Abstract—Reliable communication in wireless sensor networks constitutes an essential factor in maintaining critical systems operational. Despite this, wireless sensor networks are known to be volatile and prone to faults disrupting their normal working state. Particularly in open environments, wireless sensor networks must be able to detect arising faults to minimize subsequent failures of the network. This study deals with the detection and identification of faults in wireless sensor networks, notably faults that occur due to externally driven events, affecting network services, such as data transfers and communication between nodes. Faults commonly occurring due to such factors are loss of connectivity because of faulty node interfaces, disrupted connectivity due to obstacles, and extreme packet loss because of increased noise conditions or congestion. Detection, identification, and recovery of sensor network faults have been studied extensively in the literature. In this paper, a Machine Learning approach is used to detect and diagnose these faults. A decision tree algorithm was used to train the model. The produced model is consistently able to identify the faults on test data with an overall accuracy of 96.46%. Results also include high precision and recall values for each separate fault case, thus producing a successful fault identification model.