Wireless Sensor Networks (WSN) generates large quantities of data. For efficient application of the sensor data, discovering the required knowledge from the data is of crucial importance. Generally, the WSN generates data in form of streams, which are transmitted to the sink. The raw data generated in large quantities gives rise to higher communication overhead due to which the performance of the WSN is affected adversely. Association mining processes the data to find frequent patterns. When association mining is applied in-network of the WSN, only the frequent patterns discovered in the raw data need to be transmitted to the sink. Thus, the communication overhead is reduced significantly.