Reinforcement Learning Environment for 5G IoT Device Resource Management

Authors: Logan Sherlock, Renu Balyan

SUNY Campus: SUNY Old Westbury

Presentation Type: Poster

Location: UU 111

Presentation #: 68

Timeslot: Session C 1:45-2:45 PM

Abstract: This research project aims to develop a resource management framework for efficient allocation of 5G network resources to IoT (Internet of Things) devices. As 5G technology is increasingly integrated with IoT applications, the diverse demands and use-cases of IoT devices necessitate dynamic resource management. The current focus is the development of an IoT device environment utilizing reinforcement learning for resource adjustment. This environment is part of a larger project that also involves the implementation of Massive MIMO (Multiple Input Multiple Output) network slicing. The massive MIMO implementation will be handled by collaborating teams at other institutions and followed by collective integration efforts. The environment observes IoT device parameters including the current BER (bit-error-rate), allocated bandwidth, signal power level, device distance from transmitter, and the device angle relative to transmitter. Actions that can be taken by the RL agent on the environment include adjustments to the bandwidth and signal power level of an IoT device. Both the observation space and the action space of the reinforcement learning agent are continuous spaces (range of values). The environment is currently being tested with PPO (Proximal Policy Optimization), and DDPG (Deep Deterministic Policy Gradient) RL algorithms. Initial results show PPO models to train at a faster rate, while DDPG models have explored a wider range of states, leading to better model predictions. Future work includes refinement of the reward function and testing of additional algorithms such as DQN (Deep Q Networks).