Abstract
The adaptive detector design problem for the direct sequence code division multiple access (DS-CDMA) systems in fading multipath channels is addressed in this dissertation. An adaptive multiuser detector with decision-feedback equalizer (DFE) architecture has been proposed to perform in the fading multipath channels. A tapped-delay-line filter with time-varying tap coefficients is used to model the fading multipath channels and each tap coefficient is modeled as a Rayleigh fading process. The fading multipath channels are estimated by some channel estimation algorithm. Then, the estimates are used for the multiuser DFE design. The main results are summarized as follows: Firstly, a symbol detection equation and channel estimation equation of DS-CDMA systems are given for the purposes of detector design and channels estimation, respectively. The channel estimation equation as well as the dynamics of fading multipath channels forms a novel state-space equation which is utilized to estimate the fading multipath channels precisely. Three situations about the fading multipath channels, including the known channel model, the partially known channel model and the unknown channel model, are considered. Secondly, we discuss two kinds of adaptive multiuser DFEs. One is the directly adaptive multiuser DFE via a one-stage recursive least-squares (RLS) design algorithm. The other is the channel-estimation-based adaptive multiuser DFE via a two-stage channel-estimator/minimum-mean-square-error (CE/MMSE) design algorithm, in which the tap weights of the DFE are adjusted by a channel estimator. For the case of the known channel model, a Kalman channel estimation algorithm is presented for the fading multipath channel estimation. We take not only the estimates of channel but also the statistics of the channel estimation errors into consideration of the DFE design. For the case of partially known channel model, an H_infty channel estimation algorithm is proposed to treat the uncertainties of channel parameters. For the case of unknown channel model, a weighted RLS (WRLS) channel estimation algorithm is proposed to estimate the fading multipath channels. Then, the estimates of channels are used to design the adaptive multiuser DFE via an MMSE criterion. Thirdly, we make a comparison among the directly, the channel-estimation based adaptive multiuser DFE and the optimal MMSE DFE via an abstract analysis. The optimal MMSE DFE means that the time-varying channel response are well-known for the equalizer. We also show that the directly and the channel-estimation-based adaptive multiuser DFEs are different realizations of the optimal MMSE DFE. Owing to their implementations, the proposed two-stage channel-estimation-based adaptive multiuser DFE has better tracking ability for the time-variations of fading channels and performs better than the directly adaptive one. The signal-to-interference-and-noise-ratio (SINR) performance and the bit-error-rate (BER) performance are also presented with the consideration of the impact of the imperfect channel estimation. Finally, two computer simulations examples are given. One is simulated under the land-mobile fading channel model and the other is simulated under the Butterworth fading channel model. The BER performance are investigated with the influences of channel SNR, the length of time-dispersive channel, the number of users, the fading rate of channels, the sensitivity to the uncertainties of channel parameters.