Underwater scuba divers are in dangerous environment exposed to internal and external dangers. In order to detect those dangers, in part of CADDY project DiverNet system was designed and made to track the diver. DiverNet uses group of inertial sensors to track movement of the diver and has the ability to add additional analogue sensors to detect other states, like heart rate and breathing rate. System sends read data to the surface where it will be processed and anomalies will be detected. Anomalies are detected using three methods. First method is parametric detection. Parametric detection observes if read value of physiological state is above or under allowed values. Last set of N values is used to calculate normal distribution from which is then calculated what is overlap of new value and the expected value calculated using the distribution. If the divergence is too big, this value is observed as an anomaly. Third and most thorough method of detecting anomalies is detection using mahalanobis distribution. Mahalanobis distribution as normal distribution takes into account last N set of values, but it also takes into account relationships between different aspects of human system. For example it observes as change in one parameter affects the other. Using this approach it detects range of different states and can detect which of this states are unusual and therefore anomalies. ROS was used as a framework for implementing algorithms and it enables to view the alarms in graphical user interface which has been implemented in rqt.