Effective Network Planning and Defense Strategies to Minimize Attackers’ Success Probabilities with Various Defense Mechanisms under Quality of Service Constraints

近年來由於電腦軟、硬體及有線、無線通訊技術的進步，加上基於虛擬化及雲端技術建構而成的基礎建設，各式各樣的網路應用服務正蓬勃發展。在各種服務迅速發展之際，政府、企業也順應此潮流將許多服務電子化，使網路應用服務更根深柢固地進入我們的日常生活中。然而人們對網路高度的依頼卻也造成攻擊者的覬覦，針對伺服器和網路設備層出不窮的攻擊使政府、企業遭受財務及名譽上極大的損失。為了提升網路服務的安全性與可用性，網路在惡意攻擊下的存活度便成為一個必須考量的議題。 本篇論文將網路攻防情境以數學模式描述，並搭配模擬與鬆弛法作為解題方法，同時考量網路營運者在資源佈署與防禦策略兩方面的重要議題。使用模擬的優點在於可以考量網路中眾多攻擊者類型與攻擊模式，進而衡量網路的平均存活狀態。網路營運者在佈署階段考量如何將有限防禦資源配置在網路節點上以達到嚇阻效果，防禦策略則透過攻防間動態調整網路拓樸，並應用虛擬化環境及雲端安全服務達到即時回應的防禦效果。本篇論文的結果最終將提供網路營運者採取不同防禦策略時的參考依據。Recently, due to the improvements of computer software, hardware and both wired and wireless communication technologies, governments, enterprises and individuals can now use cheap and high performance equipment. Based on the popularity of all kinds of net-devices and the infrastructure based on virtualization and cloud computing techniques, more and more applications on the Internet have been developed. At this time of web era, governments and enterprises also follow the trend to transfer their services to the web, making web applications and services get into our life more deeply. However, the heavy reliance on the Internet has caught the eyes of malicious hackers and intruders. Intrusions targeted to the web and servers keep making governments and enterprises suffer from both financial and reputational losses. In order to increase the security and availability of web services, the survivability of attacked networks has become a great issue that must be considered. In this thesis, we apply mathematical model to describe the scenario of network attack and defense, consider important issues of network operators in both resource allocation and defense strategies, and use simulation and relaxation-based method to solve the problem. The advantage of applying simulation is that we can evaluate many kinds of attacker categories with various intrusion strategies, and then examine average survivability of the network. In resource allocation, the defender considers how to allocate his finite defense resources to nodes in order to achieve the effect of deterrence. In defense strategies, the defender uses various defense mechanisms such as dynamic topology reconfiguration, request for signature and cloud computing security services to defend and react in real-time. The relaxation-based method can help us obtain a better solution for the problem, and the experiment results provide network operators a guideline to adopt different defense strategies