Research And Application Of Endogenous Security System Based On Neural Control

The structural complexity of network information systems and the richness of data types within the system are gradually increasing.Although the research results of existing active security defense technologies are abundant,they lack a high degree of integration with the original system.Moreover,there is a lack of information linkage in processing and control,and there is a lack of practical and efficient application support when the system is subject to security threats and affects task execution.In addition,due to the limited granularity of perception within the traditional network information system,there is a lack of comprehensive perception of the system’s security status,and fine-grained regulation is not possible when building a security defense.To address the shortcomings of traditional defense from the root,it is necessary to improve the integration of security and functional elements,enhance the information linkage within the system,and provide fine-grained perception of functional components,so as to have sufficient prediction and regulation capabilities in the face of unknown threats and improve the effectiveness of system defense.This thesis fully draws on the control mechanism of the human nervous system.The key elements of the human neural control system are analogized to the endogenous security system by combining the characteristics of the human neural system with the actual environmental features of the software system.An endogenous security system model based on the human neural control mechanism is proposed by fully integrating the functional elements of components with security elements in a task-oriented manner.The dynamic execution mechanism of the endogenous security system is designed based on this model.The internal data and modules of the system task execution process are formally defined and described by analogy with the neural control process.The system data is acquired by fine-grained sensing of the underlying data through endogenous security receptors,and processing and control mechanisms are designed according to the data characteristics.The processing mechanism includes incoming and outgoing processing,and the control mechanism includes execution flow protection and security state prediction.Finally,an endogenous security mechanism based file transfer system is designed and built.The system is internally integrated with endogenous security-aware services,pivot processing,and pivot control.The effectiveness of the model and execution flow protection and security state prediction mechanism proposed in this thesis is verified by testing through simulated attack experiments.The main work of this thesis is as follows.(1)To address the problem that when traditional defense mechanisms face security threats,there is a lack of integration between security and functional modules leading to the inability to unify the execution of functions and the handling of attacks.This thesis presents an endogenous security system model based on neural control.The model is analogous to the human neural control mechanism and divides the endogenous security system into three layers according to the actual operating environment characteristics of the endogenous security system,including the task layer,the regulation layer and the central layer.The task layer mainly contains endogenous security sensory receptors,the control layer contains peripheral connections and effector function modules,and the central layer contains processing and controllers to analyze and process the underlying data and control functions.This thesis uses the neural control-liked endogenous security system model as the basis,classifies endogenous security sensory receptors and designs dynamic endogenous security systems.In this thesis,we abstractly describe the working process of endogenous security sensory receptors and the dynamic execution process of the system.(2)In response to the problem of inefficient countermeasures due to inadequate linkage of components within traditional defense mechanisms,this thesis proposes a task-oriented pivotal processing and control mechanism for endogenous security systems.The processing mechanism includes incoming and outgoing processing by the processor.The system provides data for the controller by integrating and analyzing the underlying data through the incoming processing mechanism,and realizes the underlying adjustment by processing the initial control signals and generating characteristic instructions through the outgoing processing mechanism.The control mechanism includes the function execution flow protection of the controller and the endogenous security prediction mechanism,and the protection of the internal function execution flow of the system and the prediction of the security state of the system are realized by constructing a finite state machine and a decision tree.(3)An experimental platform for an endogenous security system model with neural control is designed and built.The sensing of the system task operation is achieved through software programming,and the underlying data is processed and controlled using bytecode enhancement techniques and decision trees.The execution flow protection and security state prediction functions of the controller module are tested through simulated attack experiments to verify the feasibility of the endogenous security system model proposed in this thesis,and to test the effectiveness of the function execution flow protection mechanism and the accuracy of the endogenous security decision tree.The test results show that the execution flow protection mechanism successfully identifies the illegal execution flow and restores it to the legitimate execution flow;the prediction of the security state of the test set achieves an accuracy of 97.4790%.