Support-vector-machine-based intrusion detection method for wireless sensor network

Abstract

The invention discloses a support-vector-machine-based intrusion detection method for a wireless sensor network. A flow pattern of a wireless sensor network is established to describe a flow mode of a system network, a flow characteristic parameter is extracted from a network data flow packet according to the flow model and is processed by normalization; on the basis of the flow characteristic parameter, a support vector machine is trained and a flow mode of the system network is learned; and on-line intrusion detection is carried out by using the trained support vector machine. According to the intrusion detection method disclosed by the invention, the operation state parameter of the wireless sensor network is extracted by using the network flow model; real-time intrusion detection can be realized only by extracting a few of characteristic parameters without deep parsing on a network message. Moreover, the method is also suitable for cyclic polling type and event trigger type wireless sensor network intrusion detection. While the real-time performance is guaranteed, the detection rate is improved substantially and the false alarm rate is reduced.

Description

technical field [0001] The invention belongs to the technical field of wireless sensor network information security protection, and more specifically relates to a support vector machine-based wireless sensor network intrusion detection method. Background technique [0002] With the rapid development of wireless transmission technology, low-power processors and embedded computing technology, wireless sensor networks have been more and more widely used; since the sensors of wireless sensor networks are mostly deployed in unattended environments, it is easy to suffer malicious cyber attacks. As an intelligent system that can actively discover attacks and provide alarm information, the intrusion detection system can be used as the second line of defense of the system. [0003] Existing technologies for intrusion detection of wireless sensor networks, such as the invention patent application (CN201510606829.0), "Journal of Chongqing University of Posts and Telecommunications (Na...

AI Technical Summary

Problems solved by technology

[0003] For the existing technology of intrusion detection of wireless sensor network, such as the invention patent application (CN201510606829.0), "Journal of Chongqing University of Posts and Telecommunications (Natural Science Edition)" 2016 No. 1 paper "Intrusion detection of wireless sensor network based on information entropy Genetic Algorithm" is to analyze the header information of all packets, and then extract network features and detect attacks; this type of method can work normally when the network scale is small, but in large-scale, large-scale wireless The network will be unavailable due to taking up too many system resources

In order to reduce the power consumption of sensors as much as possible, more and more wireless sensor networks adopt the event-triggered communication mode. Sensors do not usually send data, and only transmit information to the cluster head node when a specific event is detected; compared to The working mode of periodic polling, event triggering will bring more uncertain factors to the network, making the network traffic in the system bursty; how to detect attacks in a heterogeneous environment with both periodic polling and event triggering is a huge challenge

Images