Bi-Layer Shortest-Path Network Interdiction Game for Internet of Things

Area of Science:

• Computer Science
• Network Security
• Optimization

Background:

• Internet-of-Things (IoT) systems face significant network security challenges, leading to safety risks and economic losses.
• Attacks on IoT systems often involve a logical layer (Things) and a physical layer (Internet).
• Existing models often simplify network interdiction problems, failing to capture the complexities of bi-layer systems.

Purpose of the Study:

• To address the challenge of malicious network exploitation in bi-layer logical-physical IoT systems.
• To generalize classic network interdiction problems by introducing nonlinear objective functions.
• To develop an efficient computational method for optimizing defender strategies in IoT network security.

Main Methods:

• Reformulated the bi-layer network interdiction problem.
• Applied Benders decomposition for optimization.
• Introduced a layer-mapping module and a random-search process to enhance the algorithm's efficiency and convergence.

Main Results:

• The proposed method effectively solves the generalized shortest-path network interdiction problem with nonlinear objectives.
• The layer-mapping module and random-search process significantly improved computational efficiency.
• Numerical experiments validated the effectiveness and efficiency of the developed methods.

Conclusions:

• The Benders decomposition approach, enhanced with specific modules, provides an efficient solution for defending IoT systems against sophisticated network exploitation.
• The study offers a valuable framework for understanding and mitigating security risks in complex bi-layer IoT networks.
• The findings contribute to advancing network security strategies for the growing landscape of Internet-of-Things devices.