This paper initials with a brief discussion of future applications in wireless communications and of challenges posed to cognitive radio. It creates new thinking on using adaptive blind source separation algorithm to identify the primary user. The statistics of cognitive base station are provided by new embedded real-time database Berkeley DB. An innovative cloud storage strategy for multiple base stations’ real-time measurement information, proposed in this paper, is to achieve primary user’s accurate statistics. The proposed methodology can make great improvement on the MIMO system performance in the cognitive radio networks by taking new distributed cognitive engine framework with Hadoop cluster multi-nodes. Another contribution of the proposed approach was the benefit to the cognitive performance which can be characterized by the probability of spectrum coexistence on the basis of building up a distributed cognitive framework system of cognitive multi-user nodes under homogeneous network spectrum sharing environment.