Adaptive robustness optimization against network cascading congestion induced by fluctuant load via a bilateral-adaptive strategy

Abstract

In real networks, the load is usually fluctuant and temporary overloading cannot always make nodes in completely failure. However, overload nodes will reject the flow and can lead to network cascading congestion. In this paper, we propose a bilateral-adaptive strategy to enhance network robustness against cascading congestion induced by fluctuant load. In the strategy, the traffic states of both receiving and sending nodes are considered simultaneously to decide the flow delivery and a mutual flow balance effect between nodes is realized. The network cascading congestion is modeled by combining the evolution equation of node queues and fluctuant load. BA networks are investigated. The results show that the node congestion proportion and the fluctuant degree increase with the fluctuant load. The critical load leading to global congestion increases exponentially with node capacity but is limited when the node capacity arrives at an optimal threshold. When the node capacity excesses the optimal threshold, the maximal load that the networks can bear for the adaptive strategy is approximately twice of that for the non-adaptive strategy. The maximal network saturation state for the adaptive strategy is also obtained. The distribution of the load-capacity ratio is found to be hierarchical and degree-dependent under different fluctuant loads. The effectiveness of the proposed strategy is verified on a real network. This study provides a reference for the identification of the maximal fluctuant load a network can bear and adaptive robustness optimization against cascading congestion induced by fluctuant load.