Deterministic Estimation of Channel Parameters in Asynchronous CDMA Communications Over Frequency-Selective Channels

Designing appropriate user codes for training, we propose a channel gains and delay estimation scheme for asynchronous CDMA systems over frequency-selective channels. Our design arises from a simple but important property of "complex exponentials": a circular shift in a complex exponential vector looks like multiplying it by a unit magnitude complex number. The main benefit of the algorithm is its independency of near-far ratio which arises from the fact that MUI is deterministically eliminated before estimating the channel parameters. Also, the proposed scheme doesn\´t need to estimate any statistic (such as a correlation matrix) for parameter estimation. Also since we are only involving in linear computations and search over parameter space, the computational complexity is less than other schemes which usually use subspace-decomposition or other enormous computations. We first formulate the problem in noise-free case, and then to improve the performance in noisy environments, MMSE as well as ML approaches will be formulated and investigated