Cooperative relaying strategies have become a major topic in the wireless research community and that decode-and-forward relay technology can suppress noises effectively. In order to achieve the maximum capacity of the system, the optimal allocation algorithm is proposed under the assumption that all nodes know accurate channel instantaneous state information (CSI). Because it is hard to obtain a general solution expression during deducing, we provide an iterative power allocation algorithm with a simpler structure. Numerical results show that adopting the proposed power allocation algorithm, the capacity of MIMO relay channel with DF relay scheme can improve 1.4 bps/Hz. Through analysis, a proper iterative precision value can be found to keep the good balance between system capacity and complexity. To further improve the system capacity, we use the subcarrier pairing strategy before power allocation.