Signal detection and channel estimation for pilot-aided CDMA systems

Abstract

In this thesis, we propose signal detection and channel estimation techniques for CDMA systems with pilot symbols or channels in fading channels. For pilot symbol aided CDMA systems, constrained minimum mean square error (CMMSE)-RAKE receivers for multipath fading channels are developed by extending the CMMSE receiver for flat fading channels [15]. Based on the observation that interpath interference (IPI) causes a bias of the channel estimator in [15], two types of receivers that can remove such a bias are proposed. The first one (Type 1) employs a new maximum likelihood (ML) channel estimator, while the second (Type 2) imposes multiple constraints on the MMSE optimization. Closed-form expressions of the signal-to-interference and noise ratio (SINR) and bit-error rate (BER) of these receivers are derived. As a result, it is shown that the Type 1 and 2 receivers can both outperform existing RAKE receivers and that Type 1 performs somewhat better than Type 2 at the expense of additional complexity. As a zero-delay channel estimation technique combating with fading environments, the channel estimator consisting of the PSA (pilot symbol-aided) and DD (decision-directed) predictor, is proposed. The channel estimates are obtained by combining the estimates from the PSA and the DD channel predictor. The optimal combining coefficients are analytically derived, and the BER expression when the proposed channel estimator is employed by the RAKE receiver, is also derived. As a result, it is shown that the proposed channel estimation technique can outperform the PSA in [16] and the receiver with non-coherent DPSK modulation. Finally, a per-survivor processing (PSP) sequence detection receiver is proposed for DS/CDMA systems with pilot and traffic channels. The proposed receiver jointly estimates channel parameters and data symbols from pilot and traffic channels, and is derived by modifying the receiver in [24]. As a result, the new receiver can outperform the r...