Abstract: In multiple primary and multiple secondary users’ cognitive multiple-input multiple-output networks, the secondary users unilaterally eliminate the interference between primary users and secondary users, and bring about serious loss of secondary users’ performance. To solve this problem, a joint primary users and secondary users channel matrix is proposed for triangular decomposed interference alignment. Firstly, the channel matrix of the primary users system and the secondary users system is sorted according to the user channel quality. Then, the primary and secondary user channel matrix is triangularly decomposed. Finally, the feasibility of the proposed algorithm is verified by the least mean square error algorithm. The simulation results show that the proposed algorithm can effectively improve the upper limit of secondary users’ degrees of freedom, and the system capacity, the average energy efficiency and the ability to suppress interference of secondary users network and the primary users network.