Abstract: When multiple types of users coexist in a cognitive Multiple Input Multiple Output (MIMO) system, different types of users are subject to different constraints and have different demands for time-slots. In this paper, we aim to solve the problem of energy consumption optimization in cognitive MIMO systems with multiple types of users coexisting and propose a non-orthogonal time-slots allocation method based on Sparse Code Multiple Access (SCMA) to meet the communication needs of more types of users. When allocating time-slots, a dynamic time-slots allocation algorithm is proposed to ensure that different types of users will not occupy the same time-slots by designing virtual users for time-slots allocation. Simulation results and analyses show that when multiple types of users in the system exist in both eMBB (enhanced Mobile Broadband) and URLLC (Ultra-Reliable Low Latency Communications) scenarios at the same time, the non-orthogonal dynamic time-slots allocation based on SCMA is able to reduce the energy consumption compared to the traditional orthogonal allocation, and dynamic slot allocation algorithm can significantly reduce system energy consumption compared to that of the non-orthogonal time-slot allocation of fixed time-slots.