Optimal network intrusion detection assignment in multi-level IoT systems

Due to the dynamics of edge–fog computation resource availability and network traffic, it is challenging to best assign network intrusion detection (NID) tasks to the appropriate edge–fog nodes in a multi-level IoT NID system. In this paper, we first propose an integer linear programming (ILP) formulation to find an optimal NID task assignment with the objective of minimizing the NID latency while meeting the NID accuracy requirements and computation resource constraints. Then, we propose three heuristic algorithms which are shortest detection (SD)-based, nearest neighbor (NN)-based, and genetic algorithm (GA)-based assignment to efficiently find the near-optimal solutions. Extensive simulations based on two real-world IoT network attack datasets are conducted to justify the effectiveness of our proposed algorithms in terms of the NID accuracy and latency. © 2023 Elsevier B.V.