Neural network based congestion detection protocol

Congestion in Wireless Sensor Networks (WSNs) is an important challenge. Congestion causes packets loss, energy consumption as well as delay which decrease the sensor lifetime. For this reasons, congestion detection is crucial for WSNs. In this paper, a Neural Network Congestion Detection (NNCD)is proposed. It allows each intermittent node to estimate congestion and to identify malicious nodes by the use of local information such as participants, buffer occupancy, and traffic rate as input for neural network. Simulation results show that NNCD can detect the congestion level for intermittent nodes.