In the case of sudden disaster accidents, the emergency communication network plays a veryimportant role in the disaster relief because of its reliable and stable network carrying capacity, which caneffectively solve the prominent problems such as the reduction of frequency spectrum detection rate in thesevere environment. In this paper, a cluster based cooperative spectrum sensing algorithm for emergency communication is proposed. In order to reduce the influence of channel attenuation and improve the accu.racy of spectrum perception, the nodes in the network are divided into several clusters, and the data fu.sion of two-level multi-user collaboration is completed within and among the clusters. Finally , the detec-tion rate, effectiveness and applicable environment of the algorithm are analyzed and compared, and itsfeasibility is tested by Matlab software.