Zhu J.-X.,Zhejiang Financial Vocational College |
Cho G.,Chonbuk National University
Journal of Networks | Year: 2014
An optimization algorithm for virus evolution is to research the spread process of a computer or biological virus in network system. The objective of the algorithm is mainly to control the speed of the virus evolution with limited network resource and to study how users can be infected in the network. A dynamical probabilistic system on a connected graph is adopted to model the virus evolution. A traditional virus evolution model needs to solve a non-convex optimization problem taking the spectral radius function of a nonnegative matrix as an optimization objective in the description of virus evolution model. On this basis, two novel approximation algorithms are proposed in this paper. Based on continuous convex approximation, the first one is a suboptimal with rapid speed. The second one can adopt branch-and-bound techniques to achieve a global optimal solution, which use some key inequalities of nonnegative matrix. Comparing with traditional virus evolution model, the simulation experiment shows that the improved algorithm can reach the global optimum in the process of virus evolution and has fast convergence capability in different network conditions. © 2014 ACADEMY PUBLISHER.