Distribution system restoration (DSR) is intended to restore service to interrupted customers by a sequence of switching operations. An effective DSR strategy reduces interruption duration and consequently enhances reliability of distribution systems. In this paper, a systematic method is proposed for reliability analysis of smart distribution systems with DSR technique and remote control capability. The spanning tree search algorithm is used to generate optimal DSR schemes that restore the maximal amount of load with a minimal number of switching operations. Switching sequence is decided by a set of rules. Breadth-first search technique is used to determine the restoration time for customers. A 4-feeder 1069-node unbalanced test system with microgrids, developed by Pacific Northwest National Laboratory (PNNL), is simulated to demonstrate the effectiveness of the proposed method.