A data acquisition method for anti-malicious attack sensor based on recurrent neural network

Abstract

The present invention provides a data acquisition method for anti-malicious attack sensors based on a cyclic neural network. Firstly, it proposes for the first time that the multivariate normal distribution function is used to detect abnormal data generated by malicious attack behaviors in real time, which improves the security of data transmission; secondly, When abnormal data is found, the cyclic neural network acceleration IP core based on the programmable logic gate array is called to quickly and intelligently predict the normal value of the sensor; finally, the cyclic neural network training data set is dynamically updated to improve the performance of the cyclic neural network for new sensors. The predicted fitness of the measured value ensures the rationality of the predicted data and further improves the reliability of data transmission. Moreover, this scheme also has the characteristics of low computational complexity and fast convergence speed, and is widely applicable to applications with limited computing resources. IoT edge devices.

Description

technical field [0001] The invention belongs to the security field of the Internet of Things, and in particular relates to a data collection method of anti-malicious attack sensors based on a cyclic neural network. Background technique [0002] Sensors are the main data acquisition devices at the front end of the Internet of Things, closely related to real physical objects, and provide time-series or location-related massive data for subsequent network transmission and cloud processing, so as to realize the intelligent purpose of collaborative perception and real-time monitoring of physical objects . With the widespread deployment of wireless access and sharing systems, sensor nodes are more likely to become the target of malicious attacks in cyberspace. Attackers can tamper or forge the original data of sensors, resulting in serious consequences such as misjudgment, failure, and loss of control of cloud systems. . Therefore, how to improve the accuracy, safety and reliabi...

AI Technical Summary

Problems solved by technology

[0004] However, in terms of the reasonable, safe and reliable new requirements of sensor data in the Internet of Things, the existing solutions have the disadvantages of simple algorithms and low intelligence, and cannot adapt to the new trend of digitalization, intelligence and networking of Internet of Things devices.

In particular, the above two solutions have obvious shortcomings when actively defending against increasingly severe and complex cyberspace threats. For the method of multi-source sensor fusion, the internal correlation logic of multi-source sensor data is not considered, resulting in a high false alarm rate , and the main purpose of the single-source method is to improve the anti-noise ability of the data, lack of active response strategies for simple attacks, and the collection algorithm blindly trusts the hardware circuit, making the sensor a new entry point for malicious attacks in cyberspace

Images