Sensor nodes located on the boundary of the covered area often cause difficulties in maintaining a wireless sensor network's lifespan. This study proposes a simple method to approximate boundaries in wireless sensor networks. The approach uses only connectivity information of a network while not requiring nodes localize themselves. The proposed method divides a network into clusters and forms partial boundaries, then extends detected partial boundaries until the whole boundary is formed. Using a node's 1- and 2- hop neighbors, every node can decide locally whether itself is located near the boundary. The implementation shows that the approach can efficiently determine boundaries in the coverage of a WSN.