Wireless sensor nodes are prone to frequent failures due to battery exhaustion and other external events, resulting in coverage and communication holes in the network. It is important to detect such holes and network boundaries. In this paper, we propose a distributed boundary detection algorithm that relies on local connectivity information to identify holes and boundaries of the network. The proposed mechanism does not make any assumptions regarding the computation of distances, thus reducing the computation overhead. Analytical results report a low false positive percentage of 5.8%, 4.6%, 4.1% for varying radio ranges of 20, 25 and 30 meters, respectively. We also illustrate the effect of radio range on the number of boundary nodes.