Research On Cognitive Technology For Network State In Space Information Networks

The Space Information Network is a kind of novel network with the characters of distribution, self-organization and self-management. With the development of aerospace technology, the demands of service in space become diversity, personality and broadband. To make the limited resources efficiently used to meet the tremendous business demands is becoming a significant problem.On the basis of human cognition feature, a new cognitive model for network state in Space Information Networks has been proposed. The model changes the traditional mode of passive adaption to the novel mode of active recognization, learning and accommodating to the dynamic environment. The cognitive model, based on the structure of cognitive ring, includes multi-domain environment-sensing module, intelligent decision-making module and network reconfiguration module. The cognitive information is selected from amount of low quality information collected from multi-domain environment to support the ability of self-x, just like self-organization, self-management and maintain the high performance of the Space Information Networks. The cognitive function has been divided into three layers, multi-domain local cognitive layer, regional collaborative cognitive layer and overall cognitive layer. The cognitive function of nodes has been achieved by the cross-layer design scheme of shared database.A network invulnerability assessment algorithm based on the importance of nodes has been presented, according to the cognitive model and the demand of robustness in space information networks. In the node importance evaluation algorithm, the node’s contributions to network structure and physical loading are combined to evaluate the extent of cascading failures of the nodes. The importance of nodes is gained by comparison of network performances before and after cascading failure. The degree of dispersion equates to the network invulnerability, the lower the better.According to the demands of security in Space information networks, an assessment algorithm of network safety is provided. The perceptrons are designed for the Space Information Networks to analyse and classify the threats in the networks. Not only considering the direct influences from threats, but also the indirect influences are taken into account when evaluating the risks of threats. The risk of the whole network has been disintegrated into the risks of nodes and the security services on the nodes. Using the joint probability theory, The risks of the whole network have been disintegrated into the risks of nodes and the security services on the nodes.The two assessment algorithms proposed above have been simulated by the software of Matlab. The results prove that they are suitable for the Space Information Networks and can accurately evaluate the robustness state and security state of the networks.