Biological systems exhibit remarkable adaptation and robustness in the face of widely changing environments. While, current Network Security researches on bioinspired are all limited in the layer of human body - immune system. On the basis of current research in bioinspired domain and aiming at the limitation of these relative research, this paper puts forwards Bio-inspired Multidimensional Network Security Model (BMNSM) based on "three-net paralleling", which offers a novel angle of view and an integrated space model to the research of bio-inspired network security. It analyses and abstracts the network states and proves the abstract model to be a linear state space, which infers the systematicness and feasibility of the model. It also proves the excellent performance of BMNSM by quantitative analysis of stationary performance, comparing to traditional running mode.