A MODEL FOR THE DISTRIBUTION OF COMPUTATIONAL TASKS IN CLOUD INFRASTRUCTURE INCORPORATING PERFORMANCE, COST, AND SECURITY CONSIDERATIONS

Abstract

Cloud systems (CSs)—now integral to the business processes of many organisations—face sophisticated challenges such as targeted attacks, exploitation of software vulnerabilities and the leakage of confidential data. These threats greatly increase the demands on both security and efficient resource allocation. Against a backdrop of escalating cyber-risks and rising performance requirements, it is essential to design new task-allocation models that simultaneously account for security, performance and cost within cloud infrastructures. Existing theoretical approaches often study these factors in isolation and ignore the strategic interaction between attacker and defender, limiting their practical usefulness.This paper presents a hybrid model that couples antagonistic game theory for cloud-risk assessment with multi-objective optimisation based on a modified NSGA-II algorithm. Attacker behaviour is represented by an aggressiveness parameter (λ) that influences the probability of node compromise, whereas defender behaviour relies on adaptive task-allocation mechanisms. The optimisation problem minimises three objectives: total task-placement risk (cloud security), total task-processing time (cloud performance) and total cost of resource usage (cloud cost-efficiency). Simulations carried out in a Python environment confirm the effectiveness of the method, yielding IGD = 0.2263, Spacing = 0.0106 and Hypervolume ≈ 1.3310. These metrics indicate good convergence, a uniform spread and high diversity of the Pareto-optimal front for a protected cloud system. The proposed model therefore offers a flexible trade-off among conflicting criteria and can adapt to diverse adversary-behaviour scenarios.