کنفرانس مهندسی برق ایران

صفحه اصلی / سی امین کنفرانس بین المللی مهندسی برق

Stable Target Tracking in Wireless Sensor Networks Under Malicious Cyber Attacks

نویسندگان :

Jafar Akhondali¹ Mohammad Taheri²

1- دانشکده مهندسی برق و کامپیوتر/ دانشگاه شیراز 2- دانشکده مهندسی برق و کامپیوتر/ دانشگاه شیراز

کلمات کلیدی :

Wireless Sensor Networks, Security, Location Prediction, Trajectory Clustering

چکیده :

Due to the advance of hardware in recent years, Wireless Sensor Networks has attracted many research interests. Target tracking is an important application in Wireless Sensor Networks that like any other application, suffers from security issues. Previous studies relied on mathematical methods like Kalman Filter to detect attacks. In this paper, it has been tried to detect and adjust attacked target locations by a novel methodology. At first targets are tracked over time and then the similar trajectories are clustered into the same groups. Then for each group, a separate LSTM model is trained for prediction that tries to predict the next target location by learning previous target locations. Then for new targets, the attack is detected when the prediction location by the LSTM model has an abnormal distance from the new location of the target. In order to find a corresponding LSTM model for a new target in real-time, several methods have been proposed. As the known available datasets were not complex enough for our experiments, a tool has been developed to generate customized complex trajectory groups. The results are evaluated on twenty different trajectory clusters that prove the effectiveness of the proposed methods compared to similar research by four different types of attacks that are applicable in target tracking applications.