As the wireless sensor network (WSN) evolve rapidly in many applications and fields, it has become increasingly vulnerable of various types of attacks. DoS attack is considered as one of the most dangerous attack that poses a major threat in wireless sensor network security and could have major effects and serious consequences in WSNs functionalities. Recently, intrusion detection systems become crucial security components. In this paper, we propose an approach deep learning to enhance the level of security in such network. We have evaluated and analyzed the efficiency of Deep Neural Network (DNN) in DoS detection, using the standard metrics of evaluation: accuracy, precision, F1-score and recall. Our model was carried out using a well-known dataset WSN-DS, intended for wireless sensor networks, containing four types of Dos attacks: Blackhole, Grayhole, Flooding and TDMA. The experiment result demonstrate the effectiveness of DNN in DoS detection with high accuracy achieved.