Data Driven Resource Allocation for NFV-Based Internet of Things
Xiaohua Tian, Wenguang Huang, Ziao Yu, Xinbing Wang · IEEE Internet of Things Journal · 2019
Network functions virtualization (NFV) architecture enables quick and cost-effective response of mobile network operators to various Internet-of-Things (IoT) applications, where the crux is to effectively and efficiently allocate resources to virtual network functions (VNFs). However, a systematical approach for resource allocation in virtualized mobile core network is still unavailable. In this paper, we propose a synthetic approach based on analysis of both network processing procedures and users' behaviors. Inspired by the static user behavior model adopted by equipment manufacturers' load test procedures, we construct a more practical user behavior model by analyzing over 20TB real data from an operator. With the model, we propose a matrix mapping-based dynamic resource allocation mechanism for the virtualized mobile core networks. To demonstrate the effectiveness of our approach, we conduct experiments using application and signaling records of millions of real users. Results show that the new approach significantly increases the resource utilization and system capacity of the mobile core networks.