Research On Key Technologies Of SDN-based Network Moving Target Defense

With the rapid development of computer science and the deteriorating network security situation,the traditional static network architecture is difficult to effectively deal with intelligent network attacks,which is always in the passive state.In order to break the imbalance between attack and defense,countries have successively promoted the Network Moving Target Defense(MTD).MTD realizes a dynamic,continuous,and changeful network environment by actively modifying network configuration attributes and diversifying the composition of network elements.MTD improves the uncertainty of the system and increases the cost and difficulty of attack,which opens up new ideas for network security.With the flexible and efficient characteristics of Software Defined Network(SDN),we research the critical technologies of SDN-based network moving target defense,which mainly includes the following contents:Firstly,we combine Deep Reinforcement Learning(DRL)with IP Mutation.We abstract the IP mutation and attackers’ behaviours into Markov Decision Process(MDP).We creatively divide the IP mutation into two stages:IP address block allocation and randomly selecting addresses for mutation.We use the PPO algorithm to learn the optimal strategy.Secondly,we take the scenario of multimedia video service as an example to research Route Mutation.We utilize Satisfiability Modules Theories(SMT)to model and solve the constraints of routing paths for different scales of network topologies.Then,we use the feasible solutions as an action space for further route mutation strategy learning.Thirdly,considering the blindness of the current network mutation mechanism,we propose a prediction model based on Recurrent Neural Network(RNN),which judges the possible future behaviours of the attacker based on the historical traffic data.The model sends the prediction results to the IP mutation and route mutation modules to dynamically and actively adjust the mutation strategy and period.Fourthly,we implement the network mutation system based on Mininet and Ryu controller.We realize the monitoring platform and video application based on Angular and Django architecture.Finally,we evaluate the system from multiple indicators.In summary,this thesis aims at the complex network environment,based on the emerging field of artificial intelligence,focusing on the ultimate purpose of realizing the critical technologies of network moving target defense with high security,high intelligence,and high efficiency.