High rate underwater communications have traditionally relied on equalization methods to overcome the intersymbol interference (ISI) caused by multipath propagation. An alternative technique has emerged in the form of time-reversal, which comes at virtually no cost in computational complexity, but sacrifices the data rate and relies on the use of large arrays to reduce ISI. In this paper, spatiotemporal processing for optimal multipath suppression is addressed analytically. A communication link between a single element and an array is considered in several scenarios: uplink and downlink transmission, with and without channel state information and varying implementation complexity. Transmit/receive techniques are designed which simultaneously maximize the data detection signal-to-noise ratio and minimize the residual ISI, while maintaining maximal data rate in a given bandwidth and satisfying a constraint on transmitted energy. The performance of so-obtained focusing techniques is compared to the standard ones on a shallow water channel operating in a 5 kHz bandwidth around a 15 kHz center frequency. Results demonstrate benefits of focusing techniques whose performance is not conditioned on the array size. Optimal configurations are intended as a basis for adaptive system implementation in which channel estimates will replace the actual values.