Abstract
Network resource reallocation is a common way to help restore performance of network systems subject to cascading failures. Majority of current network resource allocation strategies either give little regard to or make impractical assumptions about the relationship between capacity and load of network nodes, despite this relationship is closely related to the propagation of network failures. In this work we present and verify, an improved nonlinear network capacity-load model based on the actual relation between network capacity and load. According to the verified model and realistic dynamic characteristics of network loads, we propose a new network resource reallocation strategy for networks under attacks from the perspective of maintenance. The strategy aims to effectively reallocate new capacity to network nodes after cascading failures occur Both theoretical analysis and empirical studies are performed on three typical types of complex networks. Results show that the proposed network resource reallocation strategy is more efficient in mitigating devastating impact of cascading failures on network performance, in comparison to other three existing network resource reallocation strategies.