Abstract—Traditional power allocation schemes in Orthogonal Frequency Division Multiplexing (OFDM) based Cognitive Radio Networks (CRNs) are achieved under perfect channel state information (i.e., exact parameter information). Due to channel estimation errors and feedback delays, however, channel uncertainties are inevitable in practical CRNs. In this paper, considering bounded channel uncertainties, a robust power allocation algorithm is proposed to minimize the total transmit power of Secondary Users (SUs) subject to the interference temperature constraint of primary user and the received Signal-to-Interference-plus-Noise Ratio (SINR) constraint of SU where the non-convex optimization problem is converted into a convex optimization problem that is solved by dual decomposition theory. Numerical results demonstrate the effectiveness of the proposed algorithm by comparing with the non-robust algorithm in the aspect of suppressing the effect of parameter uncertainties.