Using Trusted Networks to Detect Anomaly Nodes in Internet of Things

The increase of Internet of Things has insert many challenges on networks studies. This kind of network suffer with problems like wide traffic, fault and monitoring problems. Regarding this, the concept of trust has gained increasing attention on academia by its comprehensiveness through the nodes behavior such as energy consumption, data transmission, and processing time. If a node has a different behavior from its neighbours it can be considered an anomalous node. In this paper, we propose a method for identifying if nodes are anomalous, using only their own monitored data by computing trust values for each node. Also, a data compression method is applied to help reduce the network traffic. This method is capable of signaling and separating anomalous data coming from different nodes in order to maintain the lowest possible interference level due to errors, frauds or malicious attacks. Our objective is to avoid errors in a posterior phase. For trust measurement we have used Subjective Logic that has been recently explored for these purposes. Experiments demonstrate that our method is feasible and help to reduce package size and spare energy.