Complex network sampling based on particle swarm optimization

Whether the sampling subnets can accurately represent the topology and dynamics of the original networks is an important research topic. To improve the quality of network sampling, this paper attempts to convert the complex network sampling process to an optimization problem, and proposes a novel sampling algorithm which is based on Particle Swarm Optimization (PSO). Exponent of power-law degree distribution and clustering coefficient of networks were set as optimization objectives. Subnets were sampled from scale-free network by random sampling method, and optimization objectives were optimized by multi-objective optimizer. Kolmogorov-Smirnov test is used to verify that whether the sampling subnets conform to strict power-law degree distribution. Simulations show that the algorithm based on intelligent optimization methods could get better sample subnets than normal sampling algorithm. The optimization objectives of the sampling algorithm proposed in this paper could be extended to other statistical properties of complex network, and the alternative algorithm other than random sampling could also be used.