Malicious attack detection in underwater wireless sensor network

Abstract

Under-water wireless sensor network (UWSN) in one of the auspicious technology for marine observation. The applications of underwater sensing has several domain that range from oil industry to aquaculture. Some of the UWSN applications includes device checking, underwater ecosystems monitoring, forecasting of natural disasters and disturbances, exploration and survey missions, as well as study of oceanic life. With the characteristics and applications platform of UWSN, security of UWSN is a critical issue and had drawn significant attention to the researchers. In order to have a functional UWSN to extract the authentic data safeguarding and protection mechanisms are crucial. Malicious node attacks has accomplished eminence and constitute the utmost challenging attacks to UWSN. Several research has been conducted to protect UWSN from malicious attacks but majority of the works depend on a defined threshold prior to the deployment or a training data set. It is a complication and challenge for UWSN that with no established security groundwork a UWSN required to detect the malicious attacks. In this paper, we support vector machine to identify the malicious attacks in a UWSN. SVM delivers good result and its training time is much smaller comparing with neural networks.