TECHNIQUE OF ATTACK DETECTION IN SELF-ORGANIZING DECENTRALIZED WIRELESS SENSOR NETWORKS

Abstract

The work is devoted to the development of an attack model and a technique for detecting attacks in self-organizing decentralized wireless sensor networks. The proposed model describes possible types of attacks and their characteristics, taking into account the properties of self-organization and decentralization. The methodology is focused on the protection of wireless sensor networks deployed on the ground, used for emergency response, and describes the stages of the process of building and configuring an attack detection mechanism based on data collection algorithms in wireless sensor networks and the use of machine learning methods. The analysis of possible types of data that need to be collected at the nodes of wireless sensor networks to detect attacks is carried out. The distinctive features of the proposed technique include the sets of features used that characterize specific types of attacking influences and allow detecting attacks with high values of the detection quality indicator. On the fragment of the hardware-software prototype of wireless sensor networks used in the work with an attack detection mechanism built into it, an experiment was conducted to check the quality of attack detection, confirming the correctness of the proposed technique.