Research On Intrusion Detection Based On Feature Signal Analysis And Trust Mechanism In WSN

Wireless Sensor Network(WSN)is sensor-based,self-organizing,multi-hop network which is typically used to sense,preprocess and wireless transfer data.With the development of sensor technology,WSN have been more and more widely used in recent years.Due to the characteristic of wireless communication,self-organization,and limited resources,WSN is vulnerable to attacks which result in a series of security issues.Attacks against WSN can be classified into external attacks and internal attacks depending on the source.External attacks usually refer to the attacks that are launched outside the WSN by attackers.They can be defended by encryption and authentication methods.However,encryption and authentication methods are usually passive defenses and consume a lot of energy which is not suitable for resource-constrained WSN.Internal attacks usually refer to attacks conducted by compromised nodes within WSN.Such attacks are difficult to defend with traditional encryption and authentication methods.The compromised nodes can obtain keys and make the encryption and authentication methods failed.To solve these problems,it is necessary to design a lightweight and active approach to prevent against the external and internal attacks to ensure the network security of WSN.The main work of this paper includes:1.The related works of RSSI,LQI,and the schemes which defend against external attacks and internal attacks are analyzed.Then,the related knowledge of WSN,mathematical morphology and trust model are briefly introduced,and the characteristics and security attributes of WSN are described,and the attacks which exist in WSN are classified.2.To solve the problem that the traditional encryption and authentication methods cannot actively defend against the attacks and the high cost of energy,an intrusion detection method based on clustering algorithm and mathematical morphology is proposed.The method clusters the RSSI and LQI data in WSN,and uses the erosion method in mathematical morphology to preprocess the LQI data,and uses the silhouette coefficient method to select the better clustering result.According to the clustering result,the change of the number of nodes in the network is reflected to determine whether there is an external intrusion node.3.To solve the problem that traditional encryption and authentication schemes are difficult to defend against internal attacks,a trust model based on Beta distribution and LQI is proposed.The model calculates the trust value of the node based on the Beta distribution,and the value is used to judge whether a node is trustworthy,and realizes the detection of some types of internal attacks.The link quality is evaluated when calculating the trust value,and the adverse effect of the poor-quality links on the trust value of the normal node is reduced.The innovations in this article include:1.An intrusion detection method based on clustering algorithm and mathematical morphology is proposed.The method realizes the active defense against external intrusion by clustering the RSSI and LQI data,and the detection method runs on the server side,which can effectively reduce the energy consumption of the sensor node and is suitable for the resource-constrained WSN.2.A trust model based on Beta distribution and LQI is proposed.The model calculates the communication trust value,the energy trust value and the data trust value based on the Beta distribution,and realizes the detection of internal attacks which against communication,energy and data,and the link quality is considered when calculating the trust value,and improve the robustness and applicability of the model.