Research On Intrusion Detection Algorithm For Low Power Internet Of Things

In recent years,as an emerging technology,the Internet of Things has gradually attracted people's attention.Wireless sensor networks(Wireless Sensor Networks,WSN),as an important part of the Internet of Things,have also developed rapidly with the Internet of Things.However,WSN is more vulnerable than traditional networks and faces more security issues.Therefore,it is particularly important to design an intrusion detection system suitable for WSN in view of the problem of limited energy and computing power of WSN nodes.First,after analyzing the structure and characteristics of WSN,summarize the security issues of WSN,and analyze the needs of WSN intrusion detection according to the characteristics of WSN.In view of the limited energy and dynamic characteristics of WSN,a hierarchical structure intrusion detection system is designed to ensure the security of WSN.Secondly,in order to meet the WSN's demand for energy saving in intrusion detection systems,analyze the deficiencies of the traditional LEACH algorithm,and improve on this basis,designed a WSN layered routing protocol based on genetic algorithm.The genetic algorithm is used to select the cluster head,which makes the distribution and energy of the cluster head more balanced.The single-hop and multi-hop inter-cluster transmission strategy is adopted to reduce the energy consumed by the cluster head nodes.The algorithm is simulated and analyzed by Python,and compared with the LEACH algorithm,which proves that the protocol in this paper is superior to the LEACH algorithm in terms of network survival time,network energy consumption,data transmission volume and energy balance,and proves the superiority of the protocol.Then,in order to meet the requirements of WSN for the security of the intrusion detection system itself,a trust mechanism was introduced into the detection system,and a secure routing protocol based on the trust mechanism was proposed.It is improved on the basis of Bayesian trust model and adopts the combination of direct trust and indirect trust to evaluate trust.The trust mechanism is used to improve the hierarchical routing protocol based on genetic algorithm,and the trust mechanism is added in the preparation stage,cluster head selection stage and inter-cluster transmission stage respectively to improve the security of the network itself.Through simulation analysis of the trust value changes of normal nodes and malicious nodes,it is proved that the improved algorithm can effectively identify malicious nodes;by comparing the performance of the routing algorithm before and after improvement,it is proved that the improved algorithm can improve the network with less overhead Security.Finally,in order to meet the requirements of WSN for the accuracy of intrusion detection system,on the basis of analyzing the characteristics of neural network,an intrusion detection algorithm based on particle swarm optimization extreme learning machine(PELM)is proposed.The shortcomings of BP neural network are analyzed,and ELM with good generalization performance is used for intrusion detection;for the problem of ELM algorithm randomly selecting the input layer weights and hidden layer thresholds,the PSO algorithm is used to optimize its initialization stage;PCA is used The algorithm performs feature selection on the simulation data to reduce the pressure of the intrusion detection algorithm.The performance of the PELM algorithm is tested by simulation,and compared with the BP algorithm and the ELM algorithm,which proves that the PELM algorithm has a high detection rate;counting the running time of the entire intrusion detection system proves that the intrusion detection system designed in this paper has a good performance Real-time.