The Intrusion Detection Technology Of Wireless Sensor Network

Wireless sensor network is widely used in numerous special areas including military, industry, enviroment monitoring and prediction and so on. Usually, wireless sensor network consists of massive sensor nodes. The security of the wireless sensor networks are faced with great challeges owing to the characteristic of the sensor nodes and the uncertainty of application scenarios. The study of intrusion detection technique has been one of the most important issues existing in the wireless sensor networks, which can prevent external attacks efficiently and detect internal attack in real time. This paper focuses on the intrusion detection technology for the physical layer attacks and introduces some methods for the intrusion detection such as utilizing geometrical idea,data mining technique and thought of mathematical statistics. In view of Sinkhole attack, Wormhole attack, denial of service attack and HELLO Flooding attack, this paper puts forward two new intrusion detection mechanism:(1) Usually, Sinkhole attack is launched by a malicious node near the Sink node which sends fictitious information to the nodes within its radio range that it can offer less number of hops and less delay than other normal routing paths to the Sink node. The compromised nodes will broadcast the false messages within their clusters, thus a large vortex is formed and more severer attack is occurred. This paper proposes a novel intrusion detection technique based on the received data packets flow. The simulation results show that the method is of relatively high detectin rate, less energy consumption and low false pasitives rate. Furthermore, the technique can be used to detect the Wormhole attack which is similar to the Sinkhole attack, and the experimental results is desired.(2) In this paper, we propose a new detection protocol based on the residual energy of nodes, which can be used to detect Sinkhole attacks, denial of service attacks and Flooding attacks. These three kinds of malicious attacks usually need send and receive more packets, which leads to larger energy consumption. The detection mechanism defines the residual energy trust value and the rate of the behavior of nodes, and the trust value of nodes is obtained by the two weights, then the malicious nodes can be detected by constantly updating the trust value of all the nodes. Simulation datas show that the detection rate of malicious nodes is nearly 97%, while the error rate is lower than 10%, the energy consumption is also significantly less.