Abstract:  Collaborative spectrum sensing can effectively improve the sensing performance through utilizing the diversity from different cognitive users. However, this performance can also be degraded since the sensing data are sent via wireless reporting channels which could severely fade the sensing results. In this paper, we investigate the relationship between the number of collaborating users and system sensing performance. The optimal number of collaborating users is calculated and analyzed from two perspectives: licensed users and cognitive users. Our theoretical results along with the simulations show that collaborating all cognitive users in the network does not necessarily achieve the optimal performance. Instead, the optimal performance is reached by collaborating a group of users that have higher sensing signal-to-noise ratios. Additionally, the optimal number of users increases as the size of the network increases, and it decreases as the reporting channel error increases.