On Enhancing Network Throughput using Reinforcement Learning in Sliced Testbeds
Daniel Pereira Monteiro, Lucas Nardelli de Freitas Botelho Saar, Larissa Ferreira Rodrigues Moreira, Rodrigo Moreira
TL;DR
The paper addresses the challenge of achieving high throughput with low latency and reliability in network slices that must meet SLAs. It introduces eMBB-Agent, an RL-enabled vertical application that uses discrete actions to adjust the reception window, validated by a Deep Q-Network. The authors evaluate how factors like channel error rate, DQN depth, and learning rate affect convergence and throughput using NS3-based simulations with NS3-GYM integration. Findings suggest that shallower DQN configurations can deliver higher throughput and faster convergence, highlighting a trade-off between model complexity and real-time adaptation, and underscoring the need for real-world validation due to simulation limitations.
Abstract
Novel applications demand high throughput, low latency, and high reliability connectivity and still pose significant challenges to slicing orchestration architectures. The literature explores network slicing techniques that employ canonical methods, artificial intelligence, and combinatorial optimization to address errors and ensure throughput for network slice data plane. This paper introduces the Enhanced Mobile Broadband (eMBB)-Agent as a new approach that uses Reinforcement Learning (RL) in a vertical application to enhance network slicing throughput to fit Service-Level Agreements (SLAs). The eMBB-Agent analyzes application transmission variables and proposes actions within a discrete space to adjust the reception window using a Deep Q-Network (DQN). This paper also presents experimental results that examine the impact of factors such as the channel error rate, DQN model layers, and learning rate on model convergence and achieved throughput, providing insights on embedding intelligence in network slicing.