Abstract: To improve throughput and spectrum utilization in wireless sensor network, the authors present design and implementation of an ARM9 based multi-interface cognition-enhanced wireless sensor node. The node adopts architecture of single processor multiple radios, which is different from the typical single processor single radio architecture. Based on STR911 processor, the node can support four interfaces, two of which are for the CC2420 radio, and the other two are for the nRF905 radio. With these four radios, the node can work on ISM 433/868/915MHz and 2.4GHz ZigBee band. The node is equipped with light and temperature sensors, which can be used as a light and temperature monitor. Compared with single interface sensor node, node with multi-interface provides more flexibility for wireless sensor network, for it can be used as a router between 2.4GHz and 433MHz sensor networks. Also multi-interface makes the design of MAC protocol for wireless sensor network easier. Experiments give the measurement and validation of the nodes throughput and cognitive capability. The results show that the throughput has increased by 68.41% than Atmega128 based node. Also the average sense delay is 1.4782ms shorter for each channel. Compared with the method of CSMA/CA, communication delay is 11.86% shorter with the cognitive method, which can avoid the impact of interference effectively.