Differential Evolution for VNE-5G Scenarios

摘要

Network slicing is a key component of the envisioned 5G network. The slicing process is mathematically known as a virtual network embedding problem (VNE). Embedding a large number of virtual networks to a real physical network over time is an NP-Hard problem, and it becomes more complex when regarding different requirements of distinct types of network slices. Thus, we propose a new VNE algorithm based on Differential Evolution called VNE-DE to map virtual networks to a physical substrate. VNE-DE seeks better allocation of resources taking into account the most common QoS parameters of slice requirements: bandwidth, delay and reliability. Furthermore, we compare it with two Genetic Algorithm approaches, and its outcomes demonstrate improvement in the accuracy of the allocation of resources regarding.