This paper studies optimal resource scheduling in simultaneous wireless information and power transfer (SWIPT) system, where a half-duplex hybrid access point (HAP) coordinates the transmissions of a downlink user and an energy harvesting embedded uplink user. In downlink, the HAP transmits information to the downlink user and simultaneously uses the same signal to charge the uplink user. In uplink, the uplink user transmits information to the HAP by utilizing the energy harvested in downlink. We formulate the joint optimization of link scheduling and power allocation as an mixed-integer programming problem. Our goal is to maximize the system sum-rate subject to the binary link selection and the physical power causality. We analyze the common feature among all the optimal link scheduling polices and solve the problem optimally by characterizing the properties of optimal power allocation. We also extend our results to multiuser scenario. Simulation results show that our scheme can significantly improve the system performance compared to the conventional schemes.