27-02-2017, 12:20 PM
A strategy based on past / present values provided by each sensor in a network to detect its malicious activity. Basically, we are going to compare at each moment the sensor output with its estimated value calculated by an auto regressive predictor. In case the difference between the two values is greater than a chosen threshold, the sensor node becomes suspicious and a decision block is activated. Wireless sensor networks (WSN) are one of the most important and promising domains of the 21st century. The recent growth of the technology (greater computational power, small chips and microprocessors with less energy consumption) opened a new world for the investigation in this field. A sensor network is a collection of small distributed devices called motes, using sensors to measure (temperature, movement, pressure, sound) and for prediction (weather forecast, fire ignition, earthquakes, military attack, building safety). Some of the most important features of wireless sensor networks are environmentally free ownership and their ability to self-organize throughout the life cycle. When deployed in any type of environment, the motes are subject to attacks and without high security the information that passes through the network could be listened to and also altered. With this, the network of wireless sensors could be damaged and prove useless for certain purposes. Safe protocols are still under investigation, none were successful enough to be standardized. Although protocol security is a very important issue in the development of wireless sensor networks, the detection of anomalies and intruders is also a significant problem. After intrusion detection, the sensor network can make decisions to investigate, search, delete, or rewrite malicious nodes if possible. If intrusion detection is not performed at the appropriate time, the captured node code could be read and rewritten for malicious purposes.