Base Station Popularity-Based Dynamic Resource Allocation for VNF

Network Function Virtualization (NFV) is emerging as a promising technology in future wireless communication system to achieve resource sharing through abstracting standardized network equipment into different types of virtual network functions (VNFs) to be placed in various network slices for diverse requirements. In order to satisfy the required QoS of serving users, each network slice deploys appropriate VNFs on different Base Stations (BSs) and orchestrates them for providing uniform service like an independent virtual network. However, due to the user mobility, the VNF on the BS may not have sufficient resource to provide QoS guaranteed services for newly accessed users. It is challenging to allocate adequate resource dynamically for VNFs to guarantee the required QoS of roaming users. In this paper, we propose a dynamic resource allocation scheme for VNF based on group mobility prediction. We first employ Markov chain and online learning method together to predict group mobility of users. Then we calculate the popularities of BSs for allocating more resource to the hotspot BSs (HBSs) with aim of enabling HBS permit the enormous service requests of approaching users. We propose a complementarity mechanism to maximize resource efficiency when implementing resource allocation for VNFs. Numerical results validate the effectiveness of our proposed dynamic resource allocation scheme.