Device-to-device (D2D)-enabled Internet-of-things promises higher spectral efficiency and system capacity by sharing the cellular spectrum and offloading the cellular traffic. However, it poses new challenges in terms of resource allocation because of interference from D2D to base station and vice versa in the downlink. Moreover, energy efficient operation of the network becomes a major challenge because of unattended operation of devices. In this study, we investigate resource allocation for the energy-harvesting-aided D2D communication underlaying cellular network. We formulate a joint subcarrier assignment and power allocation problem for multiple cellular and energy harvesting direct D2D links to maximize the overall sum rate subject to quality of service, subcarrier reuse, power, and energy harvesting constraints. The problem is formulated as a mixed integer nonlinear programming and is difficult to be solved in polynomial time. Therefore, a low complexity algorithm named energy harvesting and gain-based resource allocation (EHGRA) is proposed, which determines reuse partners considering interference among D2D and cellular links, and then, allocates power optimally to them such that the sum rate is maximized. Our performance evaluation demonstrates that energy harvesting can boost up the system performance and at the same time can achieve higher sum rate over short frame duration. Comparison shows that EHGRA performs better than the existing algorithms in terms of overall sum rate.