Research On Adaptive Strategy Of End Hopping System Based On Deep Belief Network

In recent years,with the rapid development of the Internet,the network has brought convenience to people's daily life,and it has also brought security problems.The lack of passiveness and staticness of traditional network defenses makes defenders in a disadvantageous position in network confrontation,leading to the inequality between the offensive and defensive sides.Active network defense,as a forward-looking cyber defense technology,can take defensive measures before the attack occurs,turning the defender's disadvantage.The end hopping is a new type of active cyber defense technology that breaks the traditional defensive static with dynamic thoughts.It confuses the attacker through the everchanging end information,so that the attacker can not accurately launch the attack and realize the concealment of the communication.Thereby greatly increasing the cost of attack.We have done a lot of research on the key issues involved in the end hopping.This paper mainly focuses on the single hopping strategy in the end hopping active defense technology,and introduces multiple heterogeneous hopping modes into the end hopping.The system expands the definition of the end information and gives an adaptive adjustment scheme for the hopping strategy.Furthermore,an end hopping adaptive model based on deep belief network is proposed.The process of data collection,feature extraction and state prediction in the model is described formally.The end hopping network state feature index is defined and the feature data is established.The data set is modeled by the deep belief network,and the network state of the next cycle is predicted by the method of dynamic transition probability.The heterogeneous hopping mode is selected according to the prediction result,thereby realizing the adaptive change of the end hopping mode.This paper combines the two research directions of deep learning and end hopping,applies deep learning to network situation assessment,and combines Markov chain to propose a network prediction method of dynamic transition probability.The adaptive strategy of end hopping system is realized by choosing hopping strategy according to the prediction result.After elaborating the basic techniques of deep belief network,end hopping,Markov chain,etc.This paper proposes an adaptive end hopping model based on deep belief network,which provides a formal description of each process.And programming implementation;Further,the paper analyzes the performance of the model through experiments.The experimental results show that the model network state recognition and prediction has high accuracy,and the heterogeneous hopping strategy can effectively resist different attack types.The validity and security of the end hopping adaptive model are also be explained.