Wireless sensor networks (WSN) have been identified for their utility in different areas such as defense, weather monitoring, patient monitoring, and tracking smart environments. Wireless body area network (WBAN) is one of the types of WSN technology which is used only in patient monitoring. It is a network of low-power biomedical nodes for observing vital parameters of sick patient remotely. WBAN is very much vulnerable and prone to external attacks. It would be difficult for medical professionals to diagnose correctly if the data is modified by adverse attack which leads to incorrect analysis of patient condition like false alarm. These alarms are motivated to address the challenges of adversarial attacks in WBAN and the defensive methods available. Deep learning methods are opted by many users to handle this problem with their variety of models. In healthcare monitoring systems, DL models are employed to extract valuable feats from homogenous and heterogeneous healthcare data of tracking patients’ well-being. This survey aims to bring idea of recent advances in robustness in DNN model with the safety of information. This paper is giving idea of different types of adversarial attacks and generation methods of attacks, basic defensive methods from these attacks, and different types of deep learning models which can be fooled by adversarial attacks and countermeasures from recent publication. So, robustness of different DL techniques needs to be identified.