| 摘要: | 因應工業 4.0 來臨，越來越多物聯網設備出現在工業系統中來進行智慧製造，但大 量的設備造成了網路管理人員的負擔，且物聯網設備的資源不足使得設備內建防護能力 不足，容易受到惡意人士的侵入及控制，導致資料竊取、竄改等情事發生，甚至是利用 受害節點來癱瘓整體的工控網路，種種的資安事件都可能影響到企業的營收。
但由於工控系統與傳統資訊系統的要求不同，以及防護能力不足的物聯網設備使得 工控系統漏洞百出，不過現有資安產品並無法直接應用於工控系統中，因此我們根據工 控物聯網的特性及要求來設計相應的資安策略。本論文首先提出在軟體定義網路 (Software-defined Networking, SDN) 的環境下整合大型且擁有異質性的工控物聯網網 路，並基於零信任框架來設計系統架構以建立工控系統上的可信傳輸。對於網路上的節 點採用雙因素認證 (Two-factor authentication) 的方式，以防止陌生節點加入至規範的工 控網路中。考量到設備入侵的情況，即使認證後的設備，本論文會以集中式控制器持續 蒐集設備的操作行為，實施細粒度的身分存取控管，利用信任度演算法與信任衰減的設 計來衡量設備的信任度，並基於信任度的高低來動態調整設備在網路中的存取控制策 略，以此方式來確保整體網路節點的傳輸安全以及資料完整性。
經實驗證明，基於 SDN 的安全架構替工控物聯網提升了網路管理效率，並以集中 式控制器來緩解工控物聯網可能遭受到的資安攻擊，本研究亦針對監測任務的資源消耗 進行評估，結果顯示能以相對低成本與低侵入式的方式來實現該系統，以保障工控物聯 網的網路安全性。;In response to the arrival of Industry 4.0, more and more IoT devices are appearing in industrial systems for smart manufacturing, but the large number of devices places a burden on network managers, and the lack of resources for IoT devices leads to insufficient built-in protection, and the devices are easily controlled by malicious people, which may lead to data theft, tampering, or even the use of victim nodes to paralyze the entire industrial control network. All kinds of information security incidents may affect the revenue of enterprises.
However, due to the different requirements of industrial control system and traditional information system, as well as the insufficient resources of IoT devices, which make the industrial control system full of loopholes, and the existing information security products are not directly applicable to industrial control system, so the relative information security strategy should be designed according to the characteristics and requirements of industrial control IoT. In this paper, we propose to integrate heterogeneous IoT device more effectively in a software- defined network environment, and design the system architecture based on a zero-trust architecture to establish trusted transmission on IoT systems. For the nodes on the network, we use two-factor authentication to effectively prevent unauthorized nodes from joining the supervised network. Considering the situation of device intrusion, even after device authentication, this thesis will continue to collect the operational behavior of devices with a centralized controller and implement fine-grained identity access control. The trust level algorithm and trust decay design are used to measure the trust level of devices more effectively and dynamically adjust the access control policy of devices in the network according to the trust level.
The experiment show that the SDN-based security architecture improves the network management efficiency of IoT and effectively avoids and mitigates possible information security attacks on IoT using the centralized controller. |
