Efficient sensing schemes for the active reduction of sound radiation from plates are presented based on error signals derived from spatially weighted plate velocity or near-field pressure. The schemes result in near-optimal reductions as compared to weighting procedures derived from eigenvector or singular vector analysis of the radiation operator. Efficient control configurations are suggested using a, possibly analog, front-end implementing a bank of spatial weighting functions and a digital controller with a minimized number of input and output channels. The performance of different weighting functions is compared, as well as the performance of different frequency-dependent filtering functions. Design rules are given for the sensor spacing, the number of weighting functions, the number of actuators, and the corresponding controller dimensionality.