This paper studies the robust resource allocation to improve the robustness against the channel state information (CSI) uncertainties for energy harvesting (EH)-aided device-to-device (D2D) communication underlaying cellular network. In this network, EH-aided D2D transmitters can reuse the uplink cellular sub-channels to perform transmission to the desired D2D receivers. Assuming the link gains in the interference channels are uncertain, we formulate the robust resource allocation problem with

more »

... view to maximizing the sum-rate of the D2D links while guaranteeing the power outage Quality-of-Service (QoS) requirements and interference constraints. In doing so, both deterministic and statistical uncertainty model are adopted to specify the CSI uncertainties. This resulting robust problem is essentially non-convex and intractable in its original form. Therefore, we firstly utilize the worst-case approach and chance constraint approach to find its tractable and convex forms. Then, via employing the Lagrangian dual method, we derive the semi-closed expression for the power allocation and two rules for assigning sub-channels. Based on the theoretical results, a distributed subgradient-based iterative algorithm is developed to achieve the optimally robust resource allocation. Finally, an in-depth simulation is conducted to verify the effectiveness of the proposed algorithm and evaluate the effect of CSI uncertainties and power outage QoS requirement on the sum-rate of the D2D links. INDEX TERMS Robust resource allocation, D2D communication, convex optimization, channel uncertainty, chance constraint.