Theory And Key Technology Research On Survivable Network System

Network security technology has been developing rapidly for years. There always occur successful malicious attacks due to the openness of Internet and the development of attack techniques, which makes it impossible to build absolutely secure network systems. Moreover, random failures of system components as well as unpredictable accidents do harm to network systems. Attacks, component failures and accidents all decline the quality of service (QoS) of network systems. Survivability aims to protect network system services and enable systems to fulfill their mission in a timely manner in the presence of attacks, failures, or accidents.This thesis is concerned with the theory and the key technologies of survivable network system, including the system architecture and design method, formal description of survivability, redundancy-based building method of network systems, dependability analysis of component-based software systems, and quantitative evaluation of network systems survivability. The contributions made in this thesis are listed below:1. An architecture of survivable network systems is proposed. Several survivability techniques are organized into the architecture, in which intrusions can be resisted, recognized, and tolerated to a great extent;2. To meet different application requirements, two methods are presented for building survivable network systems. These methods stem from the redundancy technique and integrated with major voting and threshold scheme respectively;3. A method is presented to analyze the dependability of component-based software system. The states of software system are described by the states of the components, the failures and the corresponding repairs by the changes of the components'state, and the state transition by a Continuous Time Markov Chain (CTMC). The effects on the system dependability of component failure and interface failure are measured and compared;4. The relationship among dependability, security and survivability is investigated. Network systems are sorted according to the redundancy techniques adapted and corresponding stochastic models are constructed, survivability based on different application requirement are evaluated.