Preventing alarm storms in WSNs anomaly detection applications

Wireless Sensor Networks (WSNs) anomaly detection applications aim at identifying events which occur in a monitored area. Forest fire, landslide or intrusion detection are some examples. In order to perform anomaly detection, wireless sensors are deployed in the area of interest and the nodes monitor a physical parameter. If this parameter exceeds a predefined threshold, the node generates an alarm and sends it in direction of the collection point (the sink). In some cases, the anomaly can span over several nodes, so each of the nodes produces one alarm, resulting in several alarms sent for the same event. These alarms are produced and sent in a very short amount of time, this phenomenon is called alarm storm and causes network congestion, collisions and increases the energy consumption. In this paper, we propose a mechanism which allows to reduce the number of redundant alarms. The proposed mechanism consists in a competition among nodes sensing the anomaly, which aims at determining the node which is the better located to actually send the alarm. The proposed mechanism is evaluated by extensive simulations, the results show that the scheme is able to reduce drastically the number of redundant alarms while not interfering on the MAC and routing processes which guarantee timeliness and reliability to anomaly detection applications.