Minimizing the expected cybersecurity loss in a software supply chain through scheduling scanning jobs

Abstract

Third-party cybersecurity providers in a software supply chain execute scanning jobs according to business policies and government regulations, tailoring responses to each node’s unique attributes and risks. Traditionally, providers might use various tools at different times without coordination, leading to resource bottlenecks. This study introduces a novel approach to scanning job scheduling, addressing the unique challenge where the marginal benefit of scanning time varies non-linearly, unlike traditional job scheduling problems. The findings reveal that upgrading a scanning strategy by one level can quantify the reduced loss, providing a foundation for efficient resource allocation in large-scale supply chains. A branch-and-bound algorithm is developed to generate the optimal schedules, serving as a benchmark for evaluating other metaheuristic algorithms. Furthermore, this study proposes an innovative genetic algorithm incorporating dynamic crossover or mutation rates, as well as mechanisms to prevent premature convergence and improve performance. This approach demonstrates practical scalability and efficiency in scheduling scanning jobs across 300 nodes, ensuring adaptability to real-world supply chains.