Research And Implementation Of Malicious Node Detection Based On Correlation Theory In WSN

The deployment of nodes in wireless sensor network(WSN)is random and unattended,which makes the nodes in WSN vulnerable to various attacks.Since environmental monitoring is based on the data collected by nodes,attackers usually launch false data injection(FDI)attacks to affect the detection results.In order to resist FDI attacks,this paper proposes a malicious node identification scheme based on correlation theory.The scheme proposed in this paper is divided into three parts.The first part is to judge whether the same attribute data is abnormal based on time correlation.The main idea of this part is that the node uses its own historical data to establish a prediction model.By comparing and analyzing the difference between the predicted value and the actual value,the abnormal situation of the same kind of sensor data is judged.The second part is to identify malicious nodes based on spatial correlation by using the abnormal conditions of the attribute data obtained in the first part.Because environmental events will make the data changes of the nodes in the adjacent area show high similarity,so as Cluster head nodes can judge whether the nodes are malicious by using the short-range spatial correlation between the attribute data of a single node and the long-range spatial correlation between the adjacent nodes.Finally,the gateway comprehensively considers the change trend of the data when the event occurs,and uses the event correlation to verify the malicious nodes.The combination of the above three correlations can achieve the purpose of detecting malicious nodes,but there are the following problems in the detection of each correlation.In the aspect of time correlation,due to the variability of the environment,the prediction results of the model are not accurate,which affects the judgment of the abnormal situation of the same kind of sensor data.In the aspect of spatial correlation,in the close spatial correlation,D-S evidence theory judges whether the node is abnormal by fusing the anomalies of each attribute data,but D-S can't deal with the problem of evidence conflict.In the long-distance spatial correlation,due to the interference of the surrounding environment,the malicious node detection may be affected by the node failure transmission.Finally,because the temporal and spatial correlation does not consider the change trend of data from the perspective of events,there will be misjudgment or missed judgment.In view of the above problems,the following contents are mainly studied(1)Time correlation: in order to improve the detection accuracy of the same type of sensor anomaly,the second-order differentiated difference filter(ddf-2)algorithm is introduced to modify the prediction model.(2)Spatial correlation: in the short-range spatial correlation,in order to solve the D-S conflict evidence problem,this paper proposes to introduce Ada Boost algorithm for classification.In the long-range spatial correlation,the calculation method of correlation coefficient is designed,and the nearness theory is introduced to optimize the weight of each correlation value,so as to weaken the impact of fault nodes.(3)Event correlation: since the temporal and spatial correlation does not consider the change trend of data from the perspective of events,the malicious nodes identified in WSN can be verified by observing the development trend of events.In this paper,the correlation theory is used to detect malicious nodes to resist FDI attacks.The feasibility of the proposed scheme is verified by experiments.The results show that the proposed scheme has advantages in recall rate,false positive rate and false negative rate.Therefore,the proposed scheme can detect malicious nodes and resist FDI attacks effectively in WSN.