A Capacity Optimization Algorithm for Network Survivability Enhancement

Load-Capacity models are usually used for analyzing the network fault-tolerance and survivability. The node's capacity reflects its redundant resources. We propose an evolutionary algorithm to search an optimized capacity allocation strategy, which makes the network achieve optimal robustness with the same resources. Our algorithm is implemented by the following two steps repeatedly: (1) Construct a new capacity allocation strategy by transferring the capacity among network nodes; and (2) Evaluate the network robustness and then decide whether or not to accept this strategy. It can be considered as a process of simulated annealing. We find that the Load-Capacity relationship seems to be neither the linear nor the preferential-attachment relation. Our works can provide algorithm support for trustworthy network design and the information security emergency response.