Subspace-based synchronization of DS-CDMA systems utilizing training sequences

In this paper, we propose several subspace-based algorithms for near-far robust synchronization of asynchronous direct-sequence code-division multiple access (DS-CDMA) systems. A tuning parameter is introduced in the form of a scalar weighting of the propagation delay estimator cost functions. The asymptotic (large data blocks) variance of the algorithms is derived and the optimum weighting strategy, in the sense that asymptotic variance is minimized, is found. The analytical results are validated by simulations. The algorithms are shown to be near-far robust, both analytically and experimentally. Performance is of course improved by using training sequences, however, not by much; furthermore, optimum weighting is shown to yield only modest performance gains.