Edge sensor detection is often used in identifying regions that are affected by various factors in wireless sensor networks. A statistical methodology based on distributed detection theory and the Neyman-Pearson criterion is developed for edge sensor detection in this research. The input sensor statistics are assumed to be identically independently distributed in our framework. Edge regions and sensors are determined using a hypothesis test, where the observation model for each hypothesis is derived. A sub-optimal distributed detection scheme, which is optimal among detectors having the same test at all local sensors, and the way of choosing the optimal operating point are described. The condition under which the proposed scheme outperforms the optimum detector based on a single sensor is presented. Furthermore, the noisy channel effect is considered, and a method to overcome this noisy effect is addressed. The performance of the proposed distributed edge sensor detection scheme is studied via computer simulation, where the ROC curves are used to demonstrate the tradeoff between the cost (in terms of the sensor density) and detection accuracy.