| Multiuser code-timing acquisition, which parallels the well acknowledged research on multiuser detection (e.g., [57] and references therein) for direct-sequence (DS) code-division multiple-access (CDMA) systems, has been receiving increasing interest recently. A multitude of near-far resistant code-timing acquisition techniques have been proposed in the last few years (see [43] [48] [64] [3] [50] [2] [30) [25] [20] [37]). However, most of these techniques assume short or periodic spreading codes, whereby the code sequence periodically repeats itself for every symbol duration. In addition, most existing schemes assume rectangular chip waveforms that are not bandlimited. This is in contrast to bandlimited chip waveforms, such as the square-root raised cosine pulse, that are used in real CDMA systems.; In this dissertation, we have made an intensive study on the code-timing estimation problem for asynchronous DS-CDMA systems and proposed a number of codetiming acquisition schemes from different perspectives. Code acquisition is essentially to determine the propagation path delays between transmitters and receivers, which are prerequisite for all the multiuser detection and estimation schemes. We have considered both time-domain filterbank based and frequency-domain filterbank based approaches. We have studied both training-based and blind schemes. We have developed both decorrelating based and interference resistant methods. We have examined both deterministic multiuser estimator and Large-Sample Maximum Likelihood based single user estimator. In addition, we have investigated the derivation of the lower bound for all unbiased estimators and the efficient implementations with respect to computational complexity. In particular, we have paid close attention to CDMA with bandlimited chip pulse modulation, which is usually neglected by some of the existing multiuser code synchronization schemes assuming rectangular chip waveforms. Note that extending these schemes to deal with bandlimited waveform is in general difficult, due to highly nonlinear complexity. Finally, we have presented the numerical results of all the proposed schemes respectively and evaluated them by comparing with the existing techniques in Monte Carlo simulations. |
