Proposal of Allocating Radio Resources to Multiple Slices in 5G using Deep Reinforcement Learning

Abstract

Fifth-generation (5G) mobile communication is expected to provide a suitable network for all service requirements. Automation of network slicing is required to respond to the dynamically changing service requirements. This paper proposes a method to allocate the radio resources that satisfy the service requirements irrespective of the number of slices utilizing reinforcement learning. From the evaluation of the proposed method using a scenario, in which the number of slices fluctuates with the passage of time, it is clarified that this method allocates the radio resources to fulfill the requirements of the service following the change in the number of slices.