The paper proposes a new middleware named Semantic Sensor Network which connects the variety of physically grounded applications to the information of the real world. Semantic Sensor Network infers and describes the state of an environment using logical expressions. The target which Semantic Sensor Network describes is an environment where wireless sensor nodes are attached to daily items. The remarkable achievement of Semantic Sensor Network is to employ explicitly the concept of a class and an instance on the sensor network so as to construct efficiently an interpretation model for sensor data and queries. We have developed two physically grounded applications on Semantic Sensor Network; one is a GUI and the other is a robotic system. They use Semantic Sensor Network to obtain information about the environment.