Resist Selective Forwarding Attacks In Wireless Sensor Network Based On Clustering And Game

Wireless sensor network(WSN)is a self-organizing network composed of a large number of sensor nodes deployed in harsh or hostile environments.It is responsible for collecting environmental information in the monitoring area and sending the processed information to the user through the transmission medium.Compared with the traditional Internet network,WSNs are very different in terminal types,network topology,data transmission,and networking methods,which cause WSNs to face more security threats than the Internet.At the same time,because sensor nodes have the characteristics of simple functions,limited computing power,limited storage capacity and limited energy resources,these make the work of WSNs in network security more challenging.Attacks in WSNs can be divided into external attacks and internal attacks.Traditional encryption technology can protect data information,and identity authentication mechanisms can verify the legitimacy of nodes.These methods can effectively defend external attacks.However,malicious nodes in internal attack have authentication and key information for network communication.Malicious nodes with legitimate identities can launch attacks on the network by participating in the communication between nodes.The selective forwarding attack is one of the most representative internal attacks,it has strong concealment and is easy to combine with other attacks,there is no way to defend it perfectly.This paper aims to propose a scheme that can detect and defend selective forwarding attacks on WSNs.The full text of the work is summarized as follows:1.Based on the existing detection methods,the requirements that the detection scheme for selective forwarding attacks need to meet are proposed.In-depth understanding of the research status of WSNs and the main types of attacks on different protocol layers.The selective forwarding attack,which has strong concealment and is difficult to be detected,is emphasized.By summarizing the advantages and disadvantages of existing security mechanisms,this paper concludes that the detection scheme for selective forwarding attacks in WSNs needs to meet the following requirements:(1)Reduce the complexity of the algorithm;(2)Reduce the mis-identification rate of normal nodes;(3)Reduce the missed detection rate of malicious nodes;(4)Maximize network throughput;(5)Reduce the algorithm's impact on network lifetime.2.A detection and defense scheme for selective forwarding attacks in WSNs based on DBSCAN and game theory is proposed.After in-depth analysis of classic clustering and existing game theory-based anomaly detection methods,this paper combines network characteristics and the advantages of various algorithms to propose a detection and defense scheme based on DBSCAN and game theory.The main innovations of this scheme include:(1)According to the difference between the behavior of the captured malicious node with legal identities and the normal node,this paper use cumulative forwarding rates(CFRs)when nodes act as cluster heads(CHs)and cumulative transmission rates(CTRs)when nodes act as member nodes(MNs)to form a data set,for the sink node(SN)to execute the clustering detection algorithm.The algorithm is handed over to the SN to execute periodically,which can reduce the burden of the monitoring node(IN)and reduce the energy consumption of nodes.(2)A single round of clustering detection results often has a higher missed detection rate.In this paper,in the network environment where malicious nodes and normal nodes coexist,the interaction between suspicious nodes and normal nodes is established as a non-cooperative game with incomplete information,and a payment matrix is established according to the behavioral strategies of all participants to solve the mixed Nash equilibrium.The combined algorithm of DBSCAN and game can significantly improve network throughput and ensure network security.A dynamic behavior monitoring mechanism is adopted in the combined algorithm,and compared with the full-time behavior monitoring,it's network lifetime is longer.3.Based on the previous research results,an improved detection and defense scheme is proposed.The DBSCAN algorithm uses global parameters for clustering,when the density difference between clusters is large,its clustering effect is poor.Hierarchical clustering can get different density clustering under different termination conditions.This paper analyzes the principle of HDBSCAN,and then constructs a detection and defense scheme for selective forwarding attacks in WSNs based on HDBSCAN and game theory.The improved combination scheme has fewer missed detections of malicious nodes,and the network throughput is further improved.4.This paper uses MATLAB 2019a as the simulation platform to construct the network layout and simulate the selective forwarding attack of malicious nodes in WSNs.The performance of the algorithm is evaluated by five indicators including the number of false alarms,the number of missed alarms,the network throughput,the number of successfully forwarded packets per unit time,and the network lifetime.Moreover,the comparison scheme is used to test the effectiveness of the proposed algorithm.The simulation results show that forcing malicious nodes to adjust their behavior strategies can effectively increase network throughput and extend network lifetime.In addition,compared with the combined scheme based on DBSCAN,the combined scheme based on HDBSCAN has greater network throughput.