UC San Diego

This dissertation examines the effect of severe narrowband interference on wireless communication systems. In single- carrier systems, the interference causes the adaptive equalizer to have an extended convergence time, where convergence is considered in terms of the bit error rate (BER). Two techniques are proposed to improve the convergence. The first method, data-aided initialization (DAI), initializes the Wiener weights from estimates derived directly from the received data and training sequences. This technique is shown to substantially reduce the number of training symbols needed for convergence. Further, two methods for obtaining the DAI weights are investigated. The use of multistage Wiener filters (MSWF) is preferable to a parametric approach to direct matrix inversion in terms of BER performance and number of training symbols needed. The second method is a two-stage system that utilizes a prediction error filter (PEF) as a pre-filter to the equalizer. It is shown that the two- stage system reduces the number of training symbols required to reach a BER of 1̄0⁻² by approximately two orders of magnitude without substantially degrading the steady-state BER performance as compared to the DFE-only case. In block-modulated multi-carrier systems the presence of a severe narrowband interference causes the degradation of a large number of subcarriers due to spectral leakage of the interference power after demodulation. Multi-carrier code division multiple access (MC-CDMA) obtains frequency diversity by spreading the data into every subcarrier, thus mitigating the effects of narrowband interference. On the other hand, orthogonal frequency division multiplexing (OFDM) requires the addition of coding and interleaving to obtain frequency diversity. The use of genie inserted erasures provides little to no improvement in BER performance, thus the PEF is proposed as an erasure insertion mechanism that notches out the tones located close to the interference, while leaving the remaining tones unaffected. This technique provides excellent results as compared to the case of no interference. This work was done at UCSD's Center for Wireless Communication, under the "Bandwidth Efficient Communications" project (CoRe research grant 06-10216) and supported by the Office of Naval Research, Code 313.