Analytical researches on a potential performance of multipath multiple-input multiple-output (MIMO) systems inspire the development of new technologies that decompose a MIMO channel into independent sub-channels on the condition of constrained transmit power. Moreover, in current studies of inter-symbol interference (ISI) MIMO systems, there is an assumption that channel state information (CSI) at receivers and/or transmitters is perfect. In this paper, we propose a hybrid design of precoding and equalization schemes based on the unweighted minimum mean square error criterion that not only eliminates the ISI but also improves the system performance. Additionally, the impact of imperfect channel knowledge at receivers on the system performance of MIMO ISI system is investigated. The simulation result shown that the proposed hybrid design of precoding and equalization with shared redundancy outperforms the conventional method in all considered scenarios. Furthermore, the proposed and the conventional schemes are extremely sensitive to the CSI factor, the performance of these systems is quickly deteriorated when the accuracy of channel estimation decreases.