Joint pre- and post-disaster resource allocation for resilience improvement in computing power network
提出一个分阶段韧性评估框架和联合资源分配模型，通过半马尔可夫链建模系统行为，优化灾前与灾后策略以提升计算能力网络的韧性，并通过案例验证其有效性。
Computing power networks play a crucial role in supporting modern digital infrastructure, whose resilience needs to be improved due to the increasingly frequent and severe disruptive events. This paper introduces a phase-based framework for resilience assessment, coupled with an integrated resource allocation model that jointly optimises pre- and post-disaster strategies to systematically quantify and enhance the resilience of computing power networks. The system behaviour is modelled as a semi-Markov chain, capturing the stochastic dynamics of multi-state transitions under disruptions. Overall resilience is evaluated as a synthesis of four principal attributes: resistance, absorption, adaptation, and restoration, each characterised by both inherent and acquired measures. Analytical expressions for these measures are derived using the theory of aggregated stochastic processes, enabling efficient and accurate resilience quantification. Furthermore, a resource allocation model is developed to optimise resilience improvement strategies under resource constraints, incorporating both pre- and post-disaster decisions. Two case studies on homogeneous and heterogeneous computing power networks are conducted to validate the applicability of the proposed evaluation framework and demonstrate the superiority of the integrated resource allocation strategy in enhancing resilience. Finally, some managerial insights are provided to designers and managers responsible for resilient computing power network design and emergency decision-making.