Joint synchronization and channel estimation in uplink OFDMA systems

We consider the uplink of an OFDMA system and address the problem of estimating the carrier frequency offset, timing error and channel response of each active user. In doing so we follow a maximum likelihood (ML) approach and assume that a training sequence is available. Unfortunately, joint ML estimation of all the above parameters involves a multi-dimensional grid-search that is difficult to implement in practical systems. Therefore we resort to the alternating-projection algorithm and replace the multidimensional search with a sequence of one-dimensional searches. Compared to other existing methods, the proposed estimator has more flexibility since it can be used with any subcarrier assignment scheme. Simulations indicate that the accuracy of the frequency estimates asymptotically achieve the relevant Cramer-Rao bound (CRB).