Malicious Node Detection in Wireless Sensor Networks Using an Autoregression Technique

In this paper we propose a strategy based on past/present values provided by each sensor of a network for detecting their malicious activity. Basically, we will compare at each moment the sensor´s output with its estimated value computed by an autoregressive predictor. In case the difference between the two values is higher then a chosen threshold, the sensor node becomes suspicious and a decision block is activated.