Interference alignment (IA) has been proposed to optimally manage the interference aiming at providing the maximum degrees of freedom of interference channels with multiple dimensions across space, time, or/and frequency for multiuser wireless communications. Therefore, an optimal IA based power loading is proposed in this paper to improve the throughput of the OFDM/FBMC based MIMO cognitive radio (CR) systems. In the proposed algorithm, all secondary users are enabled to share the available spectrum on the base of IA technique without affecting the quality-of-service of the primary system. Furthermore, a sub-optimal power loading algorithm is proposed to achieve the performance of the optimal algorithm with lower computational complexity. Moreover, spectral efficiency comparison between MIMO-OFDM and MIMO-FBMC is presented. Simulation results show that IA based power loading achieves a significant sum-rate increase of CR systems compared to traditional orthogonal multiple access techniques. Additionally, IA based power loading achieves better sum-rate improvement with FBMC than OFDM physical layer