Correlation-Based Security in Time Synchronization of Sensor Networks

It is very important to monitor the water quality of lakes since any abnormal chemical components/pollutants can possibly cause health problems. Chemical Water Sensors can be used for such long-term monitoring purpose. In this paper, we propose a scalable, low-energy, delay-tolerant Water-quAlity moniToring sEnsor netwoRk (WATER) model, which has essential differences from terrestrial radio sensor networks due to its highly variable, long propagation delay and mobility nature. In the vertical direction, we propose a light-weight time synchronization mechanism that can achieve satisfactory timestamp accuracy. On the other hand, malicious people can use many network attacks (such as Sybil attacks, wormhole attacks, replay attacks, Byzantine attacks, etc.) to mislead water quality monitoring in WATER platforms. To make our time synchronization protocol dependable, we propose a correlation-based security model to detect outlier timestamp data and identify nodes generating insider attacks, which is different from external attacks due to the complete keying material disclosure. Our correlation-based security scheme can also countermeasure many insider attacks (i.e. assuming the enemies already captured the water sensors and got to know the keying materials). Detail experiments have validated the efficiency of our security approaches. The proposed secure time synchronization mechanism (we call it WATERSync) is especially important to navy/military underwater sensor systems.