Research On Routing-and-forwarding And Traffic Management For Content-centric Cache-enabled Networks

The traditional IP network,which focuses on connection-centric services,cannot well support the requirements of users for content services and information sharing.To this end,novel network architectures such as the Content-Centric Networks(CCN)that focus on content sharing are born,and gradually developed into an important form in future network architectures.The content-centric cache-enabled networks are more suitable for content-oriented network applications and services and meet the requirement of users for the content-centric communication paradigm.Regarding CCN and its applications,this dissertation aims to improve the efficiency and security of data transmission by conducting the following four aspects of research.To address the issues of low efficiency in routing and naming,this dissertation proposes an edge routing strategy based on hybrid naming to improve the efficiency of name management and solve the problem of forwarding table explosion.A hybrid naming mechanism is proposed and an edge routing strategy is designed based on this hybrid naming mechanism to enable inter-and intra-cluster routing collaboration.The simulation results show that the proposed strategy improves the routing success ratio and improves the data transmission performance in the mobile environment.To improve the scalability of forwarding process,a greedy forwarding strategy is proposed based on network embedding to tackle the issues of low efficiency in node relationship management and network embedding.By utilizing the social relationship of nodes,a heuristic network embedding algorithm is designed to map the network into the hyperbolic space.Based on the results of network embedding,the single-path and multi-path greedy forwarding strategies are proposed.The simulation results show that the proposed embedding algorithm is significantly better than the existing network embedding algorithms,with shorter running time,higher routing success ratio,lower path stretch,and achieving accurate network embedding.Leveraging the dataset of the Internet of Things,the proposed greedy forwarding strategies are evaluated.The simulation results show that the proposed multi-path greedy forwarding strategy obtains similar performance to the best benchmark strategy in terms of delay,hop count,and Interest satisfied ratio.Therefore,the forwarding strategy has good scalability and effectively improves the efficiency of content retrieval in large-scale CCN.To improve the rationality of resource allocation during congestion control,this dissertation proposes an optimal model for global network resource pre-allocation from the resource allocation point of view.The gradient-descentbased algorithm is proposed to solve the optimization model,and the convergence of this algorithm is analyzed.Then,a joint receiver-driven and hop-by-hop traffic shaping strategy is proposed.The simulation results show that the proposed strategy reduces the delay and packet drop,improves the Interest satisfied ratio,and achieves fairness among data flows.Based on the resource pre-allocation optimization model proposed above,a joint optimization model for congestion control and caching strategy is proposed by considering the multi-rate multicast feature of CCN’s traffic and the in-network caching.By decomposing the global joint optimization problem into multiple local sub-problems,the heuristic algorithm for the transmission rate update and the caching decision is designed.Simulation results show that the heuristic algorithm outperforms the existing congestion control algorithms of CCN in terms of cache and transmission performance metrics such as the cache hit ratio and delay.Hence,combined with the in-network caching and traffic feature of CCN,the proposed algorithm improves the efficiency of CCN congestion control via rationally allocating resources.To improve the efficiency of content cache placement,this dissertation proposes a joint optimization model for cache placement and content forwarding,which jointly considers the selection of placement node,the selection of forwarding path,and the adjustment of forwarding rate during the optimization of cache placement.The reinforcement-learning-based cache placement and content forwarding algorithms are designed to solve the joint optimization model.The simulation results show that,by adapting to the dynamic nature of the network,this algorithm reduces the data transmission time of cache placement and achieves the load balance of the storage servers.The in-network caching mechanism may cause cache consistency problems while improving the efficiency of content distribution.To this end,this dissertation designs the transmission mechanism for push-type traffic to make CCN compatible with both push-and pull-type data transmission and then proposes the cache consistency strategy based on the push mechanism,which leverages the push mechanism to actively update the outdated content in the network.The analysis and simulation results show that the proposed cache consistency strategy maintains the consistency of cache contents with low overhead,and improves the performance of the in-network caching.To tackle the issue that the existing CCN malicious flow control scheme lacks the modeling and analysis of malicious flow,this dissertation proposes an epidemic model for malicious flow based on the feature of multicast traffic.By analyzing the time evolution of the epidemic model,this dissertation derives the epidemic threshold and its upper and lower bounds.The proposed epidemic model is used to model the spread of CCN malicious flow.With the monotonicity of the epidemic threshold,the impact of forwarding strategy,caching strategy,and traffic management strategy on the spread of CCN malicious flow is analyzed.To enable the effective control of malicious flow and reduce the cost,an optimal control model for malicious flow to minimize the cost is proposed.By solving the optimal control problem,a heuristic targeted immune strategy is proposed to select the targeted nodes to deploy countermeasures.The simulation results show that,compared with the existing epidemic models,the proposed epidemic model accurately characterizes the spreading process of malicious flow in CCN.The proposed targeted immunity strategy reduces the number of infected nodes by up to80% by only choosing to immunize half of the nodes.Therefore,the immune strategy effectively suppresses the spread of malicious flows and reduces the cost of controlling the spread of malicious flows.