Reinforcement Learning Approach for Resource Allocation in 5G HetNets

Heterogeneous Networks (HetNets) have been hailed as a critical technology for 5G communications, allowing for the rapid expansion of mobile traffic. HetNets can increase network capacity and serve additional users by installing small cells inside macrocells. However, resource allocation for such networks becomes more challenging than for conventional cellular networks due to interference between small-cells and macrocells, making it more difficult to provide quality of service. Deep Reinforcement Learning (DRL) has opened the door for applications in resource allocation for 5G HetNets, because of recent breakthroughs in the field. We present a unique resource allocation technique based on DRL that may be used to both small and macro cells. According to the resource allocation process, an autonomous “agent”, in our case a cell, makes judgments to determine the appropriate BS to assign to a user, and the optimal number of Resource Blocks that should be allocated, while not needing or waiting for any information.