Research On Security Defense Technology Of Internet Of Things Perceptual Layer

With the rapid development of network technology,computer technology,smart chip technology and the the widespread popularity of smart terminals represented by smart phones has emerged as the Internet of the physical world and the network world.Since its inception,the Io T has received high attention from governments,experts and scholars around the world and it has achieved remarkable results.While the Io T has achieved rapid development,its security problems have become increasingly prominent.In order to perceive the physical world,the Io T needs to deploy a large number of heterogeneous sensor nodes to obtain state information of physical world objects in a periodic form.Data collected through sensors in the Io T is called perceptual data.Perceptual data is the foundation of the Io T.Whether the perceptual data is accurate or not has a huge impact on Io T applications,and even determines the correctness of Io T decisions.Therefore,the protection of perceptual data in Io T is an urgent problem to be solved.This paper mainly foucuses on the perceptual data protection technology based on the existing research results.This paper starts with intrusion prevention technology and intrusion detection technology,and studies the Io T perceptual data protection technology.According to the Io T architecture,as the bottom of the Io T architecture,the Io T sensing layer is the basis of the Io T application.In order to meet the needs of different applications,the types of sensing nodes are diverse,making the Io T sensing layer specific.The main characteristics of the sensing nodes are: low cost,limited computing power and storage space;sensing nodes have self-organizing capabilities;communication distance is small;sensing nodes are powered by batteries.Traditional network security technology is not fully applicable to wireless sensor networks.For this reason,this paper starts with intrusion prevention technology and intrusion detection technology to study the sensing data protection technology.The datails are as followed:Identity authentication technology can effectively guarantee the reliability of data sources.Aiming at the defects of existing authentication protocols and the high computational complexity of elliptic encryption based authentication protocol,this paper proposes a multifactor identity authentication protocol based on chaotic mapping for wireless sensor network environment.The proposed protocol introduces biometrics as the third authentication factor,which greatly improves the security of the authentication protocol.With the assistance of the gateway node,the mutual authentication between the user and the sensor node is implemented,and a secure session key is negotiated,so that the user can directly access the sensor node to access the perceptual data.After BAN logic proof and AVISPA simulation experiments,the protocol is proved to have a good resistance to known attacks.Compared with the ellipse-based authentication protocol,this protocol has lower computational complexity,which can ensure security and meet the application requirements of wireless sensor networks.The perceptual data collected by the sensor node needs to be transmitted in the form of multiple hops through the wireless link.Due to the openness of the wireless link and the limited resources of the sensor node itself,the perceptual data is highly susceptible to the tampering of data integrity during transmission,which has a huge impact on Io T applications and even affects the correctness of Io T decisions.The fragile watermark in digital watermarking technology has great advantages in protecting data integrity because of its transparency and simple implementation.In view of the high computational complexity and huge resource consumption of traditional encryption technology,this paper proposes a digital integrity protection scheme based on fragile watermarking.The proposed scheme uses the improved Logistic chaotic map to generate digital watermark sequences.Due to the initial sensitivity and unpredictability of chaotic sequences,the security of digital watermarks is guaranteed.Secondly,according to the embedding location security vulnerability of existing watermark technology,this scheme proposes an embedding location randomization strategy,which greatly increases the difficulty against attackers to launch attacks.Last but not lseat,the scheme is a reversible digital watermarking algorithm,which can guarantee the lossless reduction of data and meet the requirements of digital high precision for special application scenarios.A large number of experimental results show that the proposed algorithm not only ensures data integrity,but also saves computational overhead,and is suitable for wireless sensor network environment.Identity authentication technology and digital watermarking technology are effective intrusion prevention technologies that guarantee authentication legitimacy and digital integrity by verifying data sources and data content.As the first line of defense for the security system,they can effectively defend against external attacks.However,due to the limitations of the wireless sensor network itself,it is far from enough to defend against attacks by the intrusion prevention mechanism.The attacker obtains the security key by invading or compromising the legitimate nodes inside the network,resulting in the failure of the security defense mechanism.The compromised internal node can initiate Denial of service attacks or disruption of routing mechanisms by generating incorrect routing information can cause significant damage to sensor nodes or even entire networks.To this end,this paper proposes a hybrid intrusion detection method based on semi-supervised learning to play a vital role as the second line of defense for security protection.Firstly,the density-aware fuzzy clustering algorithm is used to realize the initial division of the data set,and then the temporary class is merged according to the number of clusters.Secondly,the fuzzy membership degree generated by the fuzzy clustering algorithm is applied to the multi-class FSVM model.A large number of simulation results show that the hybrid intrusion detection method based on semi-supervised learning has an ideal detection effect on unbalanced data sets and noise data.The experimental results show that the algorithm has strong practicability.