A new concept of current sensor reduction technique called multi-branch current sensing technique (MCST) is proposed in this dissertation. In the proposed current sensing method, one more branch currents are simultaneously measured several times in a single switching cycle by using a single current sensor. After that, the current reconstruction algorithm is applied to obtain all phase currents information. Compared to traditional single current sensor techniques (SCSTs), the proposed method samples the output of the current sensor regularly, and the current sensing dead-zone is dramatically reduced. Since the current sampling is performed periodically, its implementation using a digital controller is extremely simple. Moreover, the periodical dead-zone and the dead-zone near the origin of the voltage vector space which have been a big problem in the existing methods can be completely eliminated. Accordingly, there is no need to have a complicated vector reconfiguration or current estimation algorithm. The proposed MCST also takes the advantages of a SCST such as reduced cost and elimination of the sensor gain discrepancy problem in the multiple current sensor method. The fundamental concept, implementation issues, and limitation of the proposed MCST are described based on three-phase systems first. After that, the proposed MCST is adopted to two-phase inverters and multi-phase dc-dc converters with little modifications. Computer simulations and hardware experiments have been conducted for a three-phase boost converter, a three-phase motor drive system, a two-phase two-leg inverter, a two-phase four-leg inverter with bipolar modulation, a two-phase four-leg inverter with unipolar modulation, and a four-phase dc-dc converter applications. From the simulations and the experimental results, the feasibilities of the proposed method mentioned above are fully verified. / Ph. D.