| There exist higher requirements for the carrying capacity,security and stability of the train network with the continuous increase of train-mounted equipment that transmits data by wireless communication.The current system configuration of the existing train network is not flexible enough,and the reconfigurability is unsatisfactory.If a certain control device is malfunctioned in the train,it may affect the topology of the entire train network,reduce the quality of network communication service,and cannot meet the needs of certain control tasks.At the same time,the existence of failure nodes will occupy and waste part of the network resources,causing waste of resources and even reducing network performance.TCSN aims to establish a network control system that integrates the train control network and the passenger service network.The system has higher bandwidth and reliability,and can be reconstructed when the network fails.This paper studies and analyzes the failure of a single wireless node in the train network.At the same time,it combines the characteristics of the TCSN network to research and design the topology reconstruction algorithm,and perform related simulation verification.The main contents of this paper includes the following aspects:1.It is analyzed and summarized that failure reconstruction model of TCSN train.According to the role of nodes and network structure,filed nodes were divided into different types.The failure models of critical nodes and non-critical nodes are analyzed in depth.The difference in the influence of different node failures on network communication capabilities is compared through network indicators.On this foundation,taking the importance of the failed node to the local network load and the number of communication hops as the evaluation index,an evaluation model of the importance of non-critical nodes is proposed.2.Derived from the TCSN network key node failure model,a local topology reconstruction algorithm NS-TRA is proposed,which based on neighbor sets and with cut points as the network key nodes,.The tarjan algorithm is adopted to detect the key nodes of the network,namely the cut vertex.When a failed node is discovered by the node detection mechanism,the improved backtracking DFS algorithm is used to disassemble the adjacency matrix maintained by the sink node into multiple adjacency sub-matrices which are isolation from one another.After that,the weights between the nodes could be calculated within the radius that is the longest distance between the repaired nodes and the neighbor sets of the failure nodes.The k links are preferentially selected according to the high weight,and then the adjacent sub-matrices are merged by repairing the links.The NS-TRA algorithm is simulated in the specified TCSN topology and the random topology of different numbers of nodes.It is obvious that the connectivity of the TCSN network is restored within a relatively low time complexity and the communication load of wireless nodes are reduced via the topology reconstruction algorithm in this paper.3.Based on the evaluation model of the non-critical node importance,the threshold parameters of significance are set for the TCSN failure node.The failed nodes that exceed the threshold are reconstructed using the local topology reconstruction algorithm based on the connected dominating set,namely CDS-TRA.The improved BFS algorithm is applied to extract the adjacency matrix of the failure nodes within d-hops and the mean of neighbor node degree priority rule is adopted to construct the m-CDS of the adjacency matrix in CDS-TRA.The m-CDS connection is realized by the minimum spanning tree algorithm,and the new links are updated to the adjacency matrix finally.Through the analysis of the simulation results,it can be seen that the CDS-TRA algorithm has some superiorities,including the ability to reduce the negative impact on the network life cycle after the load redistribution of the TCSN failed nodes,to achieve load balancing of local nodes,and to extend the life cycle of the entire network.As the value of m increases,the load balancing effect of local TCSN nodes are improved.As the d-hop range increases,the number of hops between the local node and the sink node is reduced,and the end-to-end communication delay is reduced. |
