Research On Grid Computing Oriented On-demand Intrusion Detection Model And Key Technonogies

Grid computing, called the third generation information technology after internet and wide world web, comes forth in the early 1990s. It would provide next generation distributed applications and services which takes a profound effect on information system research and development. At the same time, it also challenges to present information system security.As a kind of active measure of information assurance, Intrusion Detection (ID) acts as a effective complement to traditional protection technologies. With the development of computer and network technologies, Distributed Intrusion Detection (DID) has been the focus of intrusion detection and even the whole realm of network security.Although present distributed Intrusion Detection Systems (IDS) can guard against network oriented attacks, it lacks agility that on-demand fast dynamic organization intrusion detection systems, which made it as inefficient system on grid computing. So it can't effectively cope with the challenges of monitoring frequent and unpredictable changing grid computing tasks, grid computing security link structure, threats of large-scale distributed coordinated attacks, et al. In this thesis, we research several critical problems on grid computing oriented DID.This paper focuses on On-Demand Intrusion Detection Model (ODIDM) and its support techniques. For dynamic share and multi-domain integration properties of grid computing, leveraging Shared Data Environment (SDE) technique, this paper presents an on-demand intrusion detection model and ODIDM based On-Demand Intrusion Detection System (ODIDS) agilely constructed virtual intrusion detection system to jointly monitor the changing grid computing application. System prototype and its experiments express this model was acceptable.Aiming at load balancing of ODIDS, leveraging connected vessels theory, this paper presents hydraulics based dynamic load balancing approach and Cellular Automata Programming Algorithm (CAPA). Theory proof and experiments express CAPA algorithm is fast convergence.Aiming at conspiracy attacks of ODIDS, this paper presents a Lagrange polynomial based Principal and Subordinate Threshold Secret Sharing System (PSTSSS) and PSTSSS based Multi-Domain Resources Threshold Secret Sharing System(MDRTSSS).Differing with traditional threshold scheme, private keys weredivided into principal and subordinate shadows keys and can guard against conspiracy attacks. Theory proves that PSTSSS was secure, high information rate and easy implementation. So PSTSSS based intrusion tolerant system could counteract conspiracy attacks.Comparing with traditional distributed intrusion detection system, ODIDS has agility, that is, according to frequent and unpredictable changes of grid computing application ODIDS can fast reconstruct virtual intrusion detection system to cooperatively monitor grid computing task and didn't waste any resources.