statistic chip combining for a SIMO MC-CDMA anti-jam system

Joint iterative channel and covariance estimation with sufficient-statistic-based chip combining is studied for the downlink of a multicarrier code division multiple access (MC-CDMA) system operating under partial-band noise jamming (PBNJ). To estimate the channel and AWGN-plus-PBNJ covariance, we apply the expectation-maximisation (EM) algorithm which exploits the log-likelihood ratio (LLR) from the decoder. The proposed chip combiner guarantees no loss of information in the soft output generated by the demapper. In our approach, the channel estimator or the chip combiner only requires the AWGN-plus-PBNJ power of each subcarrier to suppress jamming. Therefore, significant complexity reduction is achieved because the receiver does not need to estimate the jammer state information (JSI) and individual variances of the AWGN and PBNJ on each subcarrier for chip combining. Simulation results show that the sufficient-statistic chip combiner always offers a smaller bit error rate (BER) than the minimum mean square error (MMSE) chip combiner provided that reliable estimates of the channel impulse response and noise-plus-jamming covariance are available. Copyright © 2010 John Wiley & Sons, Ltd.