A effective two-step strategy of multi-domain virtual network embedding in 5G network slicing
Xiaolei Wang, Qiang Chen, Hang Qiu · 2017
5G mobile communication network will be based on multi-domain virtual network. Due to a huge amount of underlying network information, the ordinary single domain virtual network mapping program will cause a lot of transmission overhead and computing overhead. At the same time, different network infrastructures providers out the consideration of commercial interests are reluctant to disclose specific topological information. Therefore, based on the resource quotation, a two-step cross-domain embedding strategy of multi-domain virtual network embedding is proposed to dynamically manage the network resources in each autonomous domain. Under the premise of considering the cross-domain network environment and the difference of physical link bandwidth, the cross-domain virtual network embedding is carried out to minimize the cost of virtual network embedding. The first step is to consider node resources and inter-domain bandwidth resources, which is the main overhead for cross-domain embedding. Based on the optimized discrete particle swarm algorithm, the cross-domain virtual network embedding method is proposed to realize the effective mapping. In the second step, a fast algorithm based on Kruskal minimum spanning tree is proposed for intra-domain mapping with relatively small overhead. The simulation results show that the two-step strategy of multi-domain virtual network embedding scheme proposed in this paper can reduce the overall embedding cost in the multi-domain virtual network embedding, and has good stability and expansibility.