Abstract: Aiming to improve the transmission performance of the wireless sensor network (WSN) in the over-crowded ISM (industrial scientific medical) band, this paper proposes a novel cognitive wireless sensor network (CWSN) architecture by combining the cooperative spectrum sensing (CSS) technology in cognitive radio. In the CWSN, cognitive nodes first perform joint detection by energy detection and make decisions for local multiple channels. Then the sink node makes a decision fusion based on the local sensing results from cooperative cognitive nodes, and schedules sensor nodes to communicate on the best channel according to the global decision. On the premise of meeting channel detection requirements, the decision fusion rule and the number of cognitive nodes are optimized for maximizing network throughput. Simulation results show that the optimized CSS strategy can guarantee reliable transmission with a few cognitive nodes and significantly improve the network throughput while reducing network energy consumption and device cost.