一种朴素贝叶斯学习辅助的接触图路由算法

A naive Bayesian learning assisted contact graph routing algorithm

接触图路由(CGR)是星际网络路由协议中的一个重要组成部分,星际网络中节点易受电磁干扰或自身资源不足影响而失效,高度依赖先验知识的接触图路由在节点意外失效时通信性能会急剧恶化。针对上述问题,提出了在接触图路由中引入朴素贝叶斯学习模型来预测节点可靠性的方法。首先根据接触图路由的特点建立节点快照;然后基于节点快照进行朴素贝叶斯建模,预测节点间的不可靠性概率;最后在路由决策时考虑不可靠性概率,选择不可靠性小的节点转发数据。实验结果表明,引入朴素贝叶斯学习可以有效提高接触图路由在恶劣的太空环境中应对节点意外失效的能力,实现对数据的高效传输。

The contact graph routing is an important part of the interplanetary network routing protocol.Nodes in interstellar networks are vulnerable to electromagnetic interference or lack of resources.The communication performance of contact graph routing,which highly depends on prior knowledge,will deteriorate sharply when nodes fail unexpectedly.To solve above problems,a method of introducing naive Bayesian learning model into the contact graph routing to predict node reliability is proposed.Firstly,the node snapshot is established according to the characteristics of the contact graph routing;then,naive Bayesian modeling is carried out based on node snapshot to predict the unreliability probability between nodes;finally,the unreliable probability is considered in routing decision-making,and the node with low unreliability is selected to forward data.The experimental result shows that the introduction of naive Bayesian learning can effectively improve the ability of contact graph routing to deal with the accidental failure of nodes in the harsh space environment,and realize high-efficiency transmission of data.