Skip to Main Content
The performance of a number of iterative interference mitigation receivers is investigated in the context of the CDMA downlink. Two types of receiver are considered, both based on spatio-temporal filtering but differing in the way in which the filter weights are initialised in the first iteration. In the first type, the filter weights are trained directly via time-multiplexed transmitted pilots. In the second type, the weights are computed based on explicit channel estimates derived via the transmitted pilots. Least squares and semiblind (constant modulus) optimisation criteria are utilised. It is shown that, while the receivers of the second type are initialised via filter weights which do not strictly satisfy a specific optimality criterion, they do result in superior performance in subsequent iterations. The performance of the receivers is evaluated via link-level simulations involving 2- and 7-cell indoor and urban propagation environments.