考量攻擊綜效下雲端網路存活度最大化之研究

Abstract

現今許多組織與企業透過網路科技提供使用者各種服務，因此網路已然成為日常生活中所不可或缺的元素，透過網路，我們可以和朋友溝通、購物…等，但若企業未妥善保管顧客資料或是持續對於漏洞進行修補，則會讓企業帶來許多困擾甚至失去顧客的信心。從企業的角度，他們所面對的攻擊者與駭客越來越強大，技術也越來越精進，例如：攻擊者可以發動協同攻擊，此種攻擊方式會聚集一群攻擊者以合作的方式攻擊同一個目標節點，這類型的攻擊可產生綜效，因此較傳統的攻擊更具破壞力，且成功率更高。在本篇論文中，我們提出了一個與現實世界相當的綜效模型，並且可以看到並非人越多就一定有越好的攻擊效果。 除了攻擊方式不斷演進外，資訊基礎建設也同樣在進步，雲端運算是近幾年最受關注的議題之一，除具多元化的特色外，企業也可用此種技術提供使用者多樣化且彈性的服務，與更強大的運算能力。我們會將本篇論文的網路都模擬在雲端的環境上。 為了要讓企業提供的服務不中斷，在本篇論文中，我們採用了Local defense 和Migration兩種防禦機制。前者是當虛擬機器監視器察覺到所管控的虛擬機器被攻擊者攻擊時，他會對所有自己所管理的虛擬機器增加防禦力，利用此方法可以有效的防堵攻擊者的攻擊。而Migration則是可以動態的將虛擬機器從一台實體伺服器搬移到另外一台實體伺服器中，以提高網路存活度。論文中可以發現，Migration相對於Local defense 有較好的防禦效果。 本篇論文的目標在於幫助企業找到一個有效的方法對抗多樣的外部威脅，除了傳統的防火牆、IDS與IPS…等防禦措施外，也會利用上述所提到的兩種防禦策略來增加網路的存活度。此外，在有限的資源下，幫助企業或是防禦方找到最佳的防禦策略與資源分配方式來防禦攻擊者的入侵。 本研究最後會使用數學規劃合併Monte Carlo Simulation來解決此複雜與充滿隨機性的問題，讓防禦者用最有效的資源分配方式增加網路存活度。

Many organizations rely on networks to provide various types of services for customers and users. Internet has become the basic necessity in our daily life. From the enterprises’ perspectives, since hackers and attackers are getting more and more powerful and skillful, this is a challenging problem. For instance, in nowadays, attackers could launch a collaborative attack, which is a powerful attack approach that enables a group of attackers gathering their attack power toward a single target. Further, the synergy effects of this particular type of attacks could cause more damage than traditional attack approaches. We proposed a synergy effect that described the collaborative attack that similar to real life. In this model, we also demonstrate that more people in the same group will not always bring better attack power. Not only attack approaches but also the computing infrastructure have been rapidly innovated. Cloud computing is a main trend and has caught much attentions. Our thesis will implement the network base on cloud environment. We provide two defense strategies—“Local defense mechanism” and “Migration mechanism.” In “Local defense mechanism”, when virtual machine monitor (VMM) detects a node being attacked but yet being compromised, it can increase defense resources for all the virtual machines (VMs) that belong to this VMM. As for, “Migration mechanism”, it enables VMs dynamically move from one server to another. In our thesis, we could find that “Migration mechanism” has a great defense capability than “Local defense mechanism.” Our goal is to help companies find an effective way to cope with varieties of threats. In addition to traditional defense strategies, such as firewall, IDS, and IPS, we incorporate local defense mechanism and virtual machine migration to enhance system survivability. Moreover, under limited budget, it is important for defenders or companies to seek the optimal way of allocating defense resources against attackers’ invasions. The problem is modeled as a bi-level mathematical formulation. Combining with the concept of Monte Carlo Simulation, a variety of feasible attack-defense scenarios are simulated to find effective defense strategies.