<p>The behavior of devices connected to an Internet of Things (IoT) needs to be monitored to identify anomalous nodes. In this paper, we propose a protocol for anomaly detection that uses information about the traffic generation behavior of nodes in an IoT. A feature of the proposed protocol is its ability to allow nodes to improve their traffic generation behavior. We evaluate the performance of the proposed protocol through simulations. We focus on the effect of the traffic threshold and the fraction of grey nodes on the number of anomalous nodes in the network. Moreover, the traffic generation behavior of nodes is captured in a dataset and a deep learning technique called Bidirectional Long Short-Term Memory (BiLSTM) is applied to detect anomalies. We observe that there is a significant decrease in the number of anomalies when nodes are allowed to improve their behavior.</p>